Introduction
Originally described by Leksell, stereotactic radiotherapy has evolved from an invasive frame-based gamma knife to a non-invasive linear accelerator-based radiosurgical technique. Reference Leksell1–Reference Sarkar, Pradhan and Munshi3 Stereotactic treatment plans can be produced by conformal dynamic arc, static arc, fixed gantry 3D conformal radiotherapy (3DCRT), intensity-modulated radiotherapy (IMRT) or volumetric modulated arc therapy (VMAT). Stereotactic radiosurgery/stereotactic radiotherapy (SRS/SRT) uses multiple beams distributed over a large surface area to deliver a high dose to the target volume. It is essential to maintain the spatial accuracy of the delivered dose with respect to the stereotactic coordinate system. It is necessary to do this to ensure high-dose radiation is delivered in an optimal manner to the target volume, encompassing pathological tissue while minimising radiation-induced damage to surrounding tissues. Reference Puataweepong, Dhanachai and Hansasuta4–Reference Mitsumori, Shrieve and Alexander5 Techniques such as hypofractionated SRT and SRS resulting in delivering high dose per fraction can lead to a lethal/supralethal effect which in turn can induce ablation and direct endothelial apoptosis. SRS is also known to have other radiobiological advantages such as abscopal effect which can limit tumour cell repopulation and enhance anti-tumour immunity. The safety and efficacy of SRS/SRT in treatment of different intracranial lesions is well established. Reference Puataweepong, Dhanachai and Hansasuta4–Reference Dunbar, Tarbell and Kooy6
However, SRS treatments using multiple static cobalt sources with gamma knife-based systems are gradually being replaced by linear accelerator-based systems. Reference Leksell1,Reference Norén, Greitz, Hirsch and Lax7,Reference Touboul, Al Halabi and Buffat8 Moreover, the invasive rigid frame-based systems for single-fraction radiosurgery are also being replaced by frameless approaches. Invasive frame-based systems require the placement of screws on the patient’s head and necessitate completion of treatment on the same day. Frameless masks are easy to make and are also more flexible in using imaging modalities (X-ray-based stereoscopic imaging as well as on-board Computed Tomography (CT)). Frameless masks provide the freedom to choose between single-fraction and multi-fraction treatments. Reference Sarkar, Munshi, Manikandan, Anbazhagan, Ganesh and Mohanti9,Reference Munshi, Sarkar, Roy, Ganesh and Mohanti10 Since hypofractionated schedules cut down on machine usage time and are also more convenient for patients, they can be seen as a resource-sparing strategy. Reference Sarkar, Pradhan and Munshi3
SRS and hypofractionated SRT are considered important in high-precision radiotherapy for benign brain lesions and some malignant lesions, and are used to treat tumours such as metastasis, meningiomas, acoustic neuromas, pituitary tumours, arteriovenous malformations (AVMs), glomus tumours and recurrent gliomas.
However, treating small volume tumours is always a challenge. There are multiple challenges associated with delivering therapy to a small tumour such as the error associated with small field dosimetry, hindrance to the optimisation process and difficulties in dose calculation and dose delivery. Reference Das, Ding and Ahnesjö11
The complication increases further if IMRT or volumetric modulation (VMAT/RA: rapid arc) is used. For VMAT technique dose rate, MLC (multi-leaf collimator) speed and gantry speed need to be synchronised whereas in the case of IMRT technique, dose rate and MLC position and speed need to be synchronised. Therefore, stereotactic radiosurgery using the VMAT technique is a complex process hence its advantage needs to be evaluated against the classical stereotactic delivery technique of 3DCRT or dynamic conformal arc therapy (DCAT). Most modern linear accelerators offer both type of beams, flattened (FF) and unflattened (FFF) photon beams. It is well established from the literature presented earlier by different research groups that FFF beams have no dosimetric advantage over the FF beams; however, it reduces the treatment time significantly. As in general all modern linear accelerators offer both kinds of beams, we incorporate this as a variable in this study.
This study has been designed to evaluate the dosimetric efficacy of VMAT in comparison to DCAT and 3DCRT for very small volume (≤1 cc) and small volume (≤3 cc) tumours for FF and FFF 6 MV beams.
Materials and Methods
Twenty-one patients who received single-fraction stereotactic radiosurgery during a time frame of 6 months (September 2018–February 2019), using either VMAT, DCAT or 3DCRT at our clinic, were considered for this retrospective study with seven patients each in <1, 1–2 and 2–3 cc volume category. Attempt was made to include equal sample size for each arm of the study which resulted in seven patients under the three different volumes. All patients received single-fraction stereotactic radiosurgery. The pathological distribution of the cases was ten cases of AVM, two cases of solitary brain metastasis, three cases of acoustic neuroma, four cases of meningioma and two cases of facial nerve hemangioma. An immobilisation mask was fabricated for all patients. We used a frameless mask from Brainlab (Brainlab AG, Gmbh, Bavaria, Munich, Germany). CT simulation was carried out in 256 slice CT scanner (Philips-Tru Flight PET/CT system, Ohio, Cleveland, United States) for all the patients using the Brainlab localiser frame (Brainlab AG, Gmbh). Contrast CT scan of 1-mm slice thickness is acquired as per the institutional protocol. All patients underwent a radiotherapy planning MRI (Philips Ingenia 3 T, North Brabant, Eindhoven, The Netherlands) (256 × 256 matrices, 1-mm slices). Both the image sets were transferred to the iPlan work-station (Brainlab AG, Gmbh, Bavaria, Munich, Germany) for image fusion and contouring. Stereotactic coordinate system was obtained from the Brainlab frameless base plate on CT overlay. Gross tumour volume (GTV) and the organs at risk (OARs) are delineated by a radiation oncologist in the CT–MR-fused image dataset in the iPlan radiotherapy image (v4.1.2) module which is reviewed by a neurosurgeon and a second radiation oncologist before being finalising for planning. The planning target volume was generated by expanding the GTV uniformly by 0·7 mm. Further fused dataset was transferred to the Eclipse V15.3 (Varian Medical System, Polo Alto, CA, USA) planning system for VMAT planning without changing the isocentric information. Patient plans were carried out by experienced dosimetrist using FF beam and delivered after the appropriate quality assurance. The delivery technique (RA/DCAT/3DCRT) was decided as per the institutional protocol. The choice of technique which was executed during therapy delivery is identified as the standard arm, where other two competing techniques are identified in the experimental arm. The plans in the experimental arm were created by the same dosimetrist. All plans (RA/DCAT/3DCRT) were repeated for FFF beam. A total of 126 plans were generated and compared in this dosimetric study. The competing plans were attributed to FF and FFF beams, VMAT (RA_FF and RA_FFF), dynamic conformal arc DCAT (DCA_FF and DAC_FFF) and 3DCRT (Static_FF and Static_FFF).
Plans were evaluated for target coverage (percentage of target volume receiving prescribed dose: V100%), Paddik conformity index, homogeneity index, Reference Kataria, Sharma, Subramani, Karrthick and Bisht12 dose gradient for 50% dose fall-off, total MU and MU/dose ratio (IMRT factor), normal brain receiving >12 Gy dose, dose to the OAR, beam ON time and dose received by 12 cc of the brain. Reference Benedict, Yenice and Followill13 All plans were verified for point dose. Reference Sarkar, Ghosh, Sriramprasath, Basu, Goswami and Ray14
VMAT planning
All plans were carried out using multiple VMAT arcs. A typical plan is presented in Figure 1a. The VMAT plans were based on two coplanar full arcs and three non-coplanar partial arcs (couch angle 270°, 315° and 45°). The arc configurations were G181CW178C30, G181CW178C330, G350CW179C30T270, G10CCW185C30T45 and G350CW175C30T315. In these configurations, G stands for the gantry rotation start angle, CW/CCW are for clockwise and counterclockwise gantry motions, C is collimator angle and T is the table angle. The optimisation started with an arbitrary set of parameters and was carried out until the OAR dose could be reduced without compromising on-target volume coverage and the increase of excess dose (≥110%).
Figure 1. (a) VMAT, (b) dynamic conformal arc therapy and (c) 3DCRT beam arrangement for 12 Gy prescription of 1 cc tumour of right acoustic neuroma. Abbreviations: VMAT, volumetric modulated arc therapy; 3DCRT, 3D conformal radiotherapy.
DCAT planning
A typical DCAT plan is exhibited in Figure 1b. Arc angles were replicated from RA plans, two coplanar full arcs and three non-coplanar partial arcs (couch angle 270°, 315° and 45°). However, the collimator angles were iterated to increase the dose conformity. Plans were normalised to the isocentre.
3DCRT planning
The 3DCRT planning strategy was adopted from early literature. Reference Munshi, Sarkar, Roy, Ganesh and Mohanti10 Multiple static coplanar and non-coplanar beams were used to obtain the dose coverage to planning target volume (PTV) keeping the same isocentre. The median number of beams was 17 (range 13–20). Other than the main fields, supplementary fields were added as per the requirement. Beam shaping was done in beam’s eye view using an MLC. Beam weights were determined to yield the desired dose coverage and tolerance-limited doses to OARs. A typical 20-beam plan is presented in Figure 1c. The MLC was adjusted in beam’s eye view with an automatic margin of 0·1 cm and subsequently modified as per the clinical requirements. Further manual shaping was done if required. All plans were verified for the dose to MU conversion by delivering them in a solid phantom loaded with 0·01 cc ion chamber with true gantry angles and true couch angles. Reference Benedict, Yenice and Followill13
Figure 1 shows the planning of three different techniques for the same patient.
Result
Mean ± standard deviation (SD) of the dose prescription was 16·1 ± 3·8 Gy (range 12–20 Gy). The average ± SD tumour volume (TV) in ≤1, ≤2, ≤3 cc and overall (<1, 2, 3 cc>) scenario was 0·6 ± 0·2, 1·5 ± 0·3, 2·5 ± 0·3 and 1·5 ± 0·8 cc, respectively. Table 1 shows the target coverage, conformity index and heterogeneity index for different beam models, categorised as a function of the TV. The average target coverage for all plans, when independent TV ranges are considered, was seen to vary between 91·7 ± 8·8% and 99·1 ± 0·8%, with the minimum being exhibited by Static_FF for 1 cc tumour. The highest target coverage was exhibited by DCA_FF 99·1 ± 0·8% for a 3 cc tumour. When averaged over all volume ranges, delivery techniques and beam models, a mean target coverage of 96·4 ± 4·5 (range 95·7 ± 6·1–97·5 ± 2·9%) was seen. The Paddik conformity index averaged over all volume ranges <1, 2, 3 cc> was seen to vary between 0·55 ± 0·08 and 0·68 ± 0·04 with minimum and maximum being exhibited by DCA_FFF for 1 cc and Static_FFF/RA_FFF for 3 cc tumours, respectively. The heterogeneity index varies between 1·1 ± 0·0 and 1·3 ± 0·1. The minimum heterogeneity was shown by RA_FFF for a 3 cc tumour whereas the maximum was shown by DCA_FF for 1 cc and others. The Mean (±SD) MU for RA_FF, DCA_FF and Static_FF were observed as 5,785·3 ± 2,305·3, 3,239·3 ± 831·3 and 3,250·6 ± 830·3 MU, respectively. For RA_FFF, DCA_FFF and Static_FFF, the values were 6,240·2 ± 2,477·5, 3,311·0 ± 835·3 and 3,338·0 ± 866·2 MU, respectively. The mean IMRT factor, averaged over all volume ranges, for RA_FF, DCA_FF and Static_FF was observed at 3·5 ± 0·8, 2·0 ± 0·2, 2·0 ± 0·2, respectively; RA_FFF, DCA_FFF and Static_FFF were 3·8 ± 0·8, 2·1 ± 0·2, 2·1 ± 0·2, respectively. Figure 2 shows the variation of MU and IMRT factor as a function of the technique and beam model. No significant variation was observed on the IMRT factor as a function of the TV. Figure 3 presents the brain volume receiving more than 12 Gy for RA_FF, DCA_FF, Static_FF. RA_FFF, DCA_FFF and Static_FFF were 3·4 ± 2·5, 3·3 ± 2·2, 3·2 ± 2·2, 3·4 ± 2·5, 3·3 ± 2·1 and 3·1 ± 2·1, respectively. The dose fall-off gradient varied in the range of 0·33–0·42 (% per millimeter), 0·35–0·40 and 0·38–0·45 for 1, 2 and 3 cc tumours, respectively, and is presented in Figure 4. No significant difference in dose gradient was observed between the FF and FFF beams. The difference in OAR doses as a function of the beam model and delivery technique is presented in Table 2. For TV ≤1 cc, the RA brainstem doses vary between 24 ± 37·2% and 53·4 ± 50·2% between all TVs and beam models; for DCA_FFF, the brainstem dose range was between 24·7 ± 39·7% and54·7 ± 55·5%. Similarly, for Static FF and FFF, the range was from 24·7 ± 37·9% to 55·4 ± 55·6%. Other OARs were also very slightly over the beam models and delivery techniques. In all patients, it was possible to achieve the TG-101 recommended dose constraints. Reference Flickinger, Kondziolka and Lunsford15 Point-dose measurement showed a wide variation as a function of the TV. All delivery techniques in the ≤1 cc range show a variation in the range of 7·1 ± 3·4–4·5 ± 3·1%; in 2 cc range 5·1 ± 2·7–2·5 ± 3·4%; and in 3 cc range maximum variation was presented as 3·1 ± 2·1–1·7 ± 1·4%. FFF shows a better matching for all delivery techniques than FF beam by 1·3 ± 1·1% when averaged over all TVs.
Table 1. Target coverage, conformity index and heterogeneity index as a function of different tumour volumes and deferent delivery techniques
Figure 2. MU/cGy (dose to MU factor) as a function of different delivery techniques and beam models (tumour volumes as parameters).
Figure 3. Brain volumes receiving >12 Gy as a function of different delivery techniques and beam models (tumour volumes as base parameter).
Figure 4. Gradient of 50% dose fall-off as a function of different delivery techniques and beam models (tumour volume as key parameter).
Table 2. Organ at risk doses as a function of different delivery techniques and beam models (tumour volume as key parameter)
Discussion
Stereotactic radiotherapy is an effective way of delivering a high dose to a cranial lesion. The invasive frame-based stereotactic technique is being replaced by non-invasive frameless stereotaxy which increases the convenience of the patients. A cranial lesion can be isolated or can be surrounded by a different OAR. Therefore, tumours can be divided into OAR-challenged and OAR-unchallenged groups. In an OAR-challenged group, dose build-up to the target volume is dependent on the OAR tolerance doses whereas in the OAR-unchallenged category, dose build-up is not at all, or is feebly affected by the dose to OARs. Early investigators reported that the ratio of OAR-challenged and OAR-unchallenged tumours in a large ensemble of the plan is nearly the same. Reference Sarkar, Munshi, Manikandan, Anbazhagan, Ganesh and Mohanti9 Although in an OAR-unchallenged category it is not required to reduce the doses to the specific organ, however, one needs to reduce the spillage dose which affects the normal brain parenchyma, which can lead to a tissue necrosis. Reference Flickinger, Kondziolka and Lunsford15–Reference Khataniar, Sarkar, Gupta, Agrawal, Mohanti and Munshi17 Therefore, in stereotactic planning, it is always challenging to reduce the spillage dose to the normal brain parenchyma irrespective of the presence of OAR. An improper choice of arc angle and beam angles, beam weight or MLC shape may contribute to a suboptimal treatment plan which may lead to post-therapy deterioration in the quality of life in terms of oedema, facial palsy, neurocognitive vision, hearing losses, etc. Reference Kondziolka, Nathoo, Flickinger, Niranjan, Maitz and Lunsford18,Reference Jalali, Mallick and Dutta19 In all the plan approaches for both beam models, the treatment planner was able to reduce the OAR doses to below the specified tolerance dose specified by the AAPM task group or QUENTEC. Reference Benedict, Yenice and Followill13,Reference Marks, Yorke and Jackson20
Our planning strategy
Several attempts have been made so far by a number of investigators to evaluate the plan quality as a function of different beam models (FF and FFF), MLC width and delivery technique, etc., which contribute towards a minimal variation of the dose distribution with varying parameters like MLC width, beam model and delivery technique. Reference Laing, Bentley, Nahumb, Warrington and Brada21–Reference Monk, Perks, Doughty and Plowman28 However, the dose distribution strongly depends upon the experience of the planning dosimetrist, Reference Sarkar, Munshi, Manikandan, Anbazhagan, Ganesh and Mohanti9 it was established early on that the ideal MLC width for 3DCRT-based stereotaxy required an ideal MLC width of 1·5–1·8 mm Reference Peng, Kahler and Samant29 which is probably attributed to the 3DCRT or unmodulated treatment plans. Advance techniques like VMAT are very minimally influenced by the MLC width and beam characteristics. Our study reveals a minimal variation between FF and FFF beams in the dosimetric characteristics as a function of the delivery technique. The only difference observed was for MU where FFF offered a slightly higher MU than FF beam for RA techniques until a 2 cc volume. For a 3 cc target volume, the situation reverses.
It was always challenging to build up the dose, especially for irregular-shaped target volumes. Reference Laing, Bentley, Nahumb, Warrington and Brada21 With all the advancements in the positioning systems like six-dimensional motion-enabled robotic couches and three-dimensional imaging techniques like cone beam CT, the geometrical accuracy of the dose delivery can be ensured. Reference Peng, Kahler and Samant29,Reference Sarkar, Munshi, Krishnankutty, Ganesh and Kalyan30 However, the accuracy of the MLC position (for static IMRT and 3DCRT delivery) and MLC position as a function of time (arc therapy and DCAT) is an important factor. Reference Manikandan, Sarkar, Holla, Vivek and Sujatha31
It has been established by earlier authors that a systematic error of 1 mm for all MLCs might lead to a dose error of more than 5%. Reference Manikandan, Sarkar and Nandy32 Although modern linear accelerators work in a servo-controlled feedback mechanism to ensure the reproducibility of all essential parameters like dose rate, gantry position and MLC position as planned in treatment planning system (TPS). Reference Manikandan, Sarkar and Nandy32 Nevertheless, it is essential to ensure the accuracy of all the dose delivery parameters and variables, which is done in this study by dose to MU verification. The most vulnerable part of dealing with small tumours is the dose calculation accuracy in the planning system. There is no exact method to verify the accuracy of the dose calculation. MU to dose verification measurements are also a compromised technique because even with the smallest chamber like 0·01 cc ion chamber or diode chamber, the volume is comparable with the TV and a substantial dose averaging can take place during measurement. Therefore, with an acceptable result in point dose measurement, the accuracy is questionable. Nevertheless no better method of patient-specific quality assurance is available so far other than the dose to MU verification as film or ion chamber array dosimetry is more complex with the first one being a cumbersome process and the second one being compromised by inaccuracy.
All plans in this study have been carried out by a lone dosimetrist having experience of 15 years and previous planning of 200 stereotactic cases. This study establishes that dose distribution varies between different delivery techniques like VMAT, DCAT and 3DCRT for an experienced dosimetrist.
This article is first of its kind to present the dosimetric character of small and very small tumours as a function of the delivery technique and beam characteristic (flatten/unflatten). Often dosimetrist, therapists, physicist and clinicians have no clue on how to proceed for small and very small tumours. This report may give confidence to the group for a seamless execution in the radiotherapy treatment and choice of technique as well.
The potential demerit of this study is the number of patients in each arm is less, that is, only seven. This is attributed to the fact that such patients are very sparely populated, hence with a significant weight of time we could not generate a moderate patient number. No hypothetical tumour was used in this study.
Conclusion
This study establishes the equivalence between FF and FFF beam models and different delivery techniques` for stereotactic radiosurgery in small TVs in the range of ≤1 to ≤3 cc. Dose conformity, heterogeneity, dose fall-off characteristics and OAR doses show no or very little variation. The only possible advantage for the FFF beam could be fast dose delivery due to higher dose rate (1,400 MU/minute) than flatten beam (600 MU/minute). However, it was established by early researchers that this would reduce the patient’s in-room time only by 15%. Reference Sarkar, Pradhan and Munshi3
Acknowledgements
The authors sincerely thank Dr Biplab Sarkar, PhD, Chief Medical Physicist, Apollo Gleneagles Hospitals, Kolkata, India, for his critical inputs during the experiment and in manuscript preparation and Ms. Sandhya M. Rao for the support extended in language editing.
