Dosimetric quantification of dose fall-off in liver SBRT planning using dual photon energy IMRT

Abstract

Introduction

The purpose of this study was to dosimetrically compare 6 MV, 10 MV and a hybrid of 6 and 10 MV photon beam energies in liver stereotactic body radiotherapy (SBRT) patients using a fixed-field IMRT technique. The objectives of the study were to evaluate dosimetric differences in the target volume and investigate if dose fall-off could be improved with 10MV beam energy.

Methods and materials

Sixteen (n=16) liver SBRT patients previously treated using a non-coplanar, fixed-field IMRT technique with 6 MV were replanned using 10 MV and dual photon energy (DE). Plans were generated in Pinnacle³ using a Novalis Tx with HD120 MLC. For each patient, three plans with the same beam geometry were created using 6 MV, 10 MV and DE. For DE plans, the central axis effective depth from each beam was calculated and the values averaged. Beams with an effective depth greater than the average were assigned to 10 MV. All patients were optimized with the same planning objectives and normalized such that 98% of the target received 100% of prescription dose. Metrics used for comparison were the homogeneity index, conformity indices, and dose fall-off parameters at various isodose levels.

Results

The three techniques showed comparable PTV conformity and inhomogeneity for all patients—differences in the median values 「0.6%. With regard to dose fall-off, no statistically significant differences were noted among the techniques for R₈₀, R₆₀ and R₅₀; however, 10 MV showed statistical significance in the lowest median values of R₄₀, R₃₀, and R₂₀. Ten MV and DE plans also demonstrated a statistically significant reduction in the total number of monitor units (MU) of 14.9% (p 「0.01) and 12.0% (p 「0.01) as compared to 6 MV, respectively.

Conclusion

Both dual energy and 10 MV photon beams had similar PTV dosimetric characteristics to 6 MV for liver SBRT but findings show faster dose fall-off for 10 MV and DE plans at the 40%, 30%, and 20% prescription isodose levels.

1. Introduction

Due to the tumor size and location of liver cancers, some cancer patients are suitable for surgery.¹ For those who are not surgical candidates, radiation therapy, especially stereotactic body radiation therapy (SBRT) for early stage primary lesions or oligometastatic disease, is a viable treatment for local control of liver disease.² With rapid development and focus on treating extracranial lesions with stereotactic radiation techniques and a broader understanding of liver radiobiology, liver SBRT has since become a treatment modality of choice.³ In SBRT, the primary dosimetric goal is to concentrate a large amount of dose to the target volume with a fast fall-off in the surrounding tissues. In SBRT, definitive courses of radiotherapy consist of 3 to 5 fractions of high, ablative doses typically greater than 8Gy per fraction. Due to the high doses per fraction, compact dose distributions are required to reduce the incidence of late effects in the surrounding normal tissues.⁴

Different approaches have been investigated to improve the dose fall-off such as changing beam margin⁵, using non-coplanar beams⁶ and increasing target dose inhomogeneity—i.e. prescribing to a lower isodose line. For SBRT, 6 MV photon beams are typically used due to the small beam penumbra of an open-field beam compared to other higher photon energies in conventional linear accelerators. However with the use of IMRT or VMAT, the effective penumbra realized in the treatment plan may not be so clear due to the impact of beam fluence optimization.⁷ Although SBRT treatments are typically designed utilizing 6 MV beams with intensity modulation, higher energies such as 8 to 12MV beams may be used before secondary malignancies from neutron contamination can occur.⁸ With newer digital linear accelerators possessing more than dual photon energy capabilities—i.e. 6 MV, 10/15 MV, and 18 MV, the use of intermediate photon energies such as 10 MV may prove to be beneficial.⁹

With this in mind, the primary aim of the study is to quantify the compactness of the target dose distribution for liver SBRT plans using solely 6 MV, solely 10 MV or combination of 6 and 10 MV. For the purpose of this study, compactness is quantified in terms of fall-off for both higher isodose levels (80%, 60%, 50%) and lower isodose levels (40%, 30%, 20%). One plan has a more compact dose distribution than another if it has faster fall-off for the same level of target coverage. The objective of the study is to quantify the effect of photon beam energy selection on liver SBRT dose distributions and treatment plan quality. The study seeks to characterize and quantify any potential benefits when using 10 MV.

2. Materials and Methods

2.1. Patient Selection and Simulation

This retrospective treatment planning study involved sixteen (n=16) patients previously treated for liver disease using a SBRT delivery technique. Each patient was immobilized using the Body Pro-Lok™ system (CIVCO, Orange City, IA) with their arms up, and abdominal compression via a compression plate was applied to inhibit breathing motion.¹⁰ All patients underwent a free-breathing helical CT scan for simulation using a 2.5mm slice thickness. The gross tumor volume (GTV) was segmented by a board certified radiation oncologist using the simulation CT study. The planning target volume (PTV) was defined by adding a setup margin, typically 5mm in the anterior-posterior and lateral direction and 10 mm in the superior-inferior direction. The dose prescription and fractionation varied across the sample of patients and ranged from 54 Gy (3 fractions) to 50 Gy (5 fractions).

2.2. Beam Energy Selection

Initial treatment plans were previously planned using a 6 MV photon energy for all beams. Patients were then replanned using 10 MV photon energy for all beams. For the dual energy (DE) photon plans, a simple approach based on the central axis depth to the isocenter for a given beam was used to determine whether 6 MV or 10 MV energy was used. For each patient, the effective depth along the central axis of the beam to the isocenter was determined. The effective depths of all beams for a given patient were averaged and beams possessing an effective depth less than the average were assigned an energy of 6 MV—the remaining were assigned 10 MV. This approach was used to ensure an even split between 6 MV and 10 MV photon energies for the majority of the patients. The rationale for using 10MV for the larger effective depth is principally based on the higher penetrative power of 10 MV over 6 MV.

2.3. Treatment planning and planning objectives

All patients were planned in the Pinnacle³ treatment planning system (TPS) version 9.8 (Philips Medical, Fitchburg WI) using 8 to 11, non-coplanar fixed-field IMRT beams with 6 MV, 10 MV and DE. The beam orientations were selected to minimize beam overlap and OAR irradiation. For plan optimization and dose calculation, DMPO optimization was employed with a final dose calculation using the collapsed cone, adaptive convolve algorithm. Treatment plans were calculated using heterogeneity corrections to ensure appropriate dose calculation in the lung. Plans were computed with a dose grid voxel size of 2 x 2 x 2 mm³. The gantry, collimator and couch angles varied per patient, however, not within a patient when comparing 6 MV, 10 MV, and DE plans. All other plan settings were equivalent between photon beam energies but varied among patients. All plans were optimized with similar planning objectives and normalized such that 98% of the PTV received 100% of the prescription dose. For the optimization, the DE plans were optimized in Pinnacle³ similarly to the 6 MV and 10 MV plans. Currently, Pinnacle³ allows optimization of mixed photon beam energies within a plan trial. Target planning optimization objectives included a minimum and maximum dose for PTV, a minimum dose to GTV, and maximum dose for PTV minus GTV. Various ring structures were also created with various maximum dose constraints to improve dose fall-off down to 20% of the prescription dose. For organs at risk—primarily the liver, optimization objectives were used such that the organ fulfilled the constraints set forth by the AAPM Task Group 101 normal tissue constraint guidelines.¹¹ For all of the plans, the key structures in the optimization were the ring structures as the normal tissue constraints were easily met. Plans were optimized to achieve target dose coverage, conformity, and rapid dose fall-off.

2.4 Evaluation of Plans and Statistical Analysis

Multiple metrics were employed to evaluate the quality of the treatment plans. Dose-volume histograms (DVH) were used to quantify the D_0.2cm³ which is a measure of the near-maximum dose quantified to 0.2cm³ of the structure. Additionally, the following metrics were computed for the PTV. PTV homogeneity was quantified by a homogeneity index defined as follows:

$$HI=\frac{D_{0.2c{m}^3}}{D_{pres}}$$ (1)

where D_pres is the prescription dose. The prescription dose conformality was quantified using the conformation number (CN)¹² and is defined as follows:

$$CN=\frac{V_{PTV,pres}^2}{V_{PTV}{V}_{pres}}$$ (2)

where V_PTV,pres is the volume of the PTV receiving at least the prescription dose, V_PTV is the volume of the PTV, and V_pres is the volume of the prescription isodose volume. A CN value of 1.0 denotes the optimal prescription dose conformity—accounting for spatial differences between the V_PTV and V_pres. The PITV was used as a secondary measure of prescription dose conformity due to its universal use and was computed as follows:

$$PITV=\frac{V_{pres}}{V_{PTV}}$$ (3)

Dose fall-off was quantified by the R_x and defined as follows:

$$R_X=\frac{V_{X\%pres}}{V_{PTV}}$$ (4)

where V_{X% pres} is the volume of tissue receiving at least X% of the target prescription dose. Assuming fixed target coverage, a lower R_X value shows a faster fall-off of the dose between the 100% and X% dose levels. In this study, the following isodose levels were evaluated to characterize the dose fall-off: R₂₀, R₃₀, R₄₀, R₅₀, R₆₀, and R₈₀. For statistical analysis, the statistical analysis package in Origin® 9.1 (OriginLAB^© Northhampton, MA) was used. Statistical differences were evaluated using a paired sample Wilcoxon signed rank test with a significance level of 0.05.

3. Results

For all of the patients, the average PTV was 70.8 ± 52.9cm³ with a range from 23.3cm³ to 242.5cm³. The most common dose prescription fractionation was 50 Gy in 5 fractions. Table 1 summarizes the average, minimum, and maximum central axis effective depth of the beams for each patient. Overall, the average effective depth of beams for all patients was 12.5 ± 2.3cm. The patient sample average minimum and maximum effective depths were 8.2 ± 2.0cm and 20.1 ± 4.2cm, respectively.

Table 1.

Average, minimum and maximum central axis effective depth of beams for sample patients.

	Avg. Depth (cm)	Min. Depth (cm)	Max. Depth (cm)
Pt. 1	11.2	5.8	18.1
Pt. 2	10.6	6.2	20.9
Pt. 3	14.1	9.2	20.8
Pt. 4	12.5	9.1	16.6
Pt. 5	13.4	9.3	28.4
Pt. 6	16.0	10.3	20.6
Pt. 7	16.6	11.3	24.7
Pt. 8	13.5	7.6	24.5
Pt. 9	12.9	10.0	18.3
Pt. 10	7.7	4.0	14.8
Pt. 11	12.5	6.9	21.6
Pt. 12	8.7	6.0	11.4
Pt. 13	10.9	7.1	18.5
Pt. 14	13.9	10.1	23.9
Pt. 15	13.1	9.6	21.0
Pt. 16	12.8	9.5	17.2

With regards to the planning target volume parameters, Table 2 shows the median and standard deviation of the evaluated metrics for the 6 MV, 10 MV, and DE for all patients. For all three techniques, no statistically significant differences were noted among the median values of the three techniques for the PITV, CN and HI. This validates that all three techniques are able to produce similar prescription dose conformity and homogeneity—as well as target coverage due to the normalization constraint. In general, median values of 1.022 in the PITV and 0.940 in the CN illustrate high prescription dose conformity to the target volumes for all three techniques. Median HI values of 1.28 for all three techniques show that the optimal peaking dose is about 128% of the prescription dose—optimal is defined as the HI that produces the best dose fall-off outside of the PTV. Concerning the total plan monitor units (MU), a statistically significant reduction was shown in the median values of 10 MV and DE as compared to the 6 MV plans. In particular, a 14.9% (p「0.01) and 12.0% (p「0.01) reduction was noted in the 10 MV and DE, respectively. This finding could be of clinical significance as lower MUs will inevitably shorten treatment times for patients.

Table 2.

Planning target volume (PTV) median and standard deviation (SD) values of the selected metrics for 6MV, 10MV and Dual Energy (DE) IMRT plans for the 16 sample patients. Total treatment plan MUs are also listed for each approach.

		6X		10X			Dual Energy (DE)
Structure	Metric	Median	SD	Median	SD	%∆ 6MV	Median	SD	%∆ 6MV
PTV	PITV	1.022	0.020	1.021	0.020	-0.1% (NS)	1.016	0.020	-0.6% (NS)
	CN	0.940	0.018	0.941	0.019	0.1% (NS)	0.946	0.016	0.6% (NS)
	HI	1.28	0.05	1.28	0.05	-0.5% (NS)	1.28	0.04	-0.5% (NS)
	MU	3009	585	2619	491	-14.9% (p「0.01)	2687	539	-12.0% (p「0.01)

Abbreviations: SD standard deviation, NS = no statistical significant difference (p-value 『 0.05)

Evaluating the dose fall-off, Table 3 shows the results of the median and standard deviation values of the selected dose fall-off parameter for all patients. Figure 1 shows an axial slice of a sample liver SBRT patient using a 6 MV, 10 MV and DE technique. Overall, the results show that there was no noted statistically significant difference among the median values of the three techniques for the R₈₀, R₆₀, and R₅₀. The median values of these parameters did not vary by more than 1.0% between any given pair of techniques. These findings appear to indicate that the dose fall-off between the 100% and 50% of the prescription dose are in essence invariant among energy. However, statistically significant findings were noted for the dose fall-off at the lower dose levels. In particular, statistically significant improvements in the R₄₀, R₃₀, and R₂₀ when using 10 MV and DE as compared to 6 MV were noted. The improvements were maximized using 10 MV. Specifically, using 10 MV showed median improvements of 10.6%, 18.5%, and 5.6% for the R₄₀, R₃₀, and R₂₀, respectively, while using DE showed improvements of 2.6%, 6.8%, and 2.8% for the R₄₀, R₃₀, and R₂₀, respectively. Although improvements were noted at the lower percentage dose levels, it is important to note that these dose levels still correspond to significant doses due to the high dose per fractions administered during SBRT and may not be trivial. For example, for a prescription of 54Gy in 3 fractions, the 30% isodose line corresponds to 16.2Gy, or 5.4Gy per fraction. Although these doses may not necessarily be correlated to clinical complications, the principles of ALARA (As Low As Reasonably Achievable) still apply.

Table 3.

Dose fall-off metric evaluation for 6MV, 10MV and Dual Energy (DE) IMRT plans for the 16 sample treatment plans.

			6X			10X				Dual Energy (DE)
	Metric		Median	SD		Median	SD	%∆ 6MV		Median	SD	%∆ 6MV
Dose Fall-off	R80		1.66	0.11		1.66	0.11	0.5% (NS)		1.66	0.11	0.0% (NS)
	R60		2.58	0.21		2.60	0.21	0.8% (NS)		2.55	0.22	-1.1% (NS)
	R50		3.58	0.29		3.61	0.30	0.7% (NS)		3.54	0.31	-1.0% (NS)
	R40		6.10	0.69		5.52	0.49	-10.6% (p「0.01)		5.95	0.62	-2.6% (p「0.01)
	R30		12.08	2.22		10.19	1.50	-18.5% (p「0.01)		11.31	1.87	-6.8% (p「0.01)
	R20		21.88	5.85		20.71	4.77	-5.6% (p「0.01)		21.28	4.92	-2.8% (p「0.01)

Abbreviations: SD standard deviation, NS = no statistical significant difference (p-value 『 0.05)

Figure 1.

An axial slice of a sample liver SBRT patient is shown for 6 MV (left), 10 MV (center) and dual energy DE (right) planning techniques. Prescription dose for this sample patient is 45 Gy in 3 fractions. The ring structure shown in the figure is used as an evaluation structure to ensure that the 50% prescription isodose line is within 2.0 cm from target volume edge. Isodose line levels are denoted by color below the figure.

4. Discussion

In this study, the effect of higher beam energy on liver SBRT target dose compactness was quantified in a non-coplanar, fixed-field IMRT delivery technique approach. The significant findings of the study are that dual energy and 10 MV photon energy show a faster dose fall-off for the prescription dose levels of 40%, 30%, and 20%--with 10 MV showing the best improvement of up to 18.5% in the R₃₀ parameter. Clearly, the improved compactness from the faster dose fall-off translates into a reduction of dose in involved and adjacent normal tissues. Any improvements in SBRT treatment plan quality by improving the dose fall-off may have implications for SBRT treatments. For instance, any potential gains in the dose conformity and dose fall-off may allow for a potential dose escalation to the target volume. Findings from a published multi-institutional phase 1/2 liver SBRT study showed high local control rates were correlated with more aggressive dose fractionations due to higher biological equivalent doses being delivered.²

Due to the correlation between high dose conformity/dose compactness and successful clinical outcomes of liver SBRT, an approach commonly undertaken to achieve dose conformity and compactness is the use of a greater number of beam directions—i.e. mimicking stereotactic radiosurgery delivery principles. Liver SBRT clinical trial groups have recommended the use of more than 5 IMRT beams when delivering SBRT.¹³ In this study, all of the patients were treated with 8 to 11 non-coplanar beams that were oriented in order to achieve maximum separation between beams to reduce beam overlapping. Previous studies have shown that implementing a non-coplanar beam delivery geometry may significantly improve the treatment plan quality for liver SBRT patients when using an IMRT technique.¹⁴ However, the optimal beam angle selection strategy is not straightforward and largely dependent on target location, patient size, and table/gantry clearance capabilities. This area of automatic beam orientation optimization is an active field of research.¹⁵

It is important to note that simply increasing the number of coplanar beams may not be effective as has been shown by the comparison between intensity modulated radiation therapy and VMAT in lung¹⁶ and spine¹⁷ SBRT planning. More recently, a novel 4π non-coplanar radiation delivery approach in which the radiation delivery space is increased has been introduced for delivery of liver SBRT. Dong et al. have shown that by using a 4π non-coplanar radiation delivery geometry the volume of tissue receiving at least 50% of the prescription dose decreases more rapidly as the number of beams increase in a non-coplanar arrangement as compared to a coplanar arrangement.⁶

In reviewing published literature, the effect that photon beam energy plays on dose compactness for SBRT treatment planning has not been studied. Although this study solely used fixed-field IMRT beams, VMAT plans with multiple arcs using various photon beam energies could be evaluated for dose conformity and dose compactness. A key advantage of a VMAT approach would be the faster overall treatment delivery time. However, even for our fixed-field approach, the noted reduction in MUs when using 10 MV would also lead to a reduced beam-on time when compared to 6 MV. Based on our findings and newer novel delivery approaches, it appears that the use of higher photon beam energy in conjunction with a 4π non-coplanar radiation delivery may provide optimal SBRT dose distributions for liver SBRT.

5. Conclusion

Because of the high ablative doses used in SBRT, high quality SBRT treatment plans should ensure fast dose fall-off and compact dose distributions. This retrospective treatment planning study investigated the potential improvement in dose fall-off for SBRT liver patients treated with a fixed-field, non-coplanar IMRT technique by using 6 MV, 10 MV and a dual energy (6 MV/10 MV) approach. The use of 6 MV, 10 MV, and dual energy provided comparable prescription dose conformality, target coverage, target homogeneity, and dose fall-off between the 100% and 50% prescription dose levels. Ten MV was shown to provide the best overall dose fall-off from the 40% to 20% of the prescription dose level as quantified by statistically significant reductions in the R₄₀, R₃₀, and R₂₀ as compared to 6 MV. Furthermore, 10 MV also demonstrated the largest statistically significant reduction in total plan monitor units for the three energy techniques. Overall, 10 MV photon beam energy should be considered an effective energy for the clinical treatment of liver patients with SBRT delivery utilizing an IMRT technique.