Dosimetric evaluation of abdominal compression as a method to reduce the incidence of radiation-induced pneumonitis in lung SBRT treatment

Abstract

Abdominal compression has been shown to reduce the extent of lung tumor motion but the dosimetric impact of the approach is still in need of investigation. The current work analyzes the impact of various changes in PTV volume on key metrics of the final dose distribution to normal lung. To add clinical perspective, we also provide NTCP calculations for grade 2+ pneumonitis for each case. For a total of seventeen cases, the original ITV/PTV was reduced by systematically varied amounts and SBRT plans using dynamic conformal arc and VMAT techniques were created. DVH analysis for the normal lung comparing the original plan to the one with the ITV reduced by up to 10 mm shows that the average reduction of V5, V20 and mean lung dose is 3.8%, 2.0% and 1.1 Gy, respectively, for the conformal arc plans. Corresponding values for the VMAT plans were 3.9%, 1.9% and 1.2 Gy respectively. The mean NTCP drop for the conformal arc plans was 2.0% while it was 1.9% for the VMAT plans. These results suggest that abdominal compression has a modest impact on NTCP and on dosimetric parameters typically used to predict the risk of radiation pneumonitis in patients undergoing lung SBRT.

1. Introduction

The treatment of lung cancer with stereotactic body radiation therapy (SBRT) has been the subject of numerous studies and clinical trials^1-7 and has shown significant promise for treatment of primary and metastatic lung lesions. An SBRT regimen delivers very high, ablative doses in only 3-5 fractions using a very conformal dose distribution. This is done to minimize radiation damage to nearby critical structures and healthy tissue, thus reducing the risk of sub-acute and late toxicity.

An obvious radiation sensitive structure when using radiotherapy for lung cancer is the surrounding normal lung tissue. When treating a tumor in the lung, the nature of x-ray-based radiotherapy requires that normal tissue surrounding the tumor also receive radiation dose. Radiation pneumonitis is an important possible side effect of lung SBRT and this complication has been shown to correlate with various dose-related metrics from the treatment plan^8-10. Because the lung is a large, paired parallel organ, it can tolerate a certain amount of radiation injury without adverse effect. However, injury to larger volumes of lung parenchyma will increase the probability of clinically significant complications, such as pneumonitis¹¹. It is therefore important to limit the amount of normal lung tissue included within even the low dose wash in order to minimize the possibility of radiation-induced pneumonitis.

Because intrapulmonary tumors are typically mobile due to respiratory motion, the entire potential tumor motion space (i.e. internal target volume – ITV) is typically irradiated to ensure that the prescribed dose is delivered to the target. One strategy employed for limiting the amount of irradiated normal tissue is to reduce the tumor motion envelope as much as possible, thereby reducing the overall treatment volume. Abdominal compression, via mechanical pressure applied to the abdomen, is one method commonly used to achieve this goal. While various studies^12-17 have shown that abdominal compression does generally decrease the motion envelope of the tumor, our experience has shown that the approach is not well-tolerated by all patients and, thus, cannot be effectively applied for every patient. For those patients treated with abdominal compression, it has been observed that patients sometimes exhibit additional intra-fractional motion under abdominal compression because of the discomfort associated with the compression^{16, 18}, and due to its effect on their often poor ventilation function. Such patient motion can result in possible treatment interruption and/or inaccuracy of radiation dose delivery, and has led us to study the specific dosimetric gains afforded by abdominal compression.

In this work we perform a dosimetric study on seventeen different patient cases treated with SBRT for primary or metastatic lung cancer at our center. For each patient, systematically varied levels of ITV reduction were modeled to represent the effects of abdominal compression. Clinically important dosimetric parameters such as mean lung dose, V5 and V20 were evaluated for each plan relative to the original, non-compression plan. Additionally, for each motion envelope reduction scenario, an NTCP calculation was performed to quantify the associated change in probability of radiation-induced pneumonitis.

2. Materials and Methods

For this study, we randomly selected 15 patients from a database of roughly 150 total patients treated by SBRT for lung cancer at our institution since 2007. We ensured a reasonable range in tumor size (0.5cc < GTV < 22.9cc) and ITV volume (max=62.6 cc, min=1.7 cc, mean =16.1 cc) and we chose evenly between the two standard dose prescriptions used in our clinic (i.e. 5×10 Gy and 3×18 Gy).Table 1 summarizes the patient characteristics, dose prescription, tumor location, PTV and normal lung volume characteristics (defined as total lung-GTV) for each case studied here. For the patient with the smallest total lung volume in each of the two different prescription groups, we further simulated an extreme case by modeling those patients as if they had undergone a pneumonectomy and had only one lung, with a lesion within. This was done to represent an extreme case to determine whether ITV reduction in such cases might prove beneficial, and explains the additional 2 cases which cause the total cases explored to be seventeen.

In our center, all patients undergo a simulation 4DCT scan with 2.5 mm slice thickness on a GE Lightspeed 16-slice CT simulator (GE Healthcare, Waukesha, WI, USA) with the Varian RPM system (Varian Inc, Palo Alto, CA, USA) used for respiratory signal tracking. This 4D scan is used to define the internal target volume (ITV) by physician contouring on the 4D movie loop. For all plans studied for this project, the ITV was then expanded uniformly by 5 mm to create a planning target volume (PTV). Treatment plans were designed to deliver the prescription dose to this PTV using both dynamic conformal arcs and VMAT. Dynamic conformal arc plans were created using the iPlan treatment planning system (version 4.5, Brainlab AG, Feldkirchen, Germany) for delivery on a Novalis Classic with an M3 micro MLC (Brainlab AG, Feldkirchen, Germany). The energy used for all dynamic conformal arc plan treatments was 6 MV and the Monte Carlo algorithm¹⁹ was used for all dose calculations. The VMAT plans used a single 210-degree arc and were created in the Eclipse treatment planning system with the AAA calculation algorithm (Version 13.5, Varian Medical Systems, Palo Alto, CA), using the 6MV beam on our True Beam linear accelerator, outfitted with an HD120 MLC (Varian Medical Systems, Palo Alto, CA). This investigation was limited to these two planning techniques because our clinical experience has been acquired almost exclusively with these very effective dose spreading techniques, and because we believe the DCA and VMAT approaches to represent well the trend in SBRT treatment of lung towards use of a large number of beams for the specific purpose of spreading collateral dose²⁰. Per our clinical protocol, the calculations were done on the average scan derived from the 4D CT dataset. All treatment plans were required to satisfy organ at risk dose, dose fall-off (R50% and D2cm), conformity index and tumor coverage using the appropriate guidelines for lung SBRT outlined in RTOG 0813 and/or RTOG 0915.

For the purposes of this study, the original ITV used for the patient’s clinical treatment was used to create the baseline plan. Because we do not typically use full abdominal compression, we were able to simulate decreases in ITV by considering data from Han et al.¹². Han showed that, when abdominal compression was used, the average reduction in tumor motion was 0.2 mm in the L/R direction, 0.2 mm in the A/P direction and 1.4 mm in the S/I direction. Using this as a guideline, two subsequent sets of plans were created for each case studied by simulating various levels of abdominal compression leading to ITV reductions of 3 and 7 times the average reduction reported by Han et al. (i.e. 3X => 0.6 mm L/R, 0.6 mm A/P, 4.2 mm S/I and 7X => 1.4 mm L/R, 1.4 mm A/P, 9.8 mm S/I respectively). The choice of reducing the ITV by 7 times the average reported by Han et al. was chosen to mimic an approximately 1 cm reduction in S/I direction, which is also consistent with the maximum value reported by Negoro et al.¹⁶. This allowed us to simulate abdominal compression as if it were actually applied during the whole course of treatment. The new, reduced ITVs were then grown isotropically by 5mm, to generate the new target PTVs used for planning dynamic conformal arc and VMAT plans. Table 2 summarizes the PTV volumes derived from the original ITV, as well as from the two reduced-volume ITVs. For each of the six plans computed for each patient (i.e. Dynamic Conformal Arc and VMAT treatment plans, for No Compression, 3X Compression and 7X Compression situations per patient), the target coverage was maintained identical to the original plan (95% of the PTV covered by the prescription dose and at least 99% of the PTV receiving 90% of the prescription dose). The additional plans were also required to satisfy all normal tissue protocol planning criteria. This resulted in a total of 17 patient cases times 6 treatment plans per case, or 102 total treatment plans evaluated.

Table 2.

PTV volumes resulting from the original and ‘compressed’ ITVs for each patient.

Case	Original PTV Volume (cc)	PTV volume for 3X* reduction in ITV (cc) / (%)	PTV volume for 7X* reduction in ITV (cc) / (%)
1	15.28	10.58 / 30.8%	7.16 / 53.1%
2	30.99	24.38 / 21.3%	18.41 / 40.6%
3	5.60	3.41 / 39.1%	1.92 / 65.7%
4	66.27	46.14 / 30.4%	35.17 / 46.9%
5	20.82	16.61 / 20.2%	12.87 / 38.2%
6	21.80	15.75 / 27.7%	11.99 / 45.0%
7	21.41	19.81 / 7.5%	15.45 / 27.9%
8	16.06	11.14 / 30.6%	7.27 / 54.8%
9	16.06	11.14 / 30.6%	7.27 / 54.8%
10	61.45	51.31 / 16.5%	42.76 / 30.4%
11	86.90	68.42 / 21.3%	55.62 / 36.0%
12	60.01	43.94 / 26.8%	33.66 / 43.9%
13	43.07	28.01 / 35.0%	18.94 / 56.0%
14	141.1	100.88 / 28.5%	81.44 / 42.3%
15	15.71	11.91 / 24.2%	8.68 / 44.7%
16	25.57	20.29 / 20.7%	15.23 / 40.4%
17	25.57	20.29 / 20.7%	15.23 / 40.4%

^*3X and 7X denote the amount by which the original ITV was reduced (based on the average ITV reductions reported by Han et al.)

Table 1.

Volumetric and dosimetric characteristics for each patient

Case	Sex	Age	Disease Type	Lung Lobe Location	Total Dose (Gy)	Number of Fractions	PTV Volume (cc)	Normal Lung Volume (cc)
1	M	71	Primary	LUL	54	3	15.28	4632
2	M	65	Primary	RLL	54	3	30.99	3345
3	F	56	Metastatic	RLL	54	3	5.60	2760
4	M	67	Primary	LUL	54	3	66.27	3950
5	M	65	Primary	RLL	54	3	20.82	5039
6	M	62	Metastatic	RLL	54	3	21.80	5602
7	M	65	Metastatic	RLL	54	3	21.41	4582
8	F	60	Metastatic	LUL	54	3	16.06	1921
9	F	60	Metastatic	LUL	54	3	16.06	940
10	F	71	Metastatic	LLL	50	5	61.45	2400
11	F	70	Primary	LUL	50	5	86.90	4309
12	M	81	Primary	RLL	50	5	60.01	4849
13	F	77	Primary	RUL	50	5	43.07	2157
14	M	81	Primary	RLL	50	5	141.1	3408
15	M	77	Primary	RLL	50	5	15.71	4205
16	F	34	Metastatic	LLL	50	5	25.57	2007
17	F	34	Metastatic	LLL	50	5	25.57	772

For each plan, the mean lung dose (MLD), volume of lung receiving at least 5 Gy (V5), and 20 Gy (V20) were evaluated. In addition, the normal tissue complication probability (NTCP) for grade 2+ radiation pneumonitis was evaluated using the CERR toolkit²¹. The NTCP was calculated using the Lyman-Kutcher-Burman model. The parameters used for the NTCP evaluation were TD₅₀ = 45 Gy, m = 0.52, and n =1, based on the work of Sonke et al.²². Since two different fractionation schemes were used for the patients investigated here, the dose distributions were normalized to that for a 2 Gy per fraction regimen using an (a/b) ratio of 3.0 for normal lung tissue, as was done in the study by Sonke et al.

Figure 1.

Box and Whisker plot showing V5 values for plans using the original clinical ITV as well as the plans using the reduced ITVs.

Figure 2.

Box and Whisker plot showing V20 values for plans using the original clinical ITV as well as the plans using the reduced ITVs.

Figure 3.

Box and Whisker plot showing mean lung dose values for plans using the original clinical ITV as well as the plans using the reduced ITVs.

3. Results

The dosimetric parameters for each of the plans evaluated here are summarized as box and whisker plots in Figures 1-4. For all seventeen cases evaluated here, when the ITV was reduced by 3 times the average reduction seen by Han et al¹². (i.e. 10 cc, or 25% mean reduction in PTV volume), the average reduction relative to the originally treated plan (i.e. with no abdominal compression) were 0.6 Gy in MLD, 2.0% in V5, 1.1% in V20 and 1.1% for the NTCP for dynamic conformal arc plans, and 0.6 Gy in MLD, 2.1% in V5, 1.1% in V20 and 1.1% in NTCP for VMAT plans. The corresponding values when comparing the plans with the ITV reduced by 7 times the average seen by Han et al. (i.e. 16.7 cc, or 45% mean reduction in PTV volume) to the original plan were 1.1 Gy in MLD, 3.8% in V5, 2.0% in V20 and 2.0% in NTCP for the dynamic conformal arc plans and 1.2 Gy in MLD, 3.9% in V5, 1.9% in V20 and 1.9% in NTCP for the VMAT plans. For comparison, the maximum observed decrease in MLD, V5, V20 and NTCP were 1.1 Gy, 4.4%, 2.4% and 4.1%, respectively, for the conformal arc plans when the ITV was decreased by three times the average seen by Han et al. For the same reduction in ITV, the corresponding maximum decrease for the VMAT plans were 1.3 Gy in MLD, 3.9% in V5, 2,5% in V20 and 3.5% in NTCP. When the ITV was decreased by 7 times the average reduction seen by Han et al., the maximum decrease in MLD, V5, V20 and NTCP were 2.2 Gy, 7.8%, 4.8% and 7.2% respectively for the dynamic conformal arc plans and 2.1 Gy, 7.2%, 5.3% and 6.5% respectively for the VMAT plans.

4. Discussion

In order to evaluate the dosimetric effects resulting from a reduction in ITV size due to abdominal compression, it may be helpful to evaluate the change in plan parameters that are observed from the highest reasonable level of ITV reduction that might be encountered clinically. Negoro et al.¹⁶ reported this to be 1 cm S/I, and this is the value we used for this study. By using the maximum amount of ITV reduction reported by Negoro et al. we have endeavored to model scenarios that would yield the most optimistic improvements in lung dosimetry and NTCP that should be expected with abdominal compression. Obviously, these improvements are larger than those expected to be observed for the average patient. Table 3 summarizes the decrease in the relevant planning parameters investigated here, when the plan with the largest ITV reduction is compared to the plan with the original ITV. As might be reasonably expected, all of the metrics investigated show general, albeit modest, improvement when the ITV is decreased, because the amount of normal lung being irradiated is reduced. Given that recent literature on this topic has reported a correlation between V5 and radiation pneumonitis in conventionally fractionated lung treatments^{10, 23}, it seems reasonable to evaluate V5 here, especially since arc-based approaches are known to expose large volumes of normal lung to a low dose wash. Interestingly, even for the case of maximum ITV reduction studied here, the maximum reduction in V5 was 7.8% for the conformal arc approach and 7.2% for VMAT plans, in a patient with a simulated pneumonectomy. In light of the report by Ng et al.²⁴, showing that V5 values below 40% are not expected to lead to measurable SBRT-associated toxicities, we view this as a very modest impact, for such an extreme case. We also note that the largest value of V5 observed in any of the seventeen non-compressed patient treatment plans studied here, was 33.6%, which is lower than the 40% maximum tolerated value reported by Ng et al.²⁴, thus suggesting that employing abdominal compression for further V5 reduction may not be clinically necessary.

Table 3.

Summary of decrease (dosimetric gain) in parameters investigated when the plan with the largest ITV reduction is compared to the corresponding plan with the original ITV. *St Dev is the standard deviation of the data. The p values were calculated using a 2-tail paired t-test.

		3X Reduction in ITV				7X Reduction in ITV
Plan Type		V5 (%)	V20 (%)	MLD (Gy)	NTCP (%)	V5 (%)	V20 (%)	MLD (Gy)	NTCP (%)
Dynamic Conformal Arc	Mean	-2.0	-1.1	-0.6	-1.1	-3.8	-2.0	-1.1	-2.0
	St Dev*	0.9	0.7	0.3	1.0	1.7	1.3	0.5	1.8
	Median	-1.8	-0.9	-0.5	-0.8	-3.4	-1.7	-0.9	-1.5
	Maximum	-4.4	-2.4	-1.1	-4.1	-7.8	-4.8	-2.2	-7.2
	Minimum	-0.8	-0.3	-0.3	0.0	-1.4	-0.6	-0.5	-0.1
	p value	7E-08	9E-06	6E-07	4E-04	7E-08	1E-05	4E-07	4E-04
VMAT	Mean	-2.1	-1.1	-0.6	-1.1	-3.9	-1.9	-1.2	-1.9
	St Dev*	0.9	0.7	0.3	0.9	1.6	1.3	0.5	1.7
	Median	-1.8	-0.9	-0.5	-0.8	-3.6	-1.7	-1.0	-1.4
	Maximum	-3.9	-2.5	-1.3	-3.5	-7.2	-5.3	-2.1	-6.5
	Minimum	-0.5	-0.2	-0.2	0.0	-1.5	-0.5	-0.5	-0.1
	p value	8E-08	1E-05	3E-07	2E-04	2E-08	2E-05	1E-07	2E-04

We note that a recent study by Bouilhol et al.¹⁸ has endeavored to answer some aspects of the question posed here; namely, what is the potential dosimetric value of abdominal compression? They acquired 4DCT’s with and without abdominal compression to characterize the decrease in motion envelope and performed a dosimetric analysis on the plans that resulted from abdominal compression and free breathing conditions. Unfortunately, due to the presence of motion-induced artifacts in several of their 4DCT datasets, they were only able to perform a full dosimetric comparison on 4 patients. Additionally, the volume of normal lung exposed to low doses (e.g. 5 Gy) was not presented and, given the recent observation that V5 may be correlated with probability of radiation-induced pneumonitis for certain fractionation regimens, we believe this to be a valuable metric to characterize. Lastly, the radiobiological implications of target volume reduction were not presented in their very interesting study.

To acquire better insight into the clinical impact of changes in dose distribution resulting from the reduction in ITV volumes, we performed NTCP calculations for the incidence of Grade 2+ pneumonitis. It is not surprising that the median drop in risk of grade 2+ pneumonitis was small (i.e. 1.5% for dynamic conformal arc plans and 1.4% for VMAT plans using even the largest amount of ITV shrinkage), given the modest decreases in MLD, V5 and V20 for this cohort of patients. It is questionable, in our opinion, as to whether such a small reduction in NTCP would warrant the routine use of abdominal compression, simply for the purpose of lowering the amount of normal lung tissue irradiated. We note that for the largest extent of ITV reduction studied here (59.7 cc, 56 %), the largest decrease in NTCP for grade 2+ pneumonitis was a 3.6% reduction for both conformal arc and VMAT plans, both occurring in cases where the original risk was 12%.

We also note that the modest drop in MLD and V20 observed here is consistent with MLD and V20 values reported by Bouilhol et al.¹⁸ for the 4 patients they studied. Our current work extends Bouilhol’s valuable work by modeling larger extents of ITV reduction, for significantly more patients, for important low dose wash implications (as characterized by V5), and to include an important radiobiological metric of NTCP. For the 17 cases studied here, our results demonstrate that the use of abdominal compression to reduce the amount of normal lung irradiated should not be expected to yield large amounts of dosimetrically-driven clinical benefit. This conclusion seems important, when weighed against a potential for decrease in patient compliance second to possible lack of tolerance for abdominal compression.

We note, however, that it seems only reasonable to anticipate that there will be specific situations where abdominal compression may be dosimetrically valuable. For example, where an alteration of the shape or extent of the ITV allows for direction of dose away from a particular sensitive structure, or where application of abdominal compression might yield a more consistent breathing pattern. These specific cases certainly warrant further investigation, but are beyond the scope of the present study.

5. Conclusion

The results of this study have shown that even for very optimistic amounts of abdominal compression-driven ITV reduction, only modest improvement in key normal lung dose metrics was achieved. This is further exemplified by the very small change in NTCP for grade 2+ pneumonitis calculated from the dosimetric data. Based on our analysis, the use of abdominal compression for the sole purpose of improving lung dosimetry in patients receiving lung SBRT by dynamic conformal arc or VMAT techniques should only be expected to yield modest clinical benefit, even in the extreme case of a patient with a single lung of very low volume.