Evaluation of plan quality and treatment efficiency for single‐isocenter/two‐lesion lung stereotactic body radiation therapy

Abstract Purpose/objectives To evaluate the plan quality and treatment delivery efficiency of single‐isocenter/two‐lesions volumetric modulated arc therapy (VMAT) lung stereotactic body radiation therapy (SBRT). Materials/methods Eight consecutive patients with two peripherally located early stage nonsmall‐cell‐lung cancer (NSCLC) lung lesions underwent single‐isocenter highly conformal noncoplanar VMAT SBRT treatment in our institution. A single‐isocenter was placed between the two lesions. Doses were 54 or 50 Gy in 3 and 5 fractions respectively. Patients were treated every other day. Plans were calculated in Eclipse with AcurosXB algorithm and normalized to at least 95% of the planning target volume (PTV) receiving 100% of the prescribed dose. For comparison, two‐isocenter plans (isocenter placed centrally in each target) were retrospectively created. Conformity indices (CIs), heterogeneity index (HI), gradient index (GI), gradient distance (GD), and D2cm were calculated. The normal lung V5, V10, V20, mean lung dose (MLD) and other organs at risk (OARs) doses were evaluated. Total number of monitor units (MUs), beam‐on time, and patient‐specific quality assurance (QA) results were recorded. Results The mean isocenter to tumor distance was 6.7 ± 2.3 cm. The mean combined PTV was 44.0 ± 23.4 cc. There was no clinically significant difference in CI, HI, GD, GI, D2cm, and V20 including most of the OARs between single‐isocenter and two‐isocenter lung SBRT plans, evaluated per RTOG guidelines. However, for single‐isocenter plans as the distance between the lesions increased, the V5, V10, and MLD increased, marginally. The total number of MUs and beam‐on time was reduced by a factor of 1.5 for a single‐isocenter plan compared to a two‐isocenter plan. The single‐isocenter/two‐lesions VMAT lung SBRT QA plans demonstrated an accurate dose delivery of 98.1 ± 3.2% for clinical gamma passing rate of 3%/3 mm. Conclusion The SBRT treatment of two peripherally located lung lesions with a centrally placed single‐isocenter was dosimetrically equivalent to two‐isocenter plans. Faster treatment delivery for single‐isocenter treatment can improve patient compliance and reduce the amount of intrafraction motion errors for well‐suited patients.

compliance and reduce the amount of intrafraction motion errors for well-suited patients. has a high cure rate comparable to surgery. [1][2][3][4][5][6][7] In these studies, medically inoperable patients with early-stage NSCLC who underwent SBRT had 3-yr primary tumor local control rates of up to 98% and a low risk of treatment-related toxicity.
In the setting of either multiple primary lung cancers or limited metastatic lesions to the lungs (oligometastastic), SBRT presents a relatively new treatment opportunity. Optimal treatment planning must consider microscopic disease extension around the visible mass and allow for tumor movement, primarily due to respiration. Multiple metachronous or synchronous lung cancers are relatively common and have been managed by SBRT. 8 Based on Phase I/II trials of SBRT in the management of oligometastastic lung lesions, for patients with one to three tumors, up to five tumors (with curative intent) and more than five tumors with palliative treatment have been reported. 9,10 Rusthoven and colleagues treated 38 patients, 63 total tumors, with lung SBRT of total dose of 48-60 Gy in 3 fractions. Actuarial local control rates at 1-and 2-yr after SBRT was 100% and 96% respectively. 10 SBRT to multiple lung lesions presents with technical challenges and can be treated either sequentially with separate treatment plans or synchronously to all lesions. However, the location and geometry of synchronous plans can be challenging since minor inaccuracies of patient setup can result in geometric misses. Attention must be paid to overlapping doses to organs at risk (OARs) and respiratory control is critical since different parts of the lung can move independently. Sequential treatment plans for each individual tumor, using a multi-isocentric technique requires relatively longer planning and treatment delivery time. Safe and effective delivery of SBRT of lung requires precise, highly conformal treatment planning and delivery techniques. [11][12][13] In the past decades, treatment techniques for lung SBRT included Linear accelerator-based 3D-conformal radiation therapy, intensity modulated radiation therapy (IMRT), volumetric modulated arc therapy (VMAT) (RapidArc,Varian Inc., Palo Alto, CA, USA), CyberKnife, and helical Tomotherapy (Accuray Inc., Sunnyvale, CA, USA). However, as the complexity of the technology has evolved, treatment has required very high total monitor units (MU) and relatively long treatment times to deliver a highly conformal plan and spare OARs. [14][15][16] With the recent technological advances, VMAT may provide highly conformal radiation dose delivery with faster delivery times. [17][18][19][20] The VMAT lung SBRT simultaneously optimizes gantry speed, multileaf collimator (MLC) position and high dose-rate (FFF, flattening filter free mode) to provide highly conformal dose distributions to the planning target volume (PTV) while minimizing dose to adjacent OARs. Reducing treatment time would improve patient compliance which helps reduce error due to motion, and promote more efficient clinic flow. For multiple brain metastases, recent studies have shown that single-isocenter VMAT can provide highly conformal radiosurgical dose distributions, excellent plan quality and safe and faster treatment delivery compared to conventional multiisocenter technique. [21][22][23] However, there is little literature in the medical physics community on the treatment of multiple lung lesions using single-isocenter VMAT-SBRT technique.
A few studies have examined the use of single-isocenter SBRT for multiple lung lesions. A study by Trager et al. 24 discusses the use of a technique that utilizes a single-isocenter with distinct optimizations for extracranial radiosurgery. Gulam et al. 25 examined six patients and found that the criteria set forth by Radiation Therapy Oncology Group (RTOG) study 0915 protocol was met with regard to CI, but not some other critical dosimetric parameters. A retrospective study in total eleven patients by Quan et al. 26 showed no difference in multiple dosimetric parameters between single-isocenter VMAT plans (four singleisocenter VMAT plans were compared) and multi-isocenter intensitymodulated SBRT to the lung. Still, the ability of a single-isocenter treatment to two or more lung lesions to deliver curative treatment plans in adherence with RTOG 0915 dosimetric compliance criteria has not been fully explored. In this report we present our recently adopted treatment method utilizing single-isocenter VMAT plan for SBRT of two lung lesions evaluated per RTOG 0915. For completeness, the original single-isocenter lung SBRT plans and retrospectively generated conventional two-isocenter lung SBRT plans were compared via their protocol compliance, plan quality, dose to critical structures, treatment delivery efficiency, and accuracy.

2.A | Patient setup and target delineation
A total of eight sequential patients were included in this retrospective study, all of whom had two peripherally located Stage I NSCLC SANFORD ET AL.

| 119
lesions. The patients were immobilized using Body Pro-Lok ™ platform (CIVCO system, Orange City, IA) in the supine position with their arms above their head with abdominal compression, potentially reducing diaphragmatic motion to less than or equal to 1.0 cm. Conventional 3D CT scans and respiration-correlated 4D CT scans were acquired on a GE Lightspeed 16 slice CT scanner (General Electric Medical Systems, Waukesha, WI) with 512 × 512 pixels at 2.5 mm slice thickness in the axial cine mode. Varian's Real Time Position Management Respiratory Gating System (version 1.7) was used for collection of 4D CT data. All 10 phases of 4D CT slices and respiratory motion signal were transferred to an Advantage 4D Workstation (General Electric Medical Systems, San Francisco, CA), where the maximum intensity projection (MIP) images were generated after a phase binning of the 4D CT images. In addition to the MIP images, the motion of both tumors was evaluated by an experienced physicist to affirm synchronous tumor motion that was less than 1 cm.
The regular 3D CT scan and the MIP images were imported into the  Highly conformal, clinically optimal VMAT treatment plans were generated using 3-4 non-coplanar partial arcs (5-10°, couch kicks were used for arcs) for the Truebeam linear accelerator (Varian, Palo Alto, CA) with millennium MLC and a 6 MV-FFF (1400 MU/ min) beam. A single-isocenter was placed approximately between the two lesions. As the isocenter location does not need to be exactly in the middle of the lesions, an offset allowing for the gantry to rotate in a partial arc can be made. For those arcs, collimator angles were chosen in such a way that the opening of the MLC between tumors was minimized while the gantry rotates around the patient. Additionally, jaw tracking was used to further minimize the out of field leakage dose. The isocenter to tumors distance was the maximum 3D-linear distance from the single-isocenter location to the geometric center of the individual tumor/isocenter. This distance was calculated in the TPS using the x-, y-, and z-primary coordinates of the tumor centers. This distance was estimated to evaluate the normal lung doses as a function of isocenter distance from the targets. A dose of 54 or 50 Gy in 3 and 5 fractions was prescribed to the PTV D95%. All clinical treatment plans were calculated using the Eclipse TPS with Acuros-XB (version 13.6.0, Varian Medical Systems, Palo Alto, CA) algorithm on the 3D CT images for heterogeneity corrections with a 2.0 × 2.0 × 2.0 mm 3 dose calculation grid-size. Dose to medium reporting mode was selected. All clinical plans were inversely optimized using variation of gantry rotation speed, dose rate and MLC positions. The generalized normal tissue objective (NTO) parameters were used to control the gradients for single-isocenter clinical plan. As recommended by Varian, in our department, we used the following NTO parameters for lung SBRT plans: NTO with high priority of 150 with distance to target border of 0.1 cm. Start dose of 100.0% and end dose of 40% was used with a fall-off factor of 0.5/mm. Moreover, the ring structures of 5, 10, and 20 annulus from each lesion with 5 mm gaps were generated to enforce the high dose regions (typically enforcing maximum 120% hotspot inside each ITV) and minimize the intermediate dose spillage. All the planning objectives were per RTOG 0915 guidelines. The patients were treated every other day per lung SBRT protocol.

2.B.2 | Two-isocenter VMAT plan
For comparison, the SBRT treatment plans for all patients were retrospectively replanned with a conventional two-isocenter approach.
Individual isocenters were placed in the geometric center of each tumor. For each target, the plans were generated using 3-4 noncoplanar partial arcs, similar to single-isocenter plan. Collimator rotations and jaw tracking were applied. The plan for the first tumor (PTV1) was first computed using same RTOG guidelines as described before. The plan for PTV1 was then used as the base-plan for generating the plan for the second tumor (PTV2) in order to allow full scatter contributions from both plans. All the planning objectives used were the same as the single-isocenter plan including the NTO parameters and ring structures. Dosimetric parameters for the target coverage and the adjacent OARs, including normal lung, were evaluated.

2.C | Plan evaluation
Each plan was evaluated for the target coverage and the dose to OARs. For example, using the percentage prescribed isodose volume and target size, the RTOG conformity index (CI) was calculated as follows: 27 where TV PIV is the target volume covered by the prescription isodose volume, TV is the target volume and PIV is the prescription isodose volume. CN = 1.0 would be ideal. The heterogeneity index (HI) was determined by: where D10% is the dose to the hottest 10% of the PTV and D95% is the dose to the 95% of the PTV coverage. The intermediate dose spillage was evaluated by using, gradient index (GI), D 2cm and gradient distance (GD). The GI was given by: where R50% is the ratio of 50% prescription isodose volume to the PTV and R100% is the ratio of 100% prescription isodose volume to the PTV. Per RTOG, depending on the target size, a GI of 3.0-6.0 is desirable. Similarly, D 2cm is the maximum dose, in percent of dose prescribed, at 2 cm from the PTV in any direction; and the GD, is the average distance from 100% prescription dose to 50% of the prescription dose. Although, RTOG only recommended normal lung, V20 < 10% (10-15% was acceptable with minor deviations), we have evaluated V5, V10, and mean lung dose (MLD) for normal lung for all plans.

2.D | Dose to other OARs
In addition to the lung dose, all the clinical single-isocenter plans were evaluated for dose to spinal cord, heart, esophagus, trachea, ribs, and skin per RTOG guidelines. The dose volume histogram parameters were compared between the single-isocenter and the two-isocenter plans. The mean and standard deviation values for each of the dose metrics were compared using paired t tests for single-isocenter vs two-isocenter computed dosimetric parameters for the OARs dose tolerances using an upper bound of P < 0.05.

2.E | Delivery efficiency and accuracy
The dose delivery efficiency of each lung SBRT plan was evaluated based on total number of MU and actual beam-on time. The absolute differences between single-isocenter and two-isocenter plans for normal lung V20, V10, V5, and MLD were listed in the Table 2. All patients had V20 < 10-15% for both treatment plans. A statistically insignificant difference (P = 0.09) was found for the normal lung V20 between two plans. However, V10, V5, and MLD increases slightly with single-isocenter plan compared to twoisocenter plan, giving statistically significant differences (P = 0.03, 0.01 and 0.03 respectively). Statistically significant P-values are highlighted in bold (see Table 2). Although, V10, V5, and MLD had shown statistically significant differences, the absolute differences were on the order of less than 0.8% for V20, 2.8% for V10 and 6.5% for V5) and less than 60 cGy for MLD, on average, therefore, we do not expect the differences would be clinically significant.
The ratios between single-isocenter and two-isocenter plans for the V20, V10, and V5 as a function of isocenter to tumors distance can be seen in Fig. 3. When the isocenter to tumor distance increased, the low dose volume to the normal lung, such as V5 and V10, was slightly increased. However, two of eight patients had lower values of V20 with single-isocenter plan.

3.B | Dose to other OARs
A comparison of other OARs dosimetric parameters for single-isocenter and two-isocenter plans for all eight lung SBRT patients is presented in Table 3. Critical organs such as spinal cord (D max , and D 0.35cc ), heart (D max and D 15cc ), esophagus (D max and D 5cc ), trachea (D max and D 4cc ), ribs (D max and D 1cc ), and skin (D max and D 10cc ) were evaluated per SBRT protocol guidelines.
The average values of maximum doses to spinal cord, ribs, and skin were similar (also see the average of the ratios in Table 3) between the two planning methods. Although, the average values of the absolute dose differences and ratios for heart, esophagus and trachea were slightly higher with single-isocenter plan, the average absolute dose differences were up to 1-2 Gy. While evaluating those plans per SBRT protocol's guidelines, those values met the protocol criteria, therefore, the differences were not deemed clinically significant. Almost all P-differences were F I G . 1. This shows the dose volume histogram comparison for the target coverage (for both PTV1 and PTV2). The ITVs (red) and a few OAR such as total normal lung (light blue), heart (dark blue), ribs (green), and spinal cord (orange) are shown for patient #8. Prescription dose was 54 Gy in three fractions. The square symbols representing the single-isocenter plan, and the triangle symbols representing the two-isocenter plan. Both plans were normalized to at least 95% of PTV received 100% of the prescribed dose. In this case, the isocenter to tumors distance was about 4 cm; the dosimetrically equivalent plans were generated using single-isocenter technique, as demonstrated, with similar target coverage and dose to the OARs.
F I G . 2. This is a comparison of isodose distributions in sagittal view for the same patient #8 generated via single-isocenter and twoisocenter plans. In the right panel a single-isocenter location is shown by the intersection of the cross-hair; in the left panel two-isocenter plan sum is shown for the both targets (PTV1 and PTV2). Target volumes contoured include both ITVs (red, innermost) followed by PTVs (orange and green, outermost). Higher isodose lines, such as 54 Gy (100%), 51.3 Gy (95%), 48.6 Gy (90%), 43.2 Gy (80%), exhibit sharp dose fall off for the both plans, including 27.0 Gy (50%) isodose line (blue). In both plans, the hotspot, 120% isodose line (thick-orange) was shown in the middle of the ITV. Other OARs such as ribs and lung contours are shown. Purple color rings were contoured to calculate D 2cm (%) for each target.
insignificant, except for dose to 15 cc of heart (P = 0.002) and dose to 10 cc of ribs (P = 0.02). Both the single-isocenter and two-isocenter plans were within clinically acceptable limits per RTOG 0915.

3.C | Delivery efficiency and accuracy
For single-isocenter plans, the mean values of total number of MUs and beam on time were 6014 (4013 to 10,727) and 4.3 min (2.9 to T A B L E 1 Comparison of plan evaluation parameters for single-isocenter vs two-isocenter treatment plans of all eight lung SBRT patients. Lesion 1 (PTV1 plan) and Lesion 2 (PTV2 plan) and two-isocenter (Two-iso) plan sum. However, for patient #2, the gamma pass rates were around 92% for 3%/3 mm criteria. In this case, both tumors were relatively large, and the tumors to isocenter distance was relatively large, around 9.5 cm.
In addition, the tumors were located in the bilateral lungs, therefore, the MLCs have to travel a longer distance, providing suboptimal VMAT QA pass rates; suggesting that exceeding 10 cm (isocenter to tumors distance) may not provide clinically optimal plan with singleisocenter. While reanalyzing those data with a tighter distance-toagreement (3%/2 mm) criteria, the average value of gamma pass rate was 95.8 ± 3.8% (ranged, 90.6 to 100%) that was within the departmental SBRT VMAT QA pass rate criteria (>/=90.0% pass rates).
Since, the two-isocenter plans were not used for actual patient's treatment we did not run VMAT QA for those plans.

| DISCUSSION
In this study, we have presented our initial clinical experiences of a fast, effective, and accurate treatment planning and delivery F I G . 3. Scatter plot: For all eight lung SBRT patients, the ratios of V5, V10 and V20 of normal lung doses calculated by single-isocenter and two-isocenter plans as a function of isocenter to tumors distance. For the identical planning objectives, the single-isocenter plan gave slightly higher values of V5, V10, and V20 by a factors of 1.2, 1.1, and 1.1, on average, respectively, compared to two-isocenter plan. This suggests that comparable dosimetric parameters can be obtained for the normal lung. However, single-isocenter plan would have considerably faster treatment delivery by an almost a factor of 2, eliminating the setup and verification time for the second isocenter plan.
T A B L E 3 Average values of absolute dose differences between single-isocenter and two-isocenter plans for the other major dose distribution parameters of the OARs for all eight lung SBRT patients. Absolute dose differences = single-isocenter-two-isocenter. The negative sign indicates that the results of the two-isocenter plans were larger than those of single-isocenter plans. Statistically significant P-values are highlighted in bold. a Single-isocenter/two-isocenter.

OARs
technique using single-isocenter VMAT plans for SBRT of two lung lesions following RTOG 0915 protocol guidelines. 12 Our single-isocenter VMAT plan for SBRT of two lung lesions uses 3-4 noncoplanar partial arcs with jaw tracking and patient specific collimator angles to minimize leakage dose from leaves travelling in between the tumors. Single-isocenter VMAT-SBRT plans were highly conformal and achieved adequate target coverage (see Table 1 for CI, HI, Paddick CN, GI, D 2cm , and GD) compared to conventional two-isocenter plans. For all patients, the single-isocenter plans met RTOG guidelines including normal lung V20 and were similar compared to two-isocenter plans. However, when the isocenter to tumors distance increased, the low dose volume to the normal lung, such as V5 and V10, was slightly increased as shown in Fig. 3. In addition, the other OARs such as spinal cord, heart, esophagus, trachea, ribs, and skin dose tolerances were also within protocol. The single-isocenter treatment was well-tolerated with all patients. The beam on time was 4.3 min and VMAT-SBRT QA gamma passing rates were 98.1% (3%/ 3 mm clinical gamma passing criteria), on average, demonstrating an excellent potential for a fast, reliable, and accurate delivery of singleisocenter VMAT lung SBRT treatment for two lung lesions.
The single-isocenter plan for treating multiple lung tumors has been reported by a few investigators. 29 One potential concern for single-isocenter VMAT-SBRT plan for two lung lesions was low dose spill in the normal lung, such as V20, V10, and V5. Per RTOG recommendation, all our single-isocenter/ two-lesions VMAT lung SBRT plans had V20 < 10-15%. Moreover, normal lung V5 was maintained less than 40%, on average. [31][32][33] Although, in our experience when the isocenter to tumors distance increased, the normal lung V10 and V5 slightly increased, as expected, when compared to two-isocenter plan. Our treatment planning strategy favored minimizing normal lung dose during singleisocenter VMAT planning (by optimizing patient specific collimator angles in conjunction with jaws tracking such that the leakage dose due to the leaves travelling in between two tumors could be minimized) that could potentially help reduce severe lung toxicity with careful attention to V5 and V10 during plan optimization.
The detailed information on total number of MUs and beam-on time for the both single-isocenter and two-isocenter plans for all eight lung SBRT patients. The Octavius VMAT-SBRT QA pass rates and point dose measurements for single-isocenter plans were also shown. Another potential concern for single-isocenter VMAT plan was the patient setup errors, for example tumor motion and rotational errors. In addition, due to rotational errors, for small targets and those away from the single-isocenter could potentially alter the dose distributions. For those highly conformal VMAT plans, the small deviation in motion error could potentially irradiate normal tissues, and it may increase the chance of radiation-induced toxicity or miss the target.
Our attending physician has addressed that issue by individually reviewing these target volumes and the associated tumor motion pattern and by assigning appropriate ITV to PTV margins (usually 5 mm in the medio-lateral and anterior-posterior directions and 8 to 10 mm in superior-inferior direction) to accommodate potential tumor deformation. Moreover, great care has been taken by our treating physician and the physicist to address some of the abovementioned issues, for example, being available for the patient setup (in the 3D, 4D CT simulation and each treatment), image guidance, and CBCT matching and physically authorizing each treatment fraction for all patients. However, it is worthwhile to mention here that the 8-10 mm superior-inferior expansion of the ITV to PTV is not a requirement for an effective treatment of two lung lesions using a single-isocenter plan, but this was really only a conservative preference of our treating physicians from their many years of lung SBRT experiences. Further studies are required to validate the standard 5 mm ITV to PTV expansions that would be adequate or not for this kind of treatment setting while fulfilling the RTOG compliance.
In summary, each plan was rigorously evaluated using the dosimetric parameters listed in the Tables 1, 2, and 3. All parameters were deemed acceptable for both single-isocenter and two-isocenter plans per SBRT protocolsuggesting that single-isocenter plan could be dosimetrically equivalent to two-isocenter plan and a faster and equally effective treatment delivery which can be offered to well suited patients. In the future, these patients will be followed up clinically and evaluated for local control rates and treatment related toxicity such as the effect of normal lung dose as a function of isocenter to tumors distance. Moreover, single-isocenter VMAT plan for SBRT of lung for more than two lesions will be investigated.

| CONCLUSION
This report presents our initial clinical experience with a single-isocenter for two-lesion SBRT procedure for lung tumors and compared with conventional two-isocenter plan. Treatment of peripherally located two lung lesions with centrally assigned single-isocenter was dosimetrically equivalent to two-isocenter plan. For single-isocenter plans, it was observed that as the distance between the lesions increased the normal lung V5, V10 and MLD somewhat increased.
The single-isocenter technique was fast, accurate, and very well-tolerated by all the patients, improving patient comfort and potentially reducing the amount of intrafraction motion errors for well-suited patients. Clinical follow-up of these patients is warranted to determine the tumor local control rates and treatment related toxicity.

CONFLI CT OF INTEREST
The authors declare no conflict of interest.