Skip to main content
NCBI home page
The British Journal of Radiology logoLink to The British Journal of Radiology
. 2017 Oct 17;90(1079):20170523. doi: 10.1259/bjr.20170523

Clinical outcomes of multileaf collimator-based CyberKnife for spine stereotactic body radiation therapy

Nalee Kim 1, Ho Lee 1, Jin Sung Kim 1, Jong Geal Baek 1, Chang Geol Lee 1, Sei Kyung Chang 2, Woong Sub Koom 1,✉
PMCID: PMC5963365  PMID: 28869401

Abstract

Objective:

Stereotactic body radiotherapy (SBRT) for spinal metastases is becoming a prevalent therapeutic option. We aimed to evaluate the clinical feasibility and outcomes of the recently developed multileaf collimator (MLC)-based CyberKnife (CK-M) for spine SBRT.

Methods:

We reviewed 119 patients of 144 cases with 229 lesions treated with CK between November 2014 and March 2016. The lesion features, dosimetric parameters and clinical outcomes were compared between fixed cone collimator based CK (CK-F) and CK-M.

Results:

Of 144 cases, 78 and 66 were treated with CK-F and CK-M, respectively. CK-M achieved an adequate target volume coverage that was comparable with CK-F (median 92 vs 90%; p = 0.03) even in larger targets (median 64.2 vs 46.7 cm3; p = 0.01), respectively. CK-M showed an improvement in the gradient index (p < 0.001) and no difference in conformity (p = 0.16). With CK-M, the median beam delivery time was significantly reduced by 30% (to 34 vs 48 min; p < 0.001). CK-M showed 1 year local control rates that were comparable to CK-F (77 vs 78%, respectively; p = 0.83).

Conclusion:

CK-M exhibits dosimetric data and local control that are comparable with CK-F, but with significant treatment time reduction. CK-M could be widely used in spine SBRT.

Advances in knowledge:

Given the recently developed MLC in CK, we aimed to evaluate the clinical feasibility and outcomes of MLC compared with fixed cone-based CK. MLC showed equivalent plan quality and significant treatment time reduction with comparable radiological control. We report here MLC as an effective and tolerable treatment option in CK.

Introduction

As the survival rates among patients with metastatic diseases that arise from solid malignancies increase, spine metastases are becoming a common complication afflicting over 20,000 patients annually.1 Conventional fractionated external beam radiotherapy (EBRT) of 8 Gy for a single fraction, 20 Gy for 5 fractions and 30 Gy for 10 fractions are frequently used for spine metastases.2 Although conventional EBRT produces a good pain relief response with complete pain relief attained in 0–20% of patients),3 it shows poor local control (LC) in complex and bulky metastatic lesions.4

Spine stereotactic body radiotherapy (SBRT) can deliver a locally ablative radiation dose while maximizing pain relief and radiological LC in a convenient and effective schedule. Since the initial guidelines of the American Society of Therapeutic Radiation Oncology were published,5 the patient selection criteria continues to evolve. The criteria have even been broadened to include historically contraindicated patients; such as those with three or more contiguous segments, widespread metastatic disease, pre-existing vertebral fracture and lesions with epidural disease and potential spinal instability.6

CyberKnife (CK) has a distinct feature of real-time tracking of target motion with intrafractional X-ray imaging and automatic positional adjustments. In 2015, certain CK series were equipped with the InCiseTM (Accuray Inc., Sunnyvale, CA) multileaf collimator (MLC). In addition to fixed-cone and IRISTM (IRIS) collimators, the current version of MLC consists of 41 leaf pairs with widths of 2.5 mm, forming a maximum field size of 120 mm by 102.5 mm.7 The application of MLC was postulated to higher delivery efficacy in terms of treatment time and equivalent planning in the treatment of larger and more complex tumours.8 Until now, there are minimal data evaluating only dosimetric comparisons between MLC-based treatment plans9–11 and cone/IRIS-based plans; such studies do not describe clinical outcomes. The aim of this study was to evaluate both dosimetric feasibility and clinical outcomes of CK equipped with MLC (CK-M) in SBRT for spine metastases.

methods and Materials

Study population

We identified 144 consecutive patients of 179 cases with 315 lesions treated with CK SBRT between November 2014 and March 2016. Of these, the lesions of 144 cases were enrolled after excluding those that did not complete entire sessions (n = 12), had no follow-up radiological evaluation (n = 19) or were treated with IRIS-based CK (n = 4). We retrospectively reviewed 78 cases of fixed cone collimator-based CK (CK-F) and 66 cases of CK-M. This study was approved by the Health Institutional Review Boards of Yonsei University Hospital (IRB No. 4-2016-0994). The requirement for informed consent was waived owing to the study’s retrospective nature.

Patient evaluation

Patient, tumour and dosimetric characteristics, as well as clinical outcomes, were obtained from medical records. We used the epidural spinal cord compression scales (Bilsky grade)12 for evaluating the degree of epidural disease. We grouped the cases into two categories according to Bilsky grade: epidural space compressive disease-negative (Bilsky Grades 0–1c) and positive (Bilsky Grades 2–3). Each of the 229 spinal lesions was scored according to the spine instability neoplastic score (SINS) criteria.13 The SINS system classifies patients as stable, potentially unstable (indeterminate) and unstable based on the overall score.

Treatment

All patients underwent a fine-cut (1 mm) non-contrast-enhanced treatment planning CT scan in the supine treatment position. All patients were immobilized in a customized total body vacuum bag and additional thermoplastic mask for cervical spine treatment. Planning CT data and structures were transferred to the MultiPlan 5.1.2 (Accuray Inc., Sunnyvale, CA) CK planning system. Each treatment plan was generated with fixed cones mounted on the CK M6 (Accuray Inc., Sunnyvale, CA) or the InCise MLC which was installed at Yonsei Cancer Center in August 2015. All patients were treated using the Xsight spine tracking system (Accuray, Sunnyvale, CA), which uses the bony anatomy of the spine for continuous imaging, repositioning and tracking tumours without fiducial markers.

Planning CT and axial gadolinium T1 and T2 weighted MRI were fused to aid with target and spinal cord delineation. Delineation of the gross tumour volume (GTV) and clinical target volume was consistent with the recent International Spine Radiosurgery Consortium consensus guidelines.14 The gross tumour volume was defined as a gross tumour mass or osteolytic lesion identified on the planning CT, high signal intensity lesion on T2 weighted MRI or enhanced lesion on gadolinium-enhanced T1 weighted MRI. The planning treatment volume (PTV) was defined with no margins added to the clinical target volume. We divided the PTV into five groups according to degrees:<180°, 180°, 270°, horseshoe and doughnut-shaped around the spinal cord. An additional 10 mm of the spinal cord (at minimum) was contoured above and below the PTV based on non-coplanar beam arrangements of CK. There was heterogeneity in the patterns of the prescribed dose (Table 1); all patients were radiation naïve at the treatment site. In the entire cohort, most patients received a total of 18 Gy in a single fraction [equivalent dose in 2 Gy per day fractions (EQD2) 42 Gy, 47 cases (32.6%)] or 32 Gy in four fractions [EQD2 48 Gy, 41 cases (28.5%)]. There were different patterns of prescription according to the collimator type; the most frequent prescription was 18 Gy in a single fraction (40 cases, 51.3% of CK-F) in CK-F and 32 Gy in four fractions (38 cases, 57.6% of CK-M) in CK-M, respectively. The dose was typically prescribed to the 80–90% isodose line for PTV.

Table 1.

Patient, disease and treatment characteristics

  Total
 n = 144
 Fixed
 n = 78
 MLC
 n = 66
 p-value          
n % n % n %          
Sex Male 90 62.5 56 71.8 34 51.5 0.01        
Female 54 37.5 22 28.2 32 48.5          
ECOG PS 0–1 59 41.0 34 43.6 25 37.9 0.49        
≥2 85 59.0 44 56.4 41 62.1          
Tumour histology Gastrointestinal 53 36.8 27 34.6 26 39.4 0.40        
Lung 30 20.8 21 26.9 9 13.6          
Prostate 18 12.5 8 10.3 10 15.2          
Renal 14 9.7 8 10.3 6 9.1          
Melanoma 6 4.2 1 1.3 5 7.6          
Sarcoma 5 3.5 2 2.6 3 4.5          
Breast 4 2.8 2 2.6 2 3.0          
Unknown primary 3 2.1 2 2.6 1 1.5          
Other 11 7.6 7 9.0 4 6.1          
Reason to treat Pain 102 70.8 56 71.8 46 69.7 0.58        
Pain with radiating pain 26 18.1 12 15.4 14 21.2          
Neurological symptom 0 0.0 0 0.0 0 0.0          
No symptom 16 11.1 10 12.8 6 9.1          
Status of metastasis Bone 39 27.1 22 28.2 17 25.8 0.74        
Bone + visceral 105 72.9 56 71.8 49 74.2          
Status of bone metastasis Single 32 22.2 13 16.7 19 28.8 0.08        
Multiple 112 77.8 65 83.3 47 71.2          
Bilsky grade 0–1c 104 72.2 53 67.9 51 77.3 0.21        
2–3 40 27.8 25 32.1 15 22.7          
Paraspinal space extension No 54 37.5 25 32.1 29 43.9 0.14        
Yes 90 62.5 53 67.9 37 56.1          
SINS criteria (229 lesions) Stable (0–6) 96 41.9 53 46.1 43 37.7 0.31        
Indeterminate (7–12) 121 52.8 55 47.8 66 57.9          
Unstable (13–18) 12 5.2 7 6.1 5 4.4            
Compression fracture Pre-existing fracture 47 32.6 27 34.6 20 30.3 0.58        
No fracture 97 67.4 51 65.4 46 69.7          
Treated level Median (Range) 1 (1–5) 1 (1–5) 1 (1–4) 0.10              
Single 89 61.8 53 67.9 36 54.5 0.10        
Multiple 55 38.2 25 32.1 30 45.5            
PTV type <180° 40 27.8 24 30.8 16 24.2 0.15        
180° 16 11.1 8 10.3 8 12.1          
270° 55 38.2 32 41.0 23 34.8          
Horseshoe 22 15.3 12 15.4 10 15.2          
Doughnut 11 7.6 2 2.6 9 13.6          
Total dose/no. of Fxs 12–24 Gy/1 Fx 70 48.6 60 76.9 10 15.2          
10–23 Gy/2 Fxs 5 3.5 2 2.6 3 4.5          
24–36 Gy/3 Fxs 7 4.9 4 5.1 3 4.5          
20–60 Gy/4 Fxs 54 37.5 12 15.4 42 63.6          
24–48 Gy/8 Fxs 8 5.6 0 0.0 8 12.1          
Treatment session Median (range) 2 (1–8) 1 (1–4) 4 (1–8) <0.001              
Single 70 48.6 60 76.9 10 15.2 <0.001        
Multiple 74 51.4 18 23.1 56 84.8            
Open in a new tab

ECOG PS, Eastern Cooperative Oncology Group performance status; Fxs, fractions; PTV, planning treatment volume; SINS, spine instability neoplastic score.

Dosimetric parameters

Detailed dosimetric data were retrospectively collected and centrally analysed. PTV-related dosimetric parameters were evaluated for all patients, including PTV volume, percent of prescription coverage and dose to 95% of the volume. For the spinal cord, we collected data for the maximal point dose of 0.035 ml (Dmax)15 and the dose to 0.1 ml of the thecal sac, which was defined with a 1.5 mm margin with the spinal cord. To compare dosimetric parameters among the heterogeneous prescriptions, the EQD2 to the spinal cord and PTV were calculated. An α/β value of 10 Gy was used for the tumour effect and 2 Gy was used for spinal cord late effects.

The dosimetry of each plan was categorized into PTV coverage, conformity index (CI),16 new conformity index (nCI),17 homogeneity index (HI)18 and gradient index (GI).19 The detailed definition of CI is a ratio of total volume of tissue treated with prescription dose compared with the tumour volume covered by the prescription isodose reported in the CK planning system (MultiPlan 5.1.2, v. 5.1.2; Accuray Inc., Sunnyvale, CA). In addition, nCI, also reported in CK planning system, is defined as CI multiplied by ratio of total target volume to target volume receiving prescription dose or more (CI/PTV coverage).

Additionally, we compared results based on total treatment time, beam delivery time and total delivered motor units (MUs) for evaluating treatment efficacy. The beam delivery time included a beam-on time, imaging time with 1 min intervals and robot traveling time in the CK planning system The total treatment time estimated from actual patient setup to unlock patient immobilization after treatment including patient setup time, the beam delivery time and patients’ exit time.

Follow-up and relapse

All patients underwent pre-treatment evaluation and physical examination that included radiological assessment of MRI- or CT-based images of the spine. A radiation oncologist interviewed patients 1 month after treatment and every 3 months, thereafter, to assess pain and neurological symptoms. Follow-up MRI or CT was performed every 1–3 months after the end of the treatment during the first year and every 6 months thereafter. The response of spinal metastases was classified as controlled or progressed based on the criteria suggested by the SPIne response assessment in Neuro-Oncology group.20 Local progression was defined as one of following: gross unequivocal increase in tumour volume or linear dimension, any new or progressive tumour within the epidural space and neurological deterioration attributable to pre-existing epidural disease with equivocal increased epidural disease dimensions.

Grading of treatment-related toxicity was performed at the time of follow-up based on the common terminology criteria for adverse events (CTCAE, v.4.03). Radiation-induced myelopathy (RM) was defined as neurological symptoms consistent with radiation damage without radiological evidence of a recurrent or progressive tumour affecting the spinal cord.21 At baseline, pre-existing vertebral compression fractures (VCFs) were assessed by comparison to the vertebral body (VB) height on prior imaging or, if unavailable, the average height of the immediately superior and inferior VBs. We evaluated the development of a new VCF or progression of an existing VCF at the site of treatment.

Statistical analysis

Statistical analysis was performed using IBM SPSS Statistics v. 23 (SPSS Inc., Chicago, IL). Student's t-tests were used for normal variables and Mann–Whitney U test was used for ordinal but non-nominal variables between CK-F and CK-M groups. The Pearson’s Χ2 tests or Fisher exact tests were employed to evaluate the differences of categorical data. All outcomes were measured from the day of SBRT to the time of the event. LC was defined as the time to local progression at the site specifically treated with CK. Overall survival (OS) was calculated as the time from treatment until death from any cause or last follow-up. The Kaplan–Meier method was used to estimate the LC and OS. Potential prognostic factors were compared using the log-rank test and factors with p-value less than 0.2 in univariate analyses were introduced into Cox proportional hazards model to determine the adjusted hazard ratios (HRs) with 95% confidence intervals (CIs. Logistic regression models were used to model increased VCFs as a function of collimator and other covariates. Two-tailed p values of ≤ 0.05 were considered statistically significant.

Results

Cohort characteristics

A total of 119 patients treated with CK SBRT were included in this study: 229 lesions in 144 cases were finally analysed. Patient, disease and lesion characteristics are described in Table 1. The median age of the cohort was 62 years old (range, 18–82). On pre-SBRT neurological assessment, most cases had varying degrees of site-specific pain (n = 102, 71%) without any cases of neurological symptoms (i.e. sensory change or motor weakness). The overall disease burden in the entire cohort was relatively high; visceral organ involvement was reported in 105 cases (73%) and multiple bone metastases were reported in 112 cases (78%). Most lesions were categorized as bilsky Grade 0‒1c (n = 104, 72%), extension to paraspinal space (n = 90, 63%) and either stable (n = 96, 42%) or indeterminate (n = 121, 53%) in SINS criteria. Multiple levels of spine were included in 55 cases (38%) and 270° degree of PTV type was most frequent (n = 55, 38%). There were no significant differences in disease and treatment characteristics between CK-F (n = 78, 54%) and CK-M (n = 66, 46%) except for the number of treatment sessions; median number of treatment sessions was one in CK-F (range, 1‒4 sessions) and four in CK-M (range, 1‒8 sessions), respectively (p < 0.001) (Table 1).

Dosimetric analysis

Dosimetric parameters of both collimator types are summarized in Table 2. For CK-M, plans were prescribed to high isodose levels. Coverage of PTV in CK-M was better than that of CK-F (median 90 vs 92%, respectively; p = 0.03) even with larger PTVs treated with CK-M. Despite higher prescription doses with CK-M, there were no significant differences in the doses to the spinal cord or thecal sac. The dose conformities (CI and nCI) between CK-F and CK-M suggested that they are comparable. However, CK-M showed a better homogeneous dose distribution (HI, p < 0.001) and dose gradient fall off (GI, p < 0.001).

Table 2.

A comparison of dosimetric parameters between the fixed cone (Fixed) and multileaf collimators (MLC)

  Total (n = 144)
 Fixed (n = 78)
 MLC (n = 66)
 p-value        
Median (IQR) Median(IQR) Median (IQR)        
PTV volume (cm3) 52.9 (27.6;84.8) 46.7 (23.6;69.0) 64.2 (33.1;112.3) 0.01        
Prescription total dose (Gy)a 42.0 (42.0;48.0) 42.0 (34.7;42.0) 48.0 (48.0;48.0) <0.001        
BED (Gy) 50.4 (50.4;57.6) 50.4 (41.6;50.4) 57.6 (57.6;57.6) <0.001        
PTV coverage (%) 90.4 (85.5;94.3) 90.0 (84.9;92.6) 91.8 (87.8;95.2) 0.03        
Prescription isodose (%) 78.0 (75.0;80.0) 75.1 (73.0;78.0) 80.0 (77.0;82.0) <0.001        
Cord_0.035cc (Gy)a 52.5 (38.0;62.9) 52.7 (33.3;65.4) 51.8 (44.4;62.7) 0.55        
Thecal sac_0.1cc (Gy)a 61.2 (44.9;72.7) 60.5 (37.1;75.0) 61.4 (52.7;72.5) 0.70        
CI 1.04 (0.95;1.12) 1.02 (0.93;1.10) 1.05 (0.97;1.13) 0.16        
nCI 1.27 (1.22;1.36) 1.28 (1.23;1.37) 1.27 (1.19;1.36) 0.19        
HI 1.28 (1.25;1.33) 1.33 (1.28;1.37) 1.25 (1.22;1.30) <0.001        
GI 3.47 (3.06;4.04) 3.87 (3.54;4.70) 3.04 (2.86;3.36) <0.001        
Delivery time (min) 40.0 (31.0;50.0) 48.0 (38.0;53.0) 33.5 (27.0;39.0) <0.001        
Total treatment time (min) 55.0 (45.0;69.0) 65.0 (53.0;75.0) 47.0 (42.0;55.0) <0.001        
Delivered MU 14029.9 (7955.4;20864.5) 20193.7 (14141.6;23477.6) 8162.1 (6425.1;11375.2) <0.001        
Beam number 103.5 (58.0;157.5) 152.0 (123.0;180.0) 53.5 (41.0;68.0) <0.001        
Open in a new tab

BED, biologically effective dose; CI, conformity index; GI, gradient index; HI, homogeneity index; IQR, interquartile range; MU, motor unit; nCI, new conformity index; PTV, planning treatment volume.

a

The equivalent dose in 2 Gy per day fractions was calculated for all dosimetric data using an α/β value of 10 Gy for tumour effect on PTV and a value of 2 Gy for the spinal cord late effects.

The delivered MUs for the CK-M were reduced considerably (by 60%) compared with CK-F. This decrease in MUs resulted from a reduction in the total number of beams (median 54 vs 152 for CK-M and CK-F, respectively; p < 0.001) since each beam consists of multiple segments depending on the PTV shape. Total number of beams in CK-M was 35% of that in CK-F and delivery time in CK-M showed 30% decrease compared with CK-F (median delivery time 33.5 vs 48.0 min, p < 0.001).

In subgroup analysis based on PTV type (either ≤ 270° or horseshoe and doughnut-shaped) and PTV volume (cut-off value of median volume 52.9 cm3), CK-M consistently showed better dose homogeneity (HI), dose gradient fall off (GI), shorter beam delivery time, total treatment time, reduced delivered MU and number of beams [all p < 0.05, Supplementary Tables 1 and 2 (Supplementary material available online)].

Local control (LC)

The median follow-up was 5.7 months (range, 1.1–18.8 months). Radiological progression occurred in 18 cases (12.5%). The actuarial 6 month and 1 year LC rates for the entire cohort were 88 and 78%, respectively (Figure 1). Univariate analysis results of LC are reported in Table 3. For a categorical analysis of dosimetric parameters, the median values were used as the cut-off point. There was no difference in LC according to collimator type (p = 0.83, Figure 2), PTV volume (p = 0.50), PTV coverage (p = 0.78) and other dosimetric parameters (all p > 0.05) The only significant variable associated with LC was the prescribed dose with a cut-off value of 42 Gy (p = 0.05, HR 0.40, 95% confidence interval CIs 0.16–1.02). In the Cox proportional hazards model, two independent predictors of LC were identified: multiple levels of spine (p = 0.04, HR 3.59) and prescribed dose (p = 0.01, HR 0.24, Table 3). OS at 6 months and 1 year was 66.2 and 49.8% after CK-F and 59.7 and 54.7% after CK-M, with no significant difference between collimator groups.

Figure 1.

Figure 1.

Open in a new tab

LC following treatment for the entire cohort.

Table 3.

Univariate Cox regression and competing risk analyses of factors associated with local control (LC)

Variables Univariate analysis
 Multivariate analysis
          
HR 95% CIs p-value HR 95% CIs p-value  
Paraspinal extension No Ref.   0.63      
Yes 0.80 0.31–2.02        
Bilsky grade 0–1c Ref.   0.16 Ref.   0.37
2–3 1.95 0.77–4.95   1.62 0.57–4.61  
Spine level Single Ref.   0.16 Ref.   0.04
Multiple 1.95 0.77–4.95   3.59 1.08–11.89  
Collimator Fixed Ref.   0.83      
MLC 0.89 0.30–2.64        
PTV volume <52.9 cm3 Ref.   0.50      
≥52.9 cm3 1.38 0.54–3.51        
Prescription dosea <42 Gy Ref.   0.05 Ref.   0.01
≥42 Gy 0.40 0.16–1.02   0.24 0.08–0.75  
PTV coverage <90.4% Ref.   0.78      
≥90.4% 0.88 0.34–2.22        
PTV_Dmeana <52.7 Gy Ref.   0.79      
≥52.7 Gy 0.88 0.34–2.29        
PTV_D95a <40.2 Gy Ref.   0.16      
≥40.2 Gy 0.47 0.17–1.34        
Cord_Dmaxa <52.5 Gy Ref.   0.75      
≥52.5 Gy 0.86 0.34–2.19        
Thecal sac_Dmaxa <61.2 Gy Ref.   0.41      
≥61.2 Gy 0.67 0.26–1.73        
CI <1.040 Ref.   0.60      
≥1.040 0.78 0.31–1.97        
nCI <1.275 Ref.   0.06 Ref.   0.07
≥1.275 0.39 0.15–1.05   0.39 0.14–1.09  
HI <1.280 Ref.   0.70      
≥1.280 1.23 0.43–3.50        
GI <3.470 Ref.   0.12 Ref.   0.07
≥3.470 0.45 0.17–1.22   0.38 0.13–1.07  
Open in a new tab

CI, conformity index; D95, Dose to 95% of the volume; Dmax, maximum dose; Dmean, Mean dose; HI, homogeneity index; HR, hazard ratio; GI, gradient index; PTV, planning treatment volume; CIs, confidence interval.

a

The equivalent dose in 2 Gy per day fractions was calculated for all dosimetric data using an α/β value of 10 Gy for tumour effect on PTV and a value of 2 Gy for the spinal cord late effects.

Figure 2.

Figure 2.

Open in a new tab

LC after CyberKnife (CK) by collimator options: fixed collimator (Fixed) (n = 78), MLC (n = 66).

Toxicity

The overall rate of pain relief was 82% (118 of 134) at 1 month after treatment. There was no severe acute toxicity; Grade 2 generalized weakness was observed in 15 cases, Grade 2 fatigue in 14 and Grade 2 nausea in 13. There was no clinically detectable RM; however, there were 11 cases of persistent radiating pain; 4 of progressive lesions and 7 of stable lesions without clinical aggravation. 29 of 144 cases had VCFs (20%), with 14 of 97 (14%) de novo and 15 of 47 (32%) having progressed from an existing VCF. The median time to VCF was 4.1 months (range, 0.7–19.2 months) and 60% (18 cases) occurred within 6 months after treatment. In logistic regression, both pre-treatment VCF (p = 0.02, odds ratio 2.78, 95% CIs 1.21‒6.41) and instability in SINS criteria (p = 0.04, odds ratio 2.28, 95% CIs 1.07‒5.16) were associated factors of VCFs. However, dose per fraction, PTV volume, PTV coverage and collimator option were not correlated with VCF (all p > 0.05).

Discussion

We found that the CK-M is comparable with CK-F with respect to target coverage and conformity, even with larger target volumes. Our results also showed a reduction of approximately 34% in delivery time compared with CK-F, with comparable LC and tolerable toxicities.

Spine SBRT has the advantage of delivering high total doses in few fractions with locally ablative intent. Several SBRT series with CK reported LC rates between 80 and 95%.22–27 We observed a comparable 1 year LC rate of 78%, with no difference between CK-M and CK-F. In addition, treatment of multiple spine levels was correlated with worse LC in accordance with previous study from Balagamwala et al.28 Multilevel disease might inherently be more resistant to treatment and there could be a chance of missing micrometastasis paraspinal region owing to high-conformity of spine SBRT.

We examined tumour characteristics along with several dosimetric parameters of the PTV and spinal cord according to each collimator. We observed that CK-M covered larger PTV volumes without compromising either the CI values or PTV coverage with better dose homogeneity and fall off. We observed that CK-M covered larger PTV volumes without compromising either the CI values or PTV coverage with better dose homogeneity and fall off. In addition, CK-M could deliver higher prescription dose to larger PTV volumes than CK-F without compromising spinal cord dose threshold resulting in comparable LC. Consistent with our study, Jang et al10 evaluated MLC for single or multiple brain metastases, reporting that MLC is only superior in delivery efficacy but not in conformity or dose fall off.However, their study selectively focused on brain metastases, which have significantly smaller treatment volumes and regular shapes compared with those of the spine. MLC-based treatment also exhibited dosimetric improvement when covering even larger and irregular-shaped lesions as shown in the current CK-SBRT. In accordance to the current study, Kathriarachchi et al demonstrated CK-M showed comparable conformity, target coverage, normal organ sparing, tumour control probability and normal tissue complication probability with significantly shortened treatment time in prostate radiotherapy which is relatively complicated (both larger and irregular) plan.

The main disadvantage of CK for spine SBRT is a relatively longer treatment time.29 Of the limited studies regarding delivery time, one found that each CK treatment session lasts for a median 55 min, ranging from 40 to 180 min.30 Most spine SBRT patients (including ours) experience pain,22 making it difficult to remain immobilized for long periods during beam delivery. Hence, we found that one of the important advantages of the MLC is its shorter delivery time.

Total treatment time depends on the total number of beams rather than the total MUs.10 Therefore, finding paths to reduce the total number of beams without compromising PTV coverage and conformity is crucial in improving treatment efficacy. Additionally, minimizing the number of preset nodes is also required to reduce the robot arm traveling time, which is longer than that of reorienting the machine.31 As shown in Table 2, the median MU using the CK-M was approximately 40% for the CK-F and the median beam numbers with the CK-M were lower than for the CK-F by approximately 35%, resulting in a 30% reduction in delivery time. In multiple logistic regression for delivery time (data not shown), both reduced MU (β 0.73) and reduced number of beams (β 0.53) have affected on treatment time (p < 0.001).

There are conflicting reports regarding the dose limitations to the spinal cord for spine SBRT. A recent multiinstitutional investigation32 estimated the probability of RM to be under 5% if the Dmax to the thecal sac was limited to 12.4 Gy for a single fraction. In contrast to previous studies33,34 that suggested thresholds of approximately 10 Gy to the spinal cord to prevent RM, Yamada et al35 suggested limiting the Dmax to under 14 Gy for a single fraction.Our analysis supports their conclusions; no case of RM was observed from 31 cases (22%) treated with maximum dose of 10–15 Gy to the spinal cord for a single fraction until the last follow-up (median 7.8 months, range 1.1–18.8 months). As the MLC leaf is not focused between the fully open and midline positions, increased penumbra-leakage into the neighboring field can occur. Since the CK has no secondary jaw unlike conventional linear accelerators, CK-M might deliver leaked radiation to normal organs such as the spinal cord. However, our study showed that there was no difference in the Dmax to spinal cord or thecal sac and no increase in the incidence of RM.

Although we are the first to report the clinical outcomes of patients treated with either CK-M or CK-F after equipping the CK systems with MLC, this single-institution, retrospective study is subject to selection and physician biases. These results are truly consecutive results of bridge the transition from a fixed cone-based CK to a MLC-based CK. Furthermore, because it includes patients with a variety of different primary tumour types, the prescription dose is wide, and there is heterogeneity of both patients and diseases. Relatively high disease burden as mentioned above also contribute to high censoring rate of local progression. All other treatment techniques, however, remained consistent throughout the period with the exception of the prescription dose and collimator selection. Another potential shortcoming of the study is that its results were based on a relatively short follow-up period possibly owing to the high disease burden cohort; additional follow-up is, therefore, required.

In conclusion, spine SBRT for achieving ablative LC is in greater demand. We found that the recently available CK-M used an equivalent treatment plan within a significantly shorter time compared with CK-F, even with larger target volumes. There was no difference in radiological control between the two collimators and no RM was observed. CK-M could, therefore, be an effective and tolerable treatment option in the era of spine SBRT. These preliminary results support the widespread use of CK-M for spine SBRT.

Funding Statement

This research was supported by a grant of the Korea Health Technology R&D Project through the Korea Health Industry Development Institute (KHIDI) funded by the Ministry of Health & Welfare, Republic of Korea (HI16C0819). The funders had no role in the data collection, interpretation, writing or publication of this manuscript.

Contributor Information

Nalee Kim, Email: nalkim@yuhs.ac.

Ho Lee, Email: holee@yuhs.ac.

Jin Sung Kim, Email: JINSUNG@yuhs.ac.

Jong Geal Baek, Email: JGBAEK@yuhs.ac.

Chang Geol Lee, Email: CG1023@yuhs.ac.

Sei Kyung Chang, Email: skc6603@chamc.co.kr.

Woong Sub Koom, Email: mdgold@yuhs.ac.

Supplementary Table 1.

A comparison of dosimetric parameters between the fixed cone (Fixed) and multileaf collimators (MLC) according to the shape of planning target volume (PTV)

Supplementary Table 2.

A comparison of dosimetric parameters between the fixed cone (Fixed) and multileaf collimators (MLC) according to the planning target volume (PTV) volume

References

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplementary Table 1.

A comparison of dosimetric parameters between the fixed cone (Fixed) and multileaf collimators (MLC) according to the shape of planning target volume (PTV)

Supplementary Table 2.

A comparison of dosimetric parameters between the fixed cone (Fixed) and multileaf collimators (MLC) according to the planning target volume (PTV) volume

Articles from The British Journal of Radiology are provided here courtesy of Oxford University Press

RESOURCES