Volumetric modulated arc therapy (VMAT) and intensity modulated radiation therapy (IMRT) are the principal treatment modalities for localized or locally advanced prostate cancer due to the advanced delivery technique, which can deliver a conformal dose to the target while minimizing the dose to the organs at risk (OARs) [1]. A prescribed dose of 78–80 Gy has conventionally been recommended with up to 8 weeks of fractionated external beam radiation therapy [2]. Such high doses may result in a substantially greater risk of gastrointestinal toxicity, particularly stemming from the rectal dose and bladder hematuria. Radiation doses to the prostate should be better set to reduce possible complications of normal tissue [3].

However, there is inevitably overlap between target and normal tissues because of the need for a margin of the target, considering the existence of inter- and intra-fraction motion and that normal tissue and target tissue are adjacent to each other. Therefore, normal tissue irradiation can be decreased by reducing margins around the target. However, narrower margins could lead to underdosing of the target volume by anatomical mislocalization [4].

Image-guided radiation therapy (IGRT) using cone-beam computed tomography or kilovoltage (kV/kV) gold fiducial marker (FM) imaging in combination with conformal radiation therapy (RT) has been introduced to improve the precision and accuracy of treatment delivery for prostate cancer by allowing smaller treatment margins [5]. Although inter-fraction error can be decreased by an image guidance system, intra-fraction error still occurs and it is difficult to monitor internal organ motion in real-time. Several systems have been introduced to quantify intra-fraction motion, such as megavoltage portal imaging, kilovoltage radiographs, transabdominal ultrasound, and electromagnetic tracking systems. In particular, the Calypso system has been used to monitor the intra- and inter-fraction error [6]. In Calypso, small nonradioactive electromagnetic transponders as intraprostatic FMs communicate with safe radiofrequency wavelengths. The track of an implanted marker is used to interpret prostate motion. The Calypso system can detect internal prostate motion without ionizing radiation in real-time. However, FM implantation must be performed, and obese patients are relatively unsuited to the system. Moreover, the detailed target and OARs cannot be visualized directly [7].

The ViewRay system has been introduced as new magnetic resonance-guided radiotherapy (MRgRT) and gating function [8]. MR imaging allows superior visualization of the prostate, base of the seminal vesicles, and adjacent OARs such as the rectum and bladder. The gating system using real-time cine sagittal MR images can reduce the uncertainty margin [9]. The dose to OARs would be decreased while the volume where planning target volume (PTV) and OAR overlap can be reduced. However, the ViewRay system has some disadvantages in terms of beam delivery. This system uses three Co-60 source as the radiation sources and a multi-leaf collimator (MLC) leaf of 1.05 cm wide at the isoplane (source to isoplane distance: 105 cm) [10]. Previous studies showed the limited plan quality of the ViewRay system due to larger penumbrae and lower penetrating power of Co-60, as well as large MLC width.

A new version of the ViewRay system, MRIdian Linac equipped with a linear accelerator as the radiation source, has been released with a double-focused and double-stacked MLC system. MRIdian Linac can generate 6 MV flattening filter-free (FFF) photon beams with 0.415 cm MLC width at the isoplane (source to isoplane distance: 90 cm).

Several previous studies reported comparisons of plan quality between tri-Co-60-based MRgRT (MRgRT-tri-Co-60) and linac-based VMAT. In the case of lung, spine, and cervix, the plan qualities of MRgRT-tri-Co-60 were inferior to those of linac-based VMAT. Meanwhile, for the prostate, MRgRT-tri-Co-60 can deliver a smaller dose to the bladder and rectum than linac-based VMAT while maintaining target coverage. Another report described the superior plan qualities of linac-based MR-IGRT (MRgRT-linac) compared with linac-based VMAT [5]. In this study, the clinical effectiveness of MrgRT-linac was evaluated and compared with that of MRgRT-tri-Co-60 and conventional linac.

1. Patient selection and simulation

A total of 20 prostate cancer patients treated by VMAT were retrospectively selected at random under approval from the institutional review board. CT scanning was performed with 1.5 mm slice thickness using the Brilliance CT Big Bore™ (Phillips, Cleveland, OH, USA). For every patient, immobilizers were used for fixing knees and feet.

2. VMAT planning

The prostate gland and seminal vesicle were delineated by one oncologist. The primary clinical target volume (CTV) was defined by the prostate and seminal vesicle. The boost CTV was defined by only superior. The primary PTV was expanded by adding a 2 cm margin anteriorly and posteriorly and to the left and right, and a 1 cm margin posteriorly and inferiorly for rectal dose reduction. The boost PTV was defined by adding 0.7 cm isotopically. The prescribed doses were 50.4 Gy in 28 fractions and 30.6 Gy in 17 fractions for the primary and boost plans, respectively.

All VMAT plans were generated with two full arcs using 10 MV based on TrueBeam STx^TM (Varian Medical Systems, Palo Alto, CA, USA) by the Eclipse^TM system (Varian Medical Systems), which was equipped with HD MLC. The PO was used as an optimizer. The dose distribution was calculated by Acuros XB with a 2 mm calculation grid. The plans were normalized to cover 95% of the target volume by at least 100% of the prescribed dose. The sum plan was generated by summing the primary and boost VMAT plans.

3. MRgRT-linac planning

In MRgRT-linac planning, the identical CTV and OAR structure were used, except for PTV. The PTV margin was applied in consideration of the MR-IGRT system. The primary PTV and boost PTV were expanded by adding a 3 mm margin from CTV isotopically due to the gating system using real-time MR imaging.

Step-and-shoot IMRT plans were generated with eight fields using 6 MV FFF photon beams in the treatment planning system of MRIdian system (ViewRay Inc., Oakwood Village, OH, USA). For the primary plan (total prescribed dose=50.4 Gy in 28 fractions), the gantry angles were 40°, 60°, 100°, 160°, 200°, 260°, 300°, and 320°. For the boost plan (total prescribed dose=30.6 Gy in 17 fractions), the gantry angle was slightly changed to 30°, 30°, 65°, 95°, 165°, 195°, 265°, 295°, and 330°. In IMRT optimization, the number of segments was approximately 60 by disable Romeijn optimization and the efficiency value was 0.05. All plans were optimized under the QUANTEC guidelines. For dose calculation, the 2,400,000 histories were performed with a grid size of 3 mm under a 0.35 T magnetic field.

4. MRgRT-tri-Co-60 planning

In MRgRT-tri-Co-60 planning, the CT image and structure set were identical to those in MRgRT planning. Primary and boost plans were generated using 18 fields in seven groups to the same prescribed dose with VMAT and linac-based MR-IGRT. The dose distribution was calculated by a Monte Carlo algorithm provided by ViewRay with a magnetic field. The plan was normalized following the method of VMAT and linac-based MR-IGRT.

5. Evaluation of treatment plans

Dose-volumetric parameters were assessed to evaluate the plan quality. For PTV, the near-minimum (dose received by at least 98% of the target volume, D_98%) and near-maximum doses (D_2%), and D_90%, D_50% (median dose), D_5%, minimum, and maximum doses were calculated. The conformity index (CI) and homogeneity index (HI) were also calculated as follows [11,12]:

For all three OARs, the dose-volumetric histogram was analyzed. For both OARs, the V_{14 Gy}, V_{10 Gy}, and D_10% were calculated for VMAT and MRgRTs. For the entire body, gradient index (GI) was calculated following the formula suggested by Paddick and Lippitz [13]:

For radiobiological model response evaluation, tumor control probability (TCP) and normal tissue complication probability (NTCP) were investigated based on the equivalent uniform dose (EUD). The EUD was calculated according to Niemierko’s phenomenological model as follows [14]:

vi (unit: lenss) stands for the ith partial volume that received dose Di (Gy), while a is a model parameter specific to the normal structure or tumor of interest.

TCP and NTCP were calculated using Niemierko’s EUD-based TCP and NTCP equations [14]:

where TCD₅₀ is the tumor dose controlling 50% of the tumors when the tumor is homogeneously irradiated. TCD₅₀ is the tolerance dose for a 50% complication rate at an interval of 5 years. γ₅₀ is a unit-less model parameter that is specific to the tumor of interest and describes the slope of the dose response curve. The parameters used to calculate TCP and NTCP are listed in Table 1.

Table 1 . Calculation parameters for TCP and NTCP.

Type	Organ	a	γ50	TCD50/TD50	α/β
Tumor	Prostate	−13	2.2	67.5	1.5
Critical Organ	Rectum	8.33	2.66	80	5.4
	Bladder	2	3.63	80	7.5
	Lt femur head	13	2.7	65	3
	Rt femur head	13	2.7	65	3

6. Statistical methods

Analysis of variance (ANOVA) was performed to evaluate the differences of dosimetric parameters among the three modalities. Scheffe’s method was used for post hoc comparisons. A P-value of <0.05 was considered to indicate statistical significance.

Dose distributions of representative patient were shown in Fig. 1. The dose-volumetric parameters related to the target volumes of primary, boost, and sum plans are shown for the three modalities in Table 2. In the case of primary plans, the average values of D_99%, D_95%, and D_5% consistently indicated that the doses to the CTV of the MRgRT-tri-Co-60 plans were the highest among all plan types (all with P<0.05). However, the difference among the three modalities was less than 1 Gy. For D_99% of MRgRT-linac and VMAT, no statistically significant difference was shown in Scheffe’s post hoc test. The CI of MRgRT-linac was closest to 1 (P<0.05). The HI of the three plans were similar. For D_95%, D_2%, D_1%, and CI, Scheffe’s post hoc test showed no statistically significant difference among the three modalities.

Table 2 . Dose-volumetric parameters of the PTV for primary, boost, and sum plans.

Dose-volumetric parameter	MRgRT-linac	MRgRT-tri-Co-60	VMAT	P-value
Primary plan
D99% (Gy)	49.26±0.50	49.83±0.19	48.98±0.45	<0.001
D95% (Gy)	54.50±0.88	54.92±0.38	54.41±0.46	0.026
D5% (Gy)	54.88±0.92	55.20± 0.34	54.73±0.47	0.048
D2% (Gy)	55.14±0.94	55.35±0.32	54.95±0.49	0.162
Maximum dose (Gy)	57.06±1.16	56.10±0.81	56.44±0.59	<0.001
Mean dose (Gy)	52.73±0.59	52.83±0.30	53.04±0.33	0.07
Minimum dose (Gy)	52.75±0.66	52.74±0.38	53.25±0.38	0.002
Conformity index	1.04±0.05	1.09±0.08	1.05±0.05	0.038
Homogeneity index	0.09±0.02	0.10±0.01	0.09±0.01	0.091
Volume (mL)	45.7±13.2	45.7±13.2	320.2±40.5	-
Boost plan
D99% (Gy)	30.18±0.36	29.91±0.36	29.74±0.31	<0.001
D95% (Gy)	32.19±0.46	30.95±0.29	33.06±0.48	0.031
D5% (Gy)	32.35±0.05	34.22±0.39	33.20±20.50	0.051
D2% (Gy)	32.46±0.50	34.37±0.29	33.32±0.51	0.183
Maximum dose (Gy)	33.22±0.59	34.98±0.39	34.01±0.52	<0.001
Mean dose (Gy)	31.48±0.33	32.93±0.32	32.26±0.39	<0.001
Minimum dose (Gy)	31.47±0.38	30.86±0.41	32.42±0.46	0.004
Conformity index	1.07±0.07	1.05±0.07	1.01±0.03	0.032
Homogeneity index	0.06±0.02	0.12±0.01	0.09±0.02	<0.001
Volume (mL)	36.1±11.5	36.1±11.5	79.7±20.7	-
Sum plan
D99% (Gy)	79.65±0.82	79.72±0.84	78.87±0.95	<0.001
D95% (Gy)	81.72±0.34	81.33±0.10	81.97±0.48	0.021
D5% (Gy)	86.22±1.16	88.8±0.87	87.08±0.81	0.042
D2% (Gy)	86.64±1.19	89.23±0.86	87.41±0.82	0.135
Maximum dose (Gy)	88.81±1.42	90.29±0.69	89.28±0.95	<0.001
Mean dose (Gy)	84.31±0.87	85.36±0.54	85.38±0.69	<0.001
Minimum dose (Gy)	72.85±3.57	75.89±1.76	69.78±3.41	<0.001
Conformity index	1.11±0.07	1.12±0.07	1.07±0.05	0.003
Homogeneity index	0.08±0.02	0.11±0.02	0.09±0.01	<0.001

Figure 1. Dose distributions of representative patient for MRgRT-linac (a), MRgRT-tri-Co-60 (b), and volumetric modulated arc therapy (VMAT) (c) sum plans (prescribed dose=8,100 cGy).

In the case of the boost plan, ANOVA revealed a statistically significant difference among the three modalities. However, the maximum difference among the three plans was less than 2 Gy. The CI of VMAT was the closest to 1. The HI of MRgRT-tri-Co-60 was the highest among the three plans, which was statistically significant.

For dosimetric parameters of the sum plans, the values of MRgRT-tri-Co-60 were highest, except for D_95% and mean dose. The differences among the three modalities were shown to be statistically significant for all dosimetric parameters in ANOVA. However, CI did not show statistically significant differences for the three modalities by Scheffe’s test.

The dosimetric parameters of rectum, bladder, and femoral head are listed in Table 3. All parameters related to the rectum were clinically acceptable for the three modalities. For rectum, the average dosimetric parameters of MRgRT-linac, with the exception of D_50%, were lowest among the three modalities, with statistical significance. The average value of D_50% was lowest for VMAT, although this was not statistically significant.

Table 3 . Dose-volumetric parameters of OARs and whole body for primary, boost, and sum plans.

Organ	MRgRT-linac	MRgRT-tri-Co-60	VMAT	P-value
Rectum
V80Gy (%)	0.28±0.62	0.29±0.39	3.07±2.25	<0.001
V75Gy (%)	0.79±1.07	2.17±0.92	4.89±3.41	<0.001
V70Gy (%)	1.54±1.56	5.19±1.75	7.09±5.05	<0.001
V65Gy (%)	2.62±2.12	9.04±3.01	9.73±7.14	<0.001
Maximum dose (Gy)	84.08±2.67	83.19±1.89	87.93±1.28	0.017
D50% (Gy)	35.48±6.07	38.29±4.83	33.53±10.36
D20% (Gy)	47.31±4.31	54.26±4.35	58.80±5.30
Bladder
V80Gy (%)	2.54±2.36	2.30±2.07	10.74±3.74	<0.001
V75Gy (%)	3.74±2.82	4.00±3.07	13.34±4.62	<0.001
V70Gy (%)	5.12±3.35	6.02±4.15	16.07±5.61	<0.001
V65Gy (%)	86.54±1.93	86.35±2.36	87.97±0.87	<0.001
Maximum dose (Gy)	5.87±4.19	15.91±10.09	10.56±9.18	0.007
D55% (Gy)	13.59±9.64	29.94±12.42	24.01±15.56	<0.001
D30% (Gy)	16.38±10.63	34.29±12.53	28.93±16.53	<0.001
D25% (Gy)	1.48±1.78	0.94±1.15	7.99±2.94	<0.001
Femoral head
Maximum dose (Gy)	28.00±2.47	29.90±3.05	28.81±4.34	0.002
D5% (Gy)	21.27±1.83	24.05±2.38	23.76±3.34	<0.001

Regarding the bladder, the average doses of V_80Gy, V_75Gy, V_70Gy, V_65Gy, D_55%, D_30%, and D_25% related to MRgRT-linac were lower than for the other modalities (all with P<0.001). The maximum dose to bladder was lowest for MRgRT-tri-Co-60. In Scheffe’s test, there were no statistically significant differences for MRgRT-linac and MRgRT-tri-Co-60. Although the average doses of V_80Gy, V_75Gy, V_70Gy, and V_65Gy were highest for VMAT, those of D_55%, D_30%, and D_25% were highest for MRgRT-tri-Co-60, albeit with no statistical significance by Scheffe’s test.

The average dose of D_5% to the femoral head was lowest for MRgRT-linac, which reached statistical significance. In post hoc comparisons, statistical significance was not shown for MRgRT-tri-Co-60 and MRgRT-tri-Co-60.

Table 4 shows the average EUD and TCP values of the target and the EUD and NTCP values of rectum, bladder, and femoral heads for the three modalities. For the target, the average EUD and TCP values of MRgRT-linac were lowest, showing no statistical significance. The average differences of EUD and TCP were less than 1 Gy and 1%, respectively. The average EUD and NTCP values of MRgRT-linac were lowest for bladder, rectum, and femoral head. The average EUD and NTCP values of VMAT were higher than those of MRgRT-tri-Co-60 for bladder and rectum (P<0.001), as determined by Scheffe’s post hoc test. The average NTCP values of femoral head were less than 0.0001% for the three modalities.

Table 4 . EUD of target and OAR for sum plan.

Organ		MRgRT-linac	MRgRT-tri-Co-60	VMAT	P-value
Target	EUD (Gy)	84.15	85	85.15	0.080
	TCP (%)	87.41	88.51	88.37	0.091
Bladder	EUD (Gy)	44.43	50.61	55.28	<0.001
	NTCP (%)	0.08	0.3	1.01	<0.001
Rectum	EUD (Gy)	45.4	50.76	55.46	<0.001
	NTCP (%)	0.34	0.96	2.23	<0.001
Femoral head	EUD (Gy)	15.04	18.09	17.79	<0.001
	NTCP (%)	0	0	0	<0.001

EUD, equivalent uniform dose; OAR, organ at risk; NTCP, normal tissue complication probability; MRgRt-linac, magnetic resonance-guided radiotherapy linear accelerator; MRgRT-tri-Co-60, magnetic resonance-guided radiotherapy three Co60; VMAT, volumetric modulated arc therapy..

In this study, the quality of treatment plans generated by MRgRT-linac, MRgRT-tri-Co-60, and VMAT were compared in terms of plan parameters, dosimetric parameters, and radiobiological evaluation. For the three plans, dosimetric and radiobiological parameters related to PTV were similar. In the case of MRgRT, the method of static IMRT was used to deliver the radiation by the step-and-shoot technique. Therefore, the CI of VMAT was slightly lower due to the delivery technique using two full arcs [15]. In addition, the treatment time of VMAT was shorter than that of MRgRT. Moreover, MRgRT-linac had a higher dose rate than MRgRT-tri-Co-60. However, the gantry rotation time of MRgRT-tri-Co-60 was lower than that of MRgRT-linac since MRgRT-tri-Co-60 had three sources. Therefore, the treatment times were analogous among the MRgRTs.

With regard to critical organs, the MRgRT plan was superior to the VMAT plan. The parameters of MRgRT-linac were better than those of MRgRT-tri-Co-60. The PTV of MR-based techniques was smaller than that of CT-based techniques (i.e., VMAT) due to real-time gating based on MR. The MRgRT can deliver significantly lower doses to rectum and bladder compared with the VMAT plans because of the PTV size, while the PTV coverage of the three plans was comparable. In particular, the MRgRT-linac had superior dose distribution compared with MRgRT-tri-Co-60. The MLC of MRgRT-linac had a finer resolution than that of MRgRT-tri-Co-60. In addition, the penumbra of MRgRT-linac was sharper than that of MRgRT-tri-Co-60 [5].

The trend of dosimetric parameters showed similar results to radiobiological parameters. Among the three modalities, EUD and TCP were comparable. All of the plans were optimized to cover the PTV and normalized to the prescribed dose. For NTCP of bladder and rectum, MRgRT-linac showed the lowest values. However, the differences between MRgRTs were less than 1%. For femoral head, the NTCP of all plans was close to 0%.

In this paper, we describe that MRgRT-linac could decrease the doses to rectum and bladder while maintaining PTV coverage. For prostate cancer, MRgRT-linac could reduce the PTV margin due to superior soft-tissue contrast and real-time monitoring of tumor motion. The probability of complications could be decreased because of its advantages. MRgRT-linac is effective and advantageous for prostate cancer.

Approval for this study was obtained from the Institutional Review Board of Seoul National University Hospital (IRB No. 1901–059-1002).

The authors have nothing to disclose.

The data that support the findings of this study are available on request from the corresponding author.

Conceptualization: Jong Min Park and Jung-in Kim. Data curation: Chang Heon Choi and Jin Ho Kim. Formal analysis: Chang Heon Choi and Jaeman Son. Funding acquisition: None. Investigation: Jin Ho Kim. Methodology: Chang Heon Choi. Project administration: Jin Ho Kim. Resources: Jin Ho Kim and Chang Heon Choi. Software: None. Supervision: Jong Min Park and Jung-in Kim. Validation: Chang Heon Choi. Visualization: Chang Heon Choi. Writing – original draft: Chang Heon Choi and Jaeman Son. Writing – review & editing: Jong Min Park and Jung-in Kim.