Advances in Oncology 1 (2021) 29–39

ADVANCES IN ONCOLOGY

MRI-Guided Radiation Therapy

Sangjune Laurence Lee, MD, MSE, FRCPCa,b, William A. Hall, MDc,
Zachary S. Morris, MD, PhDa, Leslie Christensen, MAd, Michael Bassetti, MD, PhDa,*
a
Department of Human Oncology, University of Wisconsin Hospital and Clinics, Madison, WI, USA; bDepartment of Oncology, Division of
Radiation Oncology, University of Calgary, Calgary, AB, Canada; cDepartment of Radiation Oncology, Medical College of Wisconsin,
Milwaukee, WI, USA; dUniversity of Wisconsin School of Medicine and Public Health, Madison, WI, USA

KEYWORDS
 MR-guided radiation therapy  MR guidance  MR linear accelerator  MR-linac  MRIgRT  Plan adaptation
 Real-time imaging

KEY POINTS
 Integrated MRI-guided radiation therapy (MRIgRT) systems have recently been developed with growing clinical adoption
since 2014.
 MRIgRT systems have superior soft tissue contrast and are capable of real-time treatment gating and on-table radiation
plan adaptation.
 MRIgRT presents many technical challenges but has great potential to improve the therapeutic ratio of radiation treatment.
 MRIgRT can be used to treat malignancies in all body sites, although long-term clinical outcomes are currently pending.
 Collective efforts will be required to demonstrate improved clinical outcomes to offset the increased cost of MRIgRT
systems.

Video content accompanies this article at http://www.advances-oncology.com/.

INTRODUCTION while radiation is being delivered. This enables daily

The development of the integrated MRI-guided radio- adaptation of the radiation plan according to changes
therapy systems (MRIgRT) is an engineering feat with in daily anatomy, and real-time respiratory gating. In
growing adoption in hospitals over the past 6 years. many circumstances, MRIgRT may therefore allow for
Currently, there are 2 MRIgRT systems in development, tighter treatment margins and this has the potential to
and 2 systems that are commercially available [1–4]. enable safe delivery of higher doses per fraction. This
ViewRay (Cleveland, OH) was the first to market with could translate into better tumor control, less radiation
a 0.35-T MRI combined with a radioactive cobalt system toxicity, and/or fewer treatment visits.
in 2014 that has since been upgraded to linear The purpose of this review was to survey the inroads
accelerator–based photon radiation (MR-linac) [2,5]. that the MRIgRT has made in the treatment of cancers in
Elekta treated its first patient with the 1.5-T MR-linac certain anatomic locations and to discuss new opportu-
system in 2017 [6]. nities for using MRIgRT to provide more effective radia-
Unlike the cone beam computed tomography tion therapy. For each body site, we highlight physics
(CBCT)-based linear accelerator, MRIgRT can provide problems and technical challenges that should be consid-
continuous real-time, high soft tissue contrast imaging ered when treating patients with MRIgRT.

*Corresponding author. Department of Human Oncology, University of Wisconsin, University Hospital L7/B36, 600 Highland
Avenue, Madison, WI 53792. E-mail address: [email protected]

https://doi.org/10.1016/j.yao.2021.02.003 www.advances-oncology.com
2666-853X/21/ © 2021 Elsevier Inc. All rights reserved. 29
30 Lee et al

PANCREATIC CANCER AND ONLINE in 5 fractions. In the study by Palacios and colleagues
ADAPTIVE RADIOTHERAPY [18], 84 patients with adrenal metastases were treated
Improvements in the effectiveness of systemic therapy with breath-hold gating and a range of ablative doses
increase the importance of local therapy for pancreatic depending on proximity to OARs. In the study by
cancer through dose-escalation strategies of intact tu- Henke and colleagues [19], 20 patients with abdominal
mors [7]. Because of the close proximity and interfrac- metastases were treated with breath-hold gating and
tion motion of the duodenum and stomach to the 50 Gy in 5 fractions. In all 3 studies, on-table adapta-
pancreas, stereotactic ablative radiotherapy (SABR) for tion significantly decreased OAR dose constraint viola-
pancreatic cancer is difficult, with the rate of grade 2 tions and improved target coverage compared to
or higher gastrointestinal (GI) acute toxicity between without adaptation. Henke and colleagues [19] re-
2% and 80% and late toxicity between 10% and 50% ported no grade 3 toxicities at 6 months of follow-up.
on convectional CBCT-based linear accelerators [8]. MRIgRT may improve the safety and efficacy of SABR
The main advantage of treating patients with pancreatic by enabling isotoxic dose escalation. In conventional
cancer with MRIgRT may be its ability to perform online radiotherapy, planning objectives aim to cover the
plan adaptation [9]. In CBCT-based treatments, a single target with a homogeneous dose. In isotoxic dose esca-
radiotherapy plan is designed and used for all fractions lation, the dose to the target is increased until the dose
because of the inability to visualize the changing anat- constraint of a proximal OAR is met, and the dose dis-
omy of the pancreas and surrounding luminal organs. tribution over the target is highly heterogeneous. Het-
With the MR-linac, the radiation plan can be adapted erogeneous dose coverage of the PTV minimizes the
while the patient is on the treatment table before the de- dose to OARs but requires better motion management
livery of each fraction so as to take into account the compared with homogeneous dose coverage (Fig. 1)
changing anatomy of the day [10]. Plan adaptation is [20]. Isotoxic target dose escalation may improve sur-
more time-consuming and labor-intensive than the vival outcomes, although clinical outcomes from
CBCT-based workflow, but can be expedited by only thoracic treatments have been mixed [21–23].
recontouring organs at risk (OARs) within a 3-cm radius
from the planning target volume (PTV) [11]. In studies
in which all fractions are adapted, the chance that dose CENTRAL NERVOUS SYSTEM CANCERS
objectives are met increases from 43.9% to 83.0% [12]. AND IMAGING BIOMARKERS
In a study of 44 patients with unresectable pancreatic MRI is already incorporated in the workflow for the
cancer treated with cobalt-based MRIgRT SABR, patients treatment of central nervous system (CNS) malig-
treated with a biological equivalent dose (BED) greater nancies and metastases due to its ability to visualize
than 70 Gy had a 2-year overall survival of 49% versus and distinguish normal brain and intracranial tumors.
30% in those treated with a BED less than 70 Gy, with MRIs are rigidly registered to the planning CT scan to
no grade 3 toxicities in the high-dose group [13]. A guide highly conformal stereotactic radiosurgery
prospective, multi-institutional trial prescribing 50 Gy (SRS), SABR, and fractionated stereotactic radiotherapy
in 5 fractions with adaptive MRIgRT to patients with (FSRT) radiation plans, often in 1 to 5 fractions in the
inoperable pancreatic cancer is currently under way case of brain metastases and spine metastases [24].
(NCT03621644). SABR to spinal bone metastases can provide improved
local control compared with conventional radiation
[25]. To avoid damaging the spinal cord, patients typi-
OLIGOMETASTASES AND ISOTOXIC cally require a diagnostic-quality MRI to be rigidly fused
PLANNING to a CT. Both images must be acquired while the patient
Improved imaging techniques, such as prostate-specific is fixed into position in a rigid mold. A pilot trial at
membrane antigen (PSMA) PET, and new laboratory Washington University in St. Louis is exploring the
testing, such as circulating tumor DNA, can improve feasibility of delivering spine SABR on the same day
the detection of metastatic disease, increasing the as simulation on the MR-linac, reducing the duration
chance of improving overall survival when combined between simulation and treatment, which can be
with the early use of SABR [14–16]. Several groups several days (NCT03878485). Although promising,
have reported their experiences with MRIgRT SABR for obtaining the highly conformal radiation dose distribu-
oligometastatic disease. In the study by Winkel and col- tions needed for SABR on MRIgRT units can be difficult
leagues [17], 14 patients with oligometastatic disease in because non-coplanar beams/arcs are not possible with
the pelvic or para-aortic region were treated with 35 Gy currently available devices.
 MRI-Guided Radiation Therapy 31

FIG. 1 Axial views of SABR dose distribution for an oligometastatic lesion in left lower lung close to the
stomach at (A) initial MRI simulation, (B) day of treatment before plan adaptation, and (C) after plan adaptation
to avoid overdosing the stomach (arrow) (D) DVH demonstrating an increase in dose to the planning target
volume (PTV) after re-optimization. See Video 1 for sagittal cine of tumor tracking. Pink 5 Planning target
volume, red 5 stomach. DVH, Dose volume histogram; SABR, stereotactic ablative radiotherapy.

Treating CNS malignancies with the MRIgRT opens as prostate-specific membrane antigen-targeted nano-
opportunities to understand changes in the tumor plexes that are both therapeutic and diagnostic under
biology during treatment with advanced MRI pulse se- MRI, is an emerging field that could help focus radia-
quences due to relative lack of motion at this body tion treatment to high-risk areas [32].
site. The treatment of 3 patients with glioblastoma
with cobalt-based MRIgRT was demonstrated by Mehta
and colleagues [26] and showed changes in the postop- HEAD AND NECK CANCER AND DEEP
erative cavity and cerebral edema volumes during the 6- LEARNING ENABLED AUTO-CONTOURING
week course of radiation therapy. By manipulating the Although radiation therapy plays an integral role in treat-
MRI acquisition parameters, information regarding tu- ing head and neck (HN) malignancies, delivery of radia-
mor cellularity, vascularity, and biochemical makeup tion is difficult because of the close proximity to
can be gathered [27]. Although advanced pulse se- radiation-sensitive normal tissues. In a study by Raghavan
quences have higher signal-to-noise ratio at 1.5 T, and colleagues [33] of 6 patients treated with cobalt-based
research is under way to incorporate these pulse se- MRIgRT, the primary tumor and parotid gland volumes
quences into 0.35-T MR-linacs [28]. In the treatment decreased, and their positions shifted significantly during
of glioblastoma, advanced imaging can identify areas the 7 weeks of radiation therapy. Chen and colleagues
at higher risk of recurrence for radiation-boosting strate- [34] reported on 18 patients with HN primaries treated
gies [29]. MRI imaging biomarkers could also introduce with cobalt-based MRIgRT with disease control and
a paradigm shift in the target dose objectives. Tradition- quality-of-life outcomes similar to conventional CBCT-
ally, a single homogeneous dose was recommended for based radiation therapy. However, even with improved
treating the tumor. Advanced physiologic imaging visualization of the tumor and OARs with daily MRI,
could identify higher-risk areas within the tumor and only a minority of patients required plan adaptation
provide rationale for heterogeneous dose distribution [34]. Because the anatomy of the HN can change over
[30]. Instead of treating each tumor to a fixed dose, weeks during radiation treatment, several investigators
treatment could be given until an MRI biomarker that are exploring the use of weekly MRIs for plan adaptation
is strongly correlated with outcomes reaches a certain to more accurately target the tumor and spare OARs
threshold (Fig. 2) [31]. Use of injectable agents, such (NCT03972072) [35].
32 Lee et al

FIG. 2 Hypothesized imaging biomarker “threshold” goal (arrow) for a radiation treatment dosing. Imaging
response during treatment could be closely correlated with a validated imaging biomarker and clinical/
pathologic end-point. (Adapted from Hall WA, Paulson ES, van der Heide UA, et al. The transformation of
radiation oncology using real-time magnetic resonance guidance: A review. Eur. J. Cancer. 2019;122:42–52.;
with permission. (Figure 3 in original).)

In the traditional workflow, a patient undergoes a have not yet been reported [50,51]. Kennedy and col-
simulation scan, which is used to contour the target leagues [52] reported the use of single-fraction, adju-
and OARs. The distribution of beams and physics qual- vant partial breast irradiation for 50 patients with
ity checks are based on these contours. This process usu- early-stage disease, with most patients treated using
ally takes several days. For HN plans, the contouring can cobalt-based MRIgRT. At 25 months’ median follow-
be especially time-consuming due to complex anatomy. up, there were no grade 3 toxicities and no in-field
Deep learning may help decrease the amount of work recurrences [52]. The results of a prospective trial
required for contouring. Deep learning is a field of arti- investigating the use of neoadjuvant single-fraction
ficial intelligence in which computers learn how to pro- ablative radiation with an MRI simulation are currently
duce contours based on a previously created set of pending [53].
contoured images [36,37]. Deep learning has been One of the challenges facing MRIgRT in breast cancer
applied to CTs of the HN to automatically contour is the electron return effect (ERE). When x-ray photons
the tumor and high-risk expansion volumes as well as hit tissue within the body, a cascade of electrons is pro-
OARs [38,39]. In a study by Tong and colleagues [40], duced that generally travels in the same direction as the
twenty-five 0.35-T MRIs were used to develop a deep incident photons and eventually deposits free radicals
learning algorithm to contour bony and soft tissue that damage DNA. In the magnetic field of MRIgRT sys-
OARs in the HN. tems, the Lorentz force pushes the moving electron par-
ticles in a perpendicular direction with a force that is
proportional to the magnitude of the magnetic field.
BREAST CANCER AND THE ELECTRON Electrons that are given off from the tissue to the air cir-
RETURN EFFECT cle back to the skin in what has been called the ERE
Hypofractionated radiation is the standard of care for (Fig. 3). Mitigating the ERE, which increases skin
early-stage breast cancer with whole breast adjuvant dose, is a challenge for MRIgRT, and is especially rele-
therapy and is gaining acceptance with locoregional vant for breast cancer treatment because of the impor-
therapy for more advanced disease [41–43]. Accelerated tance of cosmesis. To avoid unwanted ERE irradiation
partial breast irradiation (APBI) can limit the radiation outside the breast treatment fields, use of a 1-cm bolus
field to the postoperative bed with acceptable local con- shielding the upper torso is recommended for 0.35-T
trol and cosmesis compared with hypofractionated and 1.5-T MR-linacs [51,54]. The ERE has a larger
whole breast treatments [44,45]. As targeting becomes impact on targets near an air-tissue interface, such as
more focused and fractionation schedules become breast, lung, or GI cancers but can be minimized by tak-
shorter, MRI studies suggest that tracking the position ing the magnetic field effects into account during plan
of the breast and the volume of the seroma become design [55,56].
more critical [46,47]. Investigations are being conduct-
ed to determine whether neoadjuvant MRIgRT could
reduce the volume of normal breast tissue irradiated LUNG CANCER AND RESPIRATORY
or detect a pathologic complete response [48,49]. MOTION MANAGEMENT
Cobalt-based MRIgRT APBI has been used to treat pa- SABR provides a high rate of local control, usually with
tients with breast cancer, although long-term outcomes minimal morbidity, for patients with medically
 MRI-Guided Radiation Therapy 33

FIG. 3 Illustration of the electron return effect, for left whole breast irradiation by means of 2 tangential fields.
The edges of the photon beams are depicted by the blue lines. (Left) In the absence of a magnetic field,
secondary electrons leave the breast at the tissue-air interface. (Right) With a magnetic field, secondary
electrons return to the breast, increasing the skin dose. (From van Heijst T, den Hartogh M, J W Lagendijk J,
et al. MR-guided breast radiotherapy: Feasibility and magnetic-field impact on skin dose. Phys. Med. Biol.
2013;58:5917–5930 Ó Institute of Physics and Engineering in Medicine. Reproduced by permission of IOP
Publishing. All rights reserved.)

inoperable early-stage lung cancer or metastases to the delivery time of a single fraction but enables tighter
lung [57]. Although lung lesions are easy to see on treatment margins by eliminating the need for an inter-
CBCT, organs in the central mediastinum are better nal target volume to account for tumor motion during
visualized on MRI. In a study of 5 ultra-central lung le- respiration (Fig. 4).
sions, Henke and colleagues [58] reported no grade 3
acute toxicity within 6 months after cobalt-based
MRIgRT. Finazzi and colleagues [59] reported the treat- PROSTATE CANCER AND SYNTHETIC
ment of 54 patients with higher-risk lung lesions due to COMPUTED TOMOGRAPHY
central location, re-irradiation, or interstitial lung dis- Because prostate cancer has a low a=b ratio, hypofrac-
ease with either cobalt-based or linac-based MRIgRT. tionated treatments may improve the therapeutic ratio
At 12 months, local control was 95.6% with 8% grade [61]. MR-guided radiation can allow for online adaptive
3 toxicities and no grade 4 to 5 toxicities. replanning and can also inform dose escalation to intra-
MR-linacs may be able to achieve lower rates of prostatic lesions while avoiding the urethra. In a study
toxicity with respiratory gating. In CBCT-based radia- of 25 patients with mainly intermediate-risk prostate
tion, a volume containing all possible positions of the cancer treated with 35 Gy in 5 fractions with daily adap-
lung tumor during a respiratory cycle is typically irradi- tation on a 1.5-T MR-linac, 16% of patients developed
ated. Although respiratory gating with a CBCT-based acute grade 2 GI or genito-urinary (GU) toxicity, and
linear accelerator with techniques such as active breath- there were no grade 3 toxicities [62]. In another study
ing control is feasible, gating may be easier and more ac- of 101 patients with an even mix of intermediate-risk
curate on MRIgRT units because MRIgRT can image the and high-risk prostate cancer treated with 36.25 Gy in
tumor constantly throughout radiation treatment. 5 fractions with daily adaptation on a 0.35-T cobalt-
Therefore, the treatment volume can be minimized by based MRIgRT, the cumulative rate of acute grade 2 GI
irradiating the tumor only when it is in a specific loca- or GU toxicity was 28.8% without any grade 3 toxic-
tion during the respiratory cycle [60]. On the ViewRay ities [63].
system, a single sagittal slice constantly images the tu- One way to shorten the overall radiotherapy work-
mor. The patient is normally asked to perform a flow is to omit the CT scan. In the current workflow,
maximum inspiration breath-hold. The system is able the CT scan is registered to the MRI to provide electron
to track the tumor as the patient breathes. The beams density information, which is necessary for calculating
are turned on automatically when the tumor is within dosimetry. However, errors in registration and chang-
a certain window and turned off when the patient re- ing anatomy can reduce accuracy. Alternatively, the
sumes respiration (Video 1). This process increases the MRI can be used to create a “synthetic CT.” There are
34 Lee et al

FIG. 4 Coronal CT slice of a patient with lung cancer. Dose distribution of (A) respiratory gated treatment in
which the tumor is tracked and (B) conventional free-breathing treatment with larger treatment margins. (C)
Dose difference between the tracked treatment versus the conventional treatment. (From Menten MJ, Fast
MF, Nill S, et al. Lung stereotactic body radiotherapy with an MR-linac – Quantifying the impact of the
magnetic field and real-time tumor tracking. Radiother. Oncol. 2016;119:461–466.; with permission. (Figure 4
in original).)

2 broad categories of methods to create synthetic CTs. GYNECOLOGIC MALIGNANCIES AND FIELD
In voxel-based methods, information about the MRI SIZE
voxel intensity, usually from 2 or more pulse se- The use of MRI for 3-dimensional high-dose-rate
quences, is used to assign electron densities. These brachytherapy planning in the treatment of cervical can-
methods are less reliant on information about the cer has improved locoregional control and survival rates
location of the voxel within the MRI. Voxel-based while reducing late morbidity [65]. External beam radi-
methods are predominantly deep learning algorithms, ation therapy (EBRT) is commonly used adjuvantly for
in which the computer is given matched MRIs and CTs endometrial cancer and as a component of definitive
and then is able to create a CT from a new MRI. In chemoradiation therapy followed by brachytherapy
atlas-based methods, the position of each MRI voxel for cervical cancer. EBRT has traditionally been deliv-
is aligned to a predefined reference atlas of anatomic ered in a large 4-field box technique due to the potential
structures or set of reference atlases through image for large day-to-day motion of the uterus. More
registration. Each structure in the atlas is assigned to conformal techniques use large margins to encompass
a particular electron density value. Currently, there the range of motion of the uterus. MRIgRT for such gy-
are 2 synthetic CT algorithms for prostate radiotherapy necologic malignancies has the potential to reduce
approved by the Food and Drug Administration [64]. toxicity to OARs with daily adaptive planning [66].
The use of synthetic CTs could lead to more efficient PTV margins in these locations can be reduced from
workflows, in which the time between simulation 1.5 to 0.5 cm with MRIgRT [67]. Case reports on the
and treatment is decreased from several days to a sin- treatment of cervical cancer with MRIgRT without daily
gle day. adaptation show considerable movement in the PTV
 MRI-Guided Radiation Therapy 35

and shrinkage of the gross tumor volume over the and wait” strategies [76]. Rectal cancers have considerable
course of radiation [68,69]. variability in bowel and bladder filling. MRIgRT can
A challenge in treating gynecologic malignancies with enable safe dose-escalation strategies and help determine
MRIgRT is the limited field size for treatment. The inte- the best candidates for nonoperative management
gration of the linear accelerator with the MRI results in through the use of quantitative biomarkers [77,78].
a smaller maximum treatable field size compared to MRIs of extremity soft tissue sarcomas are recom-
that of a conventional linear accelerator. On the 1.5-T mended to delineate the gross tumor volume (GTV)
and 0.35-T MR-linac, the maximum superior-inferior di- and peritumoral edema for preoperative radiotherapy
rection field size is 22 cm and 24 cm, respectively. Sites [79]. The Phase III EORTC STRASS trial failed to show
with extensive targets in the cranial-caudal direction, a benefit in preoperative radiotherapy for retroperito-
including gynecologic malignancies and HN malig- neal sarcomas [80]. Alternative approaches to
nancies, do not fit in the MRIgRT treatment field in improving local control for retroperitoneal sarcomas
40% of cases [2,70]. In addition, magnetic field inhomo- include the use of MR-guided hypofractionated radia-
geneities cause geometric distortions outside the treat- tion 60 Gy in 3 to 8 fractions (NCT03972930). MR-
ment isocenter. On the 0.35-T MR-linac, although guided radiation has also been used to treat challenging
there are less than 1-mm distortions within a 10-cm cases, including a pediatric rhabdomyosarcoma of the
radius of the isocenter, at 20 to 25 cm from the isocenter diaphragm and a left ventricle cardiac fibroma [81–83].
the distortions can be up to 7 mm [71]. A potential op- Although MRI has excellent soft tissue contrast and
tion for overcoming the limited field size with MRIgRT does not involve ionizing radiation, it can suffer from
is to use 2 isocenters [70]. a range of imaging artifacts. Periodic respiratory mo-
tion can create a “ghosting” artifact, a faint repetition
of structures across the image. Strategies such as
OTHER SITES (LIVER, RECTUM, AND breath-holds or detecting the position of the dia-
SARCOMA) AND MRI ARTIFACTS phragm can reduce these motion artifacts. Another
Tumors within the liver, including hepatocellular carci- broad category of artifacts is geometric distortion
noma, cholangiocarcinoma, and metastases, are diffi- due to magnetic field inhomogeneity. Structures in
cult to visualize on CBCT because of respiratory the image can be represented in the wrong location.
motion and poor soft tissue contrast. Consequently, Certain image sequences, such as diffusion weighted
larger margins are often required to account for the un- imaging (DWI), are particularly susceptible to geo-
certainty in the tumor position. Although MRIgRT is metric distortion, and caution should be exercised
better at visualizing the tumor, soft tissue contrast can when targeting lesions based on DWI. Metal objects
further be enhanced with the use of gadohexetate intra- are generally restricted from MRIs, as they can be pro-
venous contrast [72]. In a retrospective review of 26 pa- pelled by large magnetic forces. MR-compatible
tients with metastatic liver lesions and hepatocellular metallic prostheses and devices can cause dark “band-
carcinomas treated with cobalt-based MRIgRT SABR to ing” artifacts that distort the image near the object,
a median dose of 50 Gy in 5 fractions, the 2-year overall which can be especially relevant for patients with sar-
survival was 60% and 21-month local control was 80%, coma [84]. Even iron supplements and iron-fortified
without any grade 4 toxicities [73]. In a study of 17 cereals can cause banding artifacts, and patients
patients with unresectable locally advanced cholangio- should be instructed to avoid consuming these during
carcinoma treated with cobalt-based MRIgRT SABR abdominal radiation (Fig. 5) [85].
with a median dose of 40 Gy in 5 fractions, the 2-year
overall survival was 46%, and local control was 73%,
without any grade 4 toxicities [74]. Colorectal metas- DISCUSSION
tases to the liver have lower control rates compared with This review summarizes the clinical applications of
other types of histology [75]. The potential for safe dose MRIgRT, progress made so far, and areas under investi-
escalation to colorectal metastases using an isotoxic gation. We have highlighted physics problems and tech-
approach on the MR-linac is currently being investi- nical challenges and opportunities that should be
gated (NCT04020276). further explored for MRIgRT to reach its full potential.
In locally advanced rectal cancer, the combination of Although there are many dosimetric studies, most of
preoperative chemotherapy and chemoradiation can in- the clinical experiences reported have been with 0.35-
crease the rate of complete pathologic response, and trials T cobalt-based MRIgRT, and outcomes focus mainly
are being conducted to investigate nonoperative “watch on acute toxicity rates, which have been favorable.
36 Lee et al

FIG. 5 (Left) Axial and (right) coronal views of a patient a few hours after ingesting iron-fortified breakfast
cereal causing susceptibility artifact. (From Green O, Henke LE, Parikh P, et al. Practical Implications of
Ferromagnetic Artifacts in Low-field MRI-guided Radiotherapy. Cureus. 2018;10:e2359.; with permission.
(Figure 2 in original).)

The MR-linac units are approximately twice the cost Center for Advancing Translational Sciences, NIH,
of a well-equipped conventional linear accelerator. To Award Number KL2TR001438. The content is solely
justify the purchase of these more expensive units, can- the responsibility of the author(s) and does not neces-
cer centers must coordinate efforts to demonstrate sarily represent the official views of NIH. Dr Z. S. Morris
favorable clinical outcomes [86]. MRIgRT can improve reports personal fees from ViewRay, outside the submit-
the therapeutic ratio, and the operational costs will ted work; Member of Scientific Advisory Board, Archeus
likely shrink with the use of automation to expedite Technologies; and on the Scientific Advisory Board for
the adaptive workflow and with shorter hypofractio- Seneca Therapeutics. Dr M. Bassetti reports meeting
nated treatments [87]. Multi-institutional trials and co- travel reimbursement support from ViewRay, and Clin-
ordinated efforts will be essential to establishing MR- ical Trial Support from Merck, AstraZeneca, and EMD
guided radiotherapy as a treatment option for patients Serono.
with cancer globally.

SUPPLEMENTARY DATA
CLINICS CARE POINTS Supplementary data related to this article can be found
online at https://doi.org/10.1016/j.yao.2021.02.003.
 MRIgRT systems have superior soft tissue contrast
and are capable of more accurate and precise treat-
ments with real-time imaging and treatment gating REFERENCES
during radiation delivery and with radiation plan [1] Pollard JM, Wen Z, Sadagopan R, et al. The future of
adaptation while the patient is on the treatment table. image-guided radiotherapy will be MR guided. Br J Ra-
 MRIgRT presents many technical challenges but has diol 2017;90:20160667.
great potential to improve the therapeutic ratio of ra- [2] Klüter S. Technical design and concept of a 0.35 T MR-
diation treatment. Linac. Clin Transl Radiat Oncol 2019;18:98–101.
 Collective efforts will be required to demonstrate [3] Lagendijk JJW, Raaymakers BW, van Vulpen M. The mag-
improved clinical outcomes to offset the increased netic resonance imaging–linac system. Semin Radiat On-
cost of MRIgRT systems. col 2014;24:207–9.
[4] Mutic S, Dempsey JF. The ViewRay system: magnetic
resonance–guided and controlled radiotherapy. Semin
Radiat Oncol 2014;24:196–9.
DISCLOSURE [5] Henke LE, Contreras JA, Green OL, et al. Magnetic reso-
nance image-guided radiotherapy (MRIgRT): a 4.5-year
Dr S.L. Lee reports receiving travel reimbursement sup-
clinical experience. Clin Oncol (R Coll Radiol) 2018;
port from ViewRay. W.A. Hall receives departmental
30:720–7.
research and travel support from Elekta AB, and has a K [6] Raaymakers BW, Jürgenliemk-Schulz IM, Bol GH, et al.
grant from the National Institutes of Health (NIH). First patients treated with a 1.5 T MRI-Linac: clinical
The project described was supported by the National proof of concept of a high-precision, high-field MRI
 MRI-Guided Radiation Therapy 37

guided radiotherapy treatment. Phys Med Biol 2017;62: radiotherapy: the price of dose uniformity. Acta Oncol
L41–50. 2020;59:558–64.
[7] Koay EJ, Hanania AN, Hall WA, et al. Dose-Escalated Ra- [21] Bainbridge HE, Menten MJ, Fast MF, et al. Treating
diation Therapy for Pancreatic Cancer: A Simultaneous locally advanced lung cancer with a 1.5T MR-Linac - Ef-
Integrated Boost Approach. Pr Radiat Oncol 2020;10: fects of the magnetic field and irradiation geometry on
e495–507. conventionally fractionated and isotoxic dose-escalated
[8] Reyngold M, Parikh P, Crane CH. Ablative radiation ther- radiotherapy. Radiother Oncol 2017;125:280–5.
apy for locally advanced pancreatic cancer: techniques [22] Zhao Q, Liu M, Wang Z, et al. High dose radiation ther-
and results. Radiat Oncol 2019;14:95. apy based on normal tissue constraints with concurrent
[9] Boldrini L, Cusumano D, Cellini F, et al. Online adaptive chemotherapy achieves promising survival of patients
magnetic resonance guided radiotherapy for pancreatic with unresectable stage III non-small cell lung cancer. Ra-
cancer: state of the art, pearls and pitfalls. Radiat Oncol diother Oncol 2020;145:7–12.
2019;14:71. [23] De Ruysscher D, van Baardwijk A, Wanders R, et al. Indi-
[10] Luterstein E, Cao M, Lamb J, et al. Stereotactic MRI- vidualized accelerated isotoxic concurrent chemo-
guided Adaptive Radiation Therapy (SMART) for locally radiotherapy for stage III non-small cell lung cancer: 5-
advanced pancreatic cancer: a promising approach. Cur- year results of a prospective study. Radiother Oncol
eus 2018;10:e2324. 2019;135:141–6.
[11] Bohoudi O, Bruynzeel AME, Senan S, et al. Fast and [24] Cao Y, Tseng C-L, Balter JM, et al. MR-guided radiation
robust online adaptive planning in stereotactic MR- therapy: transformative technology and its role in the
guided adaptive radiation therapy (SMART) for pancre- central nervous system. Neuro Oncol 2017;19:ii16–29.
atic cancer. Radiother Oncol 2017;125:439–44. [25] Zeng KL, Tseng C-L, Soliman H, et al. Stereotactic body
[12] Bohoudi O, Bruynzeel AME, Meijerink MR, et al. Identi- radiotherapy (SBRT) for oligometastatic spine metasta-
fication of patients with locally advanced pancreatic can- ses: an overview. Front Oncol 2019;9:337.
cer benefitting from plan adaptation in MR-guided [26] Mehta S, Gajjar SR, Padgett KR, et al. Daily tracking of
radiation therapy. Radiother Oncol 2019;132:16–22. glioblastoma resection cavity, cerebral edema, and tumor
[13] Rudra S, Jiang N, Rosenberg SA, et al. Using adaptive volume with MRI-guided radiation therapy. Cureus
magnetic resonance image-guided radiation therapy for 2018;10:e2346.
treatment of inoperable pancreatic cancer. Cancer Med [27] van der Heide UA, Thorwarth D. Quantitative imaging
2019;8:2123–32. for radiation oncology. Int J Radiat Oncol 2018;102:
[14] Maurer T, Eiber M, Schwaiger M, et al. Current use of 683–6.
PSMA–PET in prostate cancer management. Nat Rev [28] Gao Y, Han F, Zhou Z, et al. Distortion-free diffusion
Urol 2016;13:226–35. MRI using an MRI-guided Tri-Cobalt 60 radiotherapy
[15] De Michino S, Aparnathi M, Rostami A, et al. The utility system: sequence verification and preliminary clinical
of liquid biopsies in radiation oncology. Int J Radiat On- experience. Med Phys 2017;44:5357–66.
col 2020;107(5):873–86. [29] Kim MM, Parmar HA, Aryal MP, et al. Developing a pipe-
[16] Palma DA, Olson R, Harrow S, et al. Stereotactic ablative line for multiparametric MRI-guided radiation therapy:
radiotherapy versus standard of care palliative treatment initial results from a Phase II clinical trial in newly diag-
in patients with oligometastatic cancers (SABR-COMET): nosed glioblastoma. Tomography 2019;5:118–26.
a randomised, phase 2, open-label trial. Lancet 2019; [30] Kaanders JHAM, van den Bosch S, Dijkema T, et al. Ad-
393:2051–8. vances in cancer imaging require renewed radiotherapy
[17] Winkel D, Bol GH, Werensteijn-Honingh AM, et al. Target dose and target volume concepts. Radiother Oncol
coverage and dose criteria based evaluation of the first clin- 2020;148:140–2.
ical 1.5T MR-linac SBRT treatments of lymph node oligo- [31] Hall WA, Paulson ES, van der Heide UA, et al. The trans-
metastases compared with conventional CBCT-linac formation of radiation oncology using real-time mag-
treatment. Radiother Oncol 2020;146:118–25. netic resonance guidance: a review. Eur J Cancer 2019;
[18] Palacios MA, Bohoudi O, Bruynzeel AME, et al. Role of 122:42–52.
daily plan adaptation in MR-guided stereotactic ablative [32] Bhujwalla ZM, Kakkad S, Chen Z, et al. Theranostics and
radiation therapy for adrenal metastases. Int J Radiat On- metabolotheranostics for precision medicine in
col 2018;102:426–33. oncology. J Magn Reson 2018;291:141–51.
[19] Henke L, Kashani R, Robinson C, et al. Phase I trial of [33] Raghavan G, Kishan AU, Cao M, et al. Anatomic and
stereotactic MR-guided online adaptive radiation therapy dosimetric changes in patients with head and neck cancer
(SMART) for the treatment of oligometastatic or unre- treated with an integrated MRI-tri-60Co teletherapy de-
sectable primary malignancies of the abdomen. Radio- vice. Br J Radiol 2016;89:20160624.
ther Oncol 2018;126:519–26. [34] Chen AM, Hsu S, Lamb J, et al. MRI-guided radiotherapy
[20] Hansen AT, Poulsen PR, Høyer M, et al. Isotoxic dose for head and neck cancer: initial clinical experience. Clin
prescription level strategies for stereotactic liver Transl Oncol 2018;20:160–8.
38 Lee et al

[35] Bahig H, Yuan Y, Mohamed ASR, et al. Magnetic guided single ablative dose partial breast irradiation.
Resonance-based Response Assessment and Dose Adap- Int J Radiat Oncol 2020;106:821–9.
tation in Human Papilloma Virus Positive Tumors of [49] den Hartogh MD, Philippens ME, van Dam IE, et al. MRI
the Oropharynx treated with Radiotherapy (MR- and CT imaging for preoperative target volume delinea-
ADAPTOR): an R-IDEAL stage 2a-2b/Bayesian phase II tion in breast-conserving therapy. Radiat Oncol 2014;9:
trial. Clin Transl Radiat Oncol 2018;13:19–23. 63.
[36] Thompson RF, Valdes G, Fuller CD, et al. Artificial intel- [50] Acharya S, Fischer-Valuck BW, Mazur TR, et al. Magnetic
ligence in radiation oncology: a specialty-wide disruptive resonance image guided radiation therapy for external
transformation? Radiother Oncol 2018;129:421–6. beam accelerated partial-breast irradiation: evaluation
[37] Meyer P, Noblet V, Mazzara C, et al. Survey on deep of delivered dose and intrafractional cavity motion. Int
learning for radiotherapy. Comput Biol Med 2018;98: J Radiat Oncol 2016;96:785–92.
126–46. [51] Park JM, Shin KH, Kim JI, et al. Air-electron stream inter-
[38] Walker GV, Awan M, Tao R, et al. Prospective random- actions during magnetic resonance IGRT: Skin irradiation
ized double-blind study of atlas-based organ-at-risk outside the treatment field during accelerated partial
autosegmentation-assisted radiation planning in head breast irradiation. Strahlenther Onkol 2018;194:50–9.
and neck cancer. Radiother Oncol 2014;112:321–5. [52] Kennedy WR, Thomas MA, Stanley JA, et al. Single Insti-
[39] Cardenas CE, McCarroll RE, Court LE, et al. Deep tution Phase I/II prospective clinical trial of single frac-
learning algorithm for auto-delineation of high-risk tion high gradient adjuvant partial breast irradiation
oropharyngeal clinical target volumes with built-in dice for hormone sensitive stage 0-i breast cancer. Int J Radiat
similarity coefficient parameter optimization function. Oncol  Biol  Phys 2020;107(2):344–52.
Int J Radiat Oncol 2018;101(2):468–78. [53] Charaghvandi RK, van Asselen B, Philippens ME, et al.
[40] Tong N, Gou S, Yang S, et al. Shape constrained fully Redefining radiotherapy for early-stage breast cancer
convolutional DenseNet with adversarial training for with single dose ablative treatment: a study protocol.
multiorgan segmentation on head and neck CT and BMC Cancer 2017;17:181.
low-field MR images. Med Phys 2019;46:2669–82. [54] Nachbar M, Mönnich D, Boeke S, et al. Partial breast irra-
[41] Whelan TJ, Pignol J-P, Levine MN, et al. Long-term re- diation with the 1.5 T MR-Linac: first patient treatment
sults of hypofractionated radiation therapy for breast and analysis of electron return and stream effects. Radio-
cancer. N Engl J Med 2010;362:513–20. ther Oncol 2020;145:30–5.
[42] Poppe MM, Yehia ZA, Baker C, et al. 5-year update of a [55] Kubota T, Araki F, Ohno T. Impact of inline magnetic
multi institution prospective Phase II hypofractionated fields on dose distributions for VMAT in lung tumor.
post-mastectomy radiation therapy trial. Int J Radiat On- Phys Med 2019;59:100–6.
col 2020;107(4):694–700. [56] Kim A, Lim-Reinders S, McCann C, et al. Magnetic field
[43] Brunt AM, Haviland JS, Wheatley DA, et al. Hypofractio- dose effects on different radiation beam geometries for
nated breast radiotherapy for 1 week versus 3 weeks hypofractionated partial breast irradiation. J Appl Clin
(FAST-Forward): 5-year efficacy and late normal tissue ef- Med Phys 2017;18:62–70.
fects results from a multicentre, non-inferiority, rando- [57] Timmerman R, Paulus R, Galvin J, et al. Stereotactic body
mised, phase 3 trial. Lancet 2020;395(10237):1613–26. radiation therapy for inoperable early stage lung cancer.
[44] Vicini FA, Cecchini RS, White JR, et al. Long-term primary re- JAMA 2010;303:1070–6.
sults of accelerated partial breast irradiation after breast- [58] Henke LE, Olsen JR, Contreras JA, et al. Stereotactic MR-
conserving surgery for early-stage breast cancer: a rando- Guided Online Adaptive Radiation Therapy (SMART) for
mised, phase 3, equivalence trial. Lancet 2019;394:2155–64. ultracentral thorax malignancies: results of a Phase 1
[45] Whelan TJ, Julian JA, Berrang TS, et al. External beam trial. Adv Radiat Oncol 2019;4:201–9.
accelerated partial breast irradiation versus whole breast [59] Finazzi T, Haasbeek CJA, Spoelstra FOB, et al. Clinical
irradiation after breast conserving surgery in women outcomes of stereotactic MR-guided adaptive radiation
with ductal carcinoma in situ and node-negative breast therapy for high-risk lung tumors. Int J Radiat Oncol
cancer (RAPID): a randomised controlled trial. Lancet Biol Phys 2020;107(2):270–8.
2019;394:2165–72. [60] Park JM, Wu HG, Kim HJ, et al. Comparison of treatment
[46] van Heijst TC, Philippens ME, Charaghvandi RK, et al. plans between IMRT with MR-linac and VMAT for lung
Quantification of intra-fraction motion in breast radio- SABR. Radiat Oncol 2019;14:105.
therapy using supine magnetic resonance imaging. Phys [61] Daşu A. Is the a/b value for prostate tumours low
Med Biol 2016;61:1352–70. enough to be safely used in clinical trials? Clin Oncol
[47] Jeon SH, Shin KH, Park SY, et al. Seroma change during 2007;19:289–301.
magnetic resonance imaging-guided partial breast irradi- [62] Alongi F, Rigo M, Figlia V, et al. 1.5 T MR-guided and
ation and its clinical implications. Radiat Oncol 2017; daily adapted SBRT for prostate cancer: feasibility, pre-
12:103. liminary clinical tolerability, quality of life and patient-
[48] Vasmel JE, Charaghvandi RK, Houweling AC, et al. Tu- reported outcomes during treatment. Radiat Oncol
mor response after neoadjuvant magnetic resonance 2020;15:69.
 MRI-Guided Radiation Therapy 39

[63] Bruynzeel AME, Tetar SU, Oei SS, et al. A prospective primary histology suggest implications for clinical out-
single-arm phase 2 study of stereotactic magnetic reso- comes after stereotactic body radiation therapy. Int J Ra-
nance guided adaptive radiation therapy for prostate can- diat Oncol 2016;95:1399–404.
cer: early toxicity results. Int J Radiat Oncol 2019;105: [76] Cercek A, Roxburgh CSD, Strombom P, et al. Adoption
1086–94. of total neoadjuvant therapy for locally advanced rectal
[64] Edmund JM, Nyholm T. A review of substitute CT gener- cancer. JAMA Oncol 2018;4:e180071.
ation for MRI-only radiation therapy. Radiat Oncol [77] Gani C, Boldrini L, Valentini V. Online MR guided radio-
2017;12:28. therapy for rectal cancer. New opportunities. Clin Transl
[65] Rijkmans EC, Nout RA, Rutten IHHM, et al. Improved Radiat Oncol 2019;18:66–7.
survival of patients with cervical cancer treated with
[78] Chand M, Oliva Perez R. MRI linac and how it may
image-guided brachytherapy compared with conven-
potentially lead to more complete response in rectal can-
tional brachytherapy. Gynecol Oncol 2014;135:231–8.
cer. Dis Colon Rectum 2018;61:643–4.
[66] Cree A, O’Donovan A, O’Hanlon S. New horizons in
radiotherapy for older people. Age Ageing 2019;48: [79] Haas RLM, DeLaney TF, O’Sullivan B, et al. Radiotherapy
605–12. for management of extremity soft tissue sarcomas: why,
[67] Visser J, de Boer P, Crama KF, et al. Dosimetric compar- when, and where? Int J Radiat Oncol Biol Phys 2012;
ison of library of plans and online MRI-guided radio- 84:572–80.
therapy of cervical cancer in the presence of [80] Bonvalot S, Gronchi A, Le Pechoux C, et al. STRASS
intrafraction anatomical changes. Radiat Oncol 2019; (EORTC 62092): A phase III randomized study of preop-
14:126. erative radiotherapy plus surgery versus surgery alone for
[68] Asher D, Padgett KR, Llorente RE, et al. Magnetic patients with retroperitoneal sarcoma. J Clin Oncol
resonance-guided external beam radiation and brachy- 2019;37:11001.
therapy for a patient with intact cervical cancer. Cureus [81] Kishan AU, Cao M, Mikaeilian AG, et al. Dosimetric
2018;10:e2577. feasibility of magnetic resonance imaging-guided tri-co-
[69] Boldrini L, Chiloiro G, Pesce A, et al. Hybrid MRI guided balt 60 preoperative intensity modulated radiation ther-
radiotherapy in locally advanced cervical cancer: case apy for soft tissue sarcomas of the extremity. Pr Radiat
report of an innovative personalized therapeutic Oncol 2015;5:350–6.
approach. Clin Transl Radiat Oncol 2020;20:27–9. [82] Henke LE, Green OL, Schiff J, et al. First reported case of
[70] Chuter RW, Whitehurst P, Choudhury A, et al. Technical pediatric radiation treatment with magnetic resonance
note: investigating the impact of field size on patient se- image guided radiation therapy. Adv Radiat Oncol
lection for the 1.5T MR-Linac. Med Phys 2017;44: 2019;4:233–6.
5667–71. [83] Gach HM, Green OL, Cuculich PS, et al. Lessons learned
[71] Nejad-Davarani SP, Kim JP, Du D, et al. Large field of from the first human low-field MRI guided radiation
view distortion assessment in a low-field MR-linac. therapy of the heart in the presence of an implantable
Med Phys 2019;46:2347–55. cardiac defibrillator. Pr Radiat Oncol 2019;9:274–9.
[72] Wojcieszynski AP, Rosenberg SA, Brower JV, et al. Gadox-
[84] van der Heide UA, Frantzen-Steneker M, Astreinidou E,
etate for direct tumor therapy and tracking with real-time
et al. MRI basics for radiation oncologists. Clin Transl Ra-
MRI-guided stereotactic body radiation therapy of the
diat Oncol 2019;18:74–9.
liver. Radiother Oncol 2016;118:416–8.
[73] Rosenberg SA, Henke LE, Shaverdian N, et al. A multi- [85] Green O, Henke LE, Parikh P, et al. Practical implications
institutional experience of MR-guided liver stereotactic of ferromagnetic artifacts in low-field MRI-guided radio-
body radiation therapy. Adv Radiat Oncol 2019;4:142–9. therapy. Cureus 2018;10:e2359.
[74] Luterstein E, Cao M, Lamb JM, et al. Clinical outcomes [86] Noble DJ, Burnet NG. The future of image-guided radio-
using magnetic resonance–guided stereotactic body radi- therapy-is image everything? Br J Radiol 2018;91:
ation therapy in patients with locally advanced cholan- 20170894.
giocarcinoma. Adv Radiat Oncol 2019;5(2):189–95. [87] Bayouth JE, Low DA, Zaidi H. MRI-linac systems will
[75] Ahmed KA, Caudell JJ, El-Haddad G, et al. Radiosensi- replace conventional IGRT systems within 15 years.
tivity differences between liver metastases based on Med Phys 2019;46:3753–6.