MR‐linac is the best modality for lung SBRT

The introduction of MR‐linac hybrid modality started a new era in image guidance for radiotherapy. It provides superior soft tissue contrast compared to conventional x‐ray imaging, and offers the ability to track and gate treatment delivery, yet challenges and limitations remain in various aspects.1 Our previous Parallel Opposed issues discussed the additional physicist qualification and residency requirements for the use of MRI in radiotherapy settings.2, 3 Herein, we continued our discussion on the clinical use of MR‐linac and its potential improvement in treatment efficacy for lung stereotactic body radiotherapy (SBRT). Dr. Andrew Godley believes that “MR‐linac is the best modality for lung SBRT,” while Dr. Dandan Zheng explained her doubts and concerns. 
 
Dr. Andrew Godley received his Ph.D. in High Energy Physics from the University of Sydney in 2001. Dr Godley continued in particle physics at the University of South Carolina and Fermilab, until 2007 when he trained in medical physics at the Medical College of Wisconsin. Dr. Godley spent 7 years as a staff physicist at the Cleveland Clinic before joining the Miami Cancer Institute in 2018 to help develop their MR‐linac and SBRT programs. 
 
Dr. Dandan Zheng received her Ph.D. in Applied Science from the University of California Davis in 2007. After conducting a postdoctoral training in Radiation Oncology at Virginia Commonwealth University, she joined the department as an assistant professor in 2009. In 2012 Dr. Zheng moved to University of Nebraska Medical Center where she is currently an associate professor and the medical physics residency program director.  Dr. Zheng is an avid peer reviewer/associate editor for many journals, and her research interests include image guidance, radiomics and machine learning, stereotactic RT, dose algorithms, and motion management.

This cine imaging incurs no additional radiation exposure. Although some conventional systems acquire planar imaging during treatment, the radiation dose over a large area would be excessive to achieve the continuous monitoring of the MR-linac. Second, while using external surrogates facilitates continuous monitoring, these surrogates require at least an additional 1 mm margin to ensure correlation with the tumor. 5 Third, the precise gating window of the MRlinac provided by its continuous tumor tracking limits or eradicates multileaf collimator (MLC) interplay issues, which was reported to affect delivery accuracy of other modalities 6 and thus reduce target coverage. 7 Lastly, the tumor based gating eliminates the need to define an internal target volume (ITV), which is conventionally used to ensure tumor coverage during the breathing cycle, but increases the volume of normal lung irradiated. Further, traditional ITVs are just a snapshot of tumor motion observed over tens of seconds in a 4D CT. When compared to a treatment length cine-MRI based ITV, the four-dimensional computed tomography (4D CT) based ITV typically captures only 45%-90% of the required volume, 8 leading to potential under dosing. The MR-linac continuous tracking accounts for the breathing motion including large excursions of the tumor, so that this motion does not have to be forcibly limited.
With the above-mentioned target monitoring and tracking, immobilization for MR-linacs does not then feature compression or breath control, leading to higher patient comfort during treatment. Due to the continuous imaging of the tumor, the patient can be coached by the therapists during treatment, or even see and respond to their own cine-MRI, in order to achieve optimal breath holds. By consistently placing the tumor in the gating window, the gating duty cycle

2.B | Dr. Dandan Zheng
Since its clinical introduction, the MR-linac has quite often attracted the spotlight as the state-of-the-art treatment modality in the era of image-guided adaptive radiotherapy. MRI-guidance may provide the much needed breakthrough for disease sites such as pancreas, where other radiotherapy treatment modalities lack efficacy due to doselimiting normal tissue toxicity, poor on-board x-ray contrast, and weak internal motion-external surrogate correlation, etc. When it comes to the titled statement for lung SBRT, however, I cannot agree that the MR-linac represents the best modality for reasons elaborated below.

2.B.1 | Technical limitations of the linac component
Integrating MRI and linac poses tremendous technical challenges on both systems. [14][15][16][17][18] While the integration has finally been made feasible after decades of R&Ds, the linac component in an MR-linac is still not up to the highest standard of modern linacs. The essential technologies pertaining to treatment time reduction for lung SBRT, such as volumetric modulated arc therapy (VMAT) and high dose rates (up to 2400 MU/min), 19,20 are not available on the MR-linac.
Currently, the MR-linac is only capable of step-and-shoot, low dose rate (450 MU/min for Elekta Unity or 600 MU/min for ViewRay MRIdian), low gantry speed, and low MLC speed, all of which substantially lengthen the treatment duration. This prolonged treatment time poses contraindication for patients who cannot tolerate due to back pain or poor performance status, and increases treatment variations and motion uncertainty. In addition, the MR-linac has a limited lateral range of isocenter and forces off-axis geometry for those tumors located outside the range. This can potentially lead to excessive monitor units and hence high integral dose to normal tissue. 21 The couch correction range is also limited with no option of couch rotation and, in some models, lateral translation. Furthermore, the electron return effect is much more pronounced in lung and can lead to considerable dose distortion at tissue-air interfaces. Some early studies have shown increased dose in lung and skin, as well as compromises to the accuracy of both patient plan and QA dosimetry. 11,22,23

2.B.2 | Technical limitations of the MR component
The most desirable features to those early adopters might be the real-time internal imaging capability with an MR scanner. However, cautions should be taken as they are not standard diagnostic MR scanners either. First, the MR-linac often uses lower magnetic field strengths than diagnostic MRIs (e.g., ViewRay MRIdian uses a 0.35 T scanner), therefore rendering compromised MR image quality. Second, the limited real-time tracking range excludes those cases with very peripheral lesions. Third, real-time target monitoring is based on 2D MR cine images, therefore omitting the volumetric or the thirddimension motion information. Fourth, the MR sequences typically used for lung are prone to irregular breathing artifacts and banding artifacts. 24 Fifth, in the desirable adaptive radiotherapy application, MRI cannot provide the CT-number information for heterogeneity correction which is especially important for lung SBRT dose calculation. Any method to address this fundamental limitation introduces additional uncertainties which are more pronounced for lung because lung and bone feature little MR signal. 25 Last but not least, tissue heterogeneity in lung can induce pronounced geometric distortions in MR images due to local susceptibility differences, and these errors are in addition to other MR distortions such as those due to main magnetic field and gradient field nonlinearity. 26

2.B.3 | Incompatible patient populations
For the patient population with any contraindication against MRI, the MR-linac is not an option. These include patients with claustrophobia and those with ferromagnetic implants such as prostheses, cardiac pacemakers and defibrillators, artificial heart valves, and cochlea implants. In addition, due to the bore size limitation, the MR-linac also excludes large patients and prevents the use of many immobilization devices and body positions.

2.B.4 | Alternative modalities
Currently, there are many successful alternative modalities and technologies for lung SBRT that the MR-linac cannot rival. As mentioned previously, standard linacs are capable of delivering highly conformal lung SBRT plans at very high speeds with VMAT and high dose rates. 19  Whereas the excess cost may not be justified for lung SBRT where the standard treatment modalities are much more cost-effective and highly successful with >90% local control rates and low toxicity to normal tissues, along with adequate image-guidance and tracking. 27,28 In conclusion, MR-linac is clearly not a viable lung SBRT treat-  In summary, the only compelling advantage of conventional linacs is their delivery speed, but without continuous tumor tracking, this just allows you to miss the target quicker. Hence I maintain that MR-linac is the best modality for lung SBRT. More importantly, while the MR-linac should be further advanced as a frontrunner modality toward that vision for diseases like pancreatic cancer, it is not the case for lung SBRT. For pancreas, current radiotherapy provides relatively poor efficacy with organ toxicity partly due to low soft-tissue imaging contrast, while MRI may offer the important soft-tissue contrast resolution that is superior to the competing x-ray technologies. Whereas for lung, contrast is not as challenging, and current SBRT can already achieve 98% tumor control as cited by Dr. Godley. 4 As discussed above and will be stressed again below, numerous existing technical issues in MR-linacs make it far from being the ultimate solution toward that vision. But even hypothetically assuming all issues would have been perfectly resolved someday, arguing for the MR-linac as the best modality for lung SBRT would still be like arguing for proton therapy for prostate: Are there enough clinically meaningful gains to justify the higher cost?
Let's revisit MR-linacs' three main advantages laid out by Dr. Godley.

3.B.1 | Comfort
With VMAT, high dose rates and fast collimators, lung SBRT treat- requirements: motion prediction to address latency, repositioning of the beam, and dose adaptation for changing anatomy. With the presence of the magnetic field, repositioning of the beam will be particularly challenging. In contrast, CyberKnife TM has already developed a working real-time tracking solution using its robotic linac and Synchrony TM system and had years of documented clinical success. 27 Meanwhile, standard linacs also have developed solutions such as MLC-tracking and couch-tracking, 35,36 but these solutions would be difficult to adopt on MR-linacs due to the interfering magnetic field.
The so-called "MR-tracking" is not really "tracking" but gating, which substantially lengthens the treatment duration and has fallen out of fashion for lung SBRT because of it. In fact, the MR-gating may actually be managing the extra motion uncertainty created by the lengthened treatment but does not necessarily improve the overall accuracy.
The much larger (compared with the conventional ITV) MR-ITV created with treatment-length cine MRIs that Dr. Godley described earlier might serve as a proof to this. 8 Since the current high tumor control rate argues against the tumor underdosing using the conventional ITV that Dr. Godley cautioned about, I would offer another more likely reason for the larger MRI-ITV volumes: an effect of the increased motion uncertainty due to the much longer "treatment length"!

3.B.3 | Tumor monitoring
Although incapable of real-time tumor targeting, the expensive MRlinac does provide the desirable real-time tumor monitoring. However, for lungs, such monitoring capability is not unique to MRI due to high x-ray imaging contrast and minimally invasive fiducial implantation procedures. Continuous x-ray monitoring based on tumors or fiducials has long been established and achieved high precision for lung. Significantly shortened treatment delivery and SBRT hypo-fractionation have further alleviated concerns about additional radiation exposure. Meanwhile, exposure-free radiofrequency tracking has been FDA-approved for lung and can be integrated with real-time MLC-tracking. 35 In addition to these mature technologies, the feasibility of other real-time lung-tumor monitoring has also been demonstrated with radiation-dose-free Compton-scatter imaging using the therapy beam 37 and with PET emission guidance. 38 In summary, MR-linacs should be better used where they are needed. For lung SBRT, competing technologies are highly successful, and offer higher cost-effectiveness and better solutions.