A 3D‐conformal technique is better than IMRT or VMAT for lung SBRT

Arguing against the Proposition is Harish K. Malhotra, Ph.D. Dr. Malhotra joined the faculty of Roswell Park Cancer Institute (RPCI) in 2002 and is a Senior Physicist in the Department of Radiation Medicine. He serves as Assistant Professor in the Department of Molecular and Cellular Biophysics in RPCIˈs Graduate Division of the State University of New York. Dr. Malhotra earned his doctorate in physics from the University of Mumbai, India, and completed a post-doctoral fellowship in Medical Physics at Montefiore Medical Center, Albert Einstein College of Medicine, New York. He is certified in Therapeutic Radiologic Physics by the American Board of Radiology and is licensed by the State of New York to practice Therapeutic Medical Physics. Dr. Malhotra is a Section Editor of the Journal of Applied Clinical Medical Physics, has served on the Board of Chancellors of the ACMP and as President of the Upstate New York Chapter of the AAPM. He has published about 30 papers in refereed journals and has been the major advisor to five M.S. students. Recently there has been an increase in use of IMRT and VMAT to treat SBRT of the lung. Comparison treatment planning studies have shown that IMRT and VMAT usually result in more critical structure sparing than 3DCRT.1,2 However, lung SBRT is a complex treatment, involving multiple steps (imaging, planning, delivery, and quality assurance of each step) with multiple layers of uncertainties. Debating the relative merits and drawbacks of different techniques for lung SBRT should take into account every aspect of the process. A good technique not only generates good plans but also delivers them without deviations. 3DCRT has unique advantages over IMRT and VMAT in this regard. First, 3DCRT is not susceptible to the interplay effect between MLC leaf motion and tumor respiratory motion. Similarly, compared to IMRT and VMAT 3DCRT is less affected by patient movement during treatment, MLC positioning errors, and the limited accuracy of MLC modeling in treatment planning systems. 3DCRT therefore is expected to have a better agreement in target coverage between the plan and delivered treatment than IMRT and VMAT, especially when there is large respiratory tumor motion or significant modulation in IMRT or VMAT. Second, 3DCRT allows for cine MV imaging during treatment delivery. These images not only verify tumor position in real-time during beam-on, but also provide information that can be used for determining the true 4D dose delivered to the tumor3 and for developing more effective lung SBRT treatments, such as tumor trailing,4 probability-based treatment planning,5 and adaptive radiation therapy.6 Cine MV imaging is not applicable to IMRT or VMAT since the moving MLC leaves block the majority of the beam view. Furthermore, in our current health care environment where cost is increasingly important, cost-effectiveness needs to be taken into consideration when selecting the technique. 3DCRT is less expensive than IMRT and VMAT. Obviously not all lung SBRT cases are best treated with 3DCRT. Lung tumors vary in location, size, motion, and grade. Some may be more effectively treated using a specific modality than others. Nevertheless, institutional experiences at Duke University Medical Center suggest that the majority of lung SBRT cases (∼70%) can be treated effectively with 3DCRT.7 IMRT and VMAT are used only for carefully selected patients with large tumor size, small tumor motion, or whose dose sparing cannot be achieved with 3DCRT due to the proximity of critical structures. ASTRO recently initiated its national Choosing Wisely campaign to question the use of treatments that are commonly ordered but may not always be appropriate. In echo with this initiative, I conclude my opening statement by recommending that one should not routinely use IMRT or VMAT for lung SBRT without considering 3DCRT. The arrival of 3DCRT into mainstream clinics in the 1980s was an important milestone in the evolution of radiation therapy. Parameters like number of fields and their spatial orientation with respect to tumor and organs at risk, and limited beam modulation options (beam energy, weight, wedges, blocks, etc.), were iteratively adjusted to get a clinically acceptable treatment plan but one which may not have been the optimal plan. The skill and experience of the planner played an important role in such 3DCRT planning. The turning point in the evolution of modern radiation therapy came in the 1990s with the advent of IMRT. The synergy of an inverse optimization engine, better dose computation algorithms, linear accelerators supporting precise dynamic motions, and the ability to target the tumor better with image guided radiation therapy (IGRT), etc., significantly boosted technical abilities. The optimization engine has been much more forgiving (number of beams and their spatial orientation) and searched for global minima, thereby producing consistently better plans. VMAT further improved work flow efficiency by ensuring that the entire treatment could be delivered in 1–2 arcs. Recently, lung SBRT has shown excellent local control.8 In the past, SBRT has most often been delivered using 3DCRT. Though SBRT in general has been well tolerated, Fakiris et al.9 reported 17% grade 3–5 toxicities, while Timmerman et al.8 reported 12.7% grade 3 toxicities. There are also reported cases of rib fractures.10 Many 3DCRT plans have dosimetric minor deviations which could be improved with IMRT/VMAT. In a recent study, IMRT/VMAT plans were shown to consistently outperform their respective 3DCRT plans in almost every dosimetric aspect.11 SBRT for advanced central lung cancers requiring nodal irradiation along with primary disease is also being attempted.12 In our preliminary experience, this requires treating complex geographically separated volumes requiring different simultaneous prescription doses for the primary and nodal beds that can be handled better with IMRT/VMAT. There is concern that in the absence of any synchronization between a moving tumor and dynamic motions inherent in IMRT/VMAT techniques, there may be significant underdosage of portions of the tumor volume. In ten patients, Rao et al.13 found that both VMAT and IMRT plans experienced a negligible MLC interplay effect with a 400 MU/min repetition rate. Similarly, two arcs and ≥2 fractions have been shown to reduce the MLC interplay effect to an apparently clinically insignificant level for FFF beams.1 If respiratory gating is employed or a simple abdominal compression is used to dampen the tumor motion, the results are expected to be even better. Whilst it is possible to generate an acceptable SBRT plan for peripheral lung tumors using 3DCRT, it becomes much more challenging for central lung tumors, including advanced tumors, because of the close proximity of critical structures, which can be better handled consistently with IMRT/VMAT. If low dose to large volumes of normal tissues is not of clinical concern, VMAT can even deliver it quickly making the procedure better tolerable to the patient, with reduced intrafraction motion, and less loss of biological effect due to prolonged fraction delivery.14 Improvement in work flow efficiency and optimal use of departmental resources is the icing on the cake. My opponent argues in favor of IMRT/VMAT mainly from three perspectives: plan quality, planning ability, and work flow efficiency. First, I agree that IMRT/VMAT usually exhibits a dosimetric advantage over 3DCRT, especially when the target volume is geographically complex or is in close proximity to critical structures. In the majority of lung SBRT cases, however, the target volume has a simple geometry and 3DCRT can generate plans that are clinically equivalent to those with IMRT/VMAT. Any slight dosimetric advantages of IMRT/VMAT will be diminished in practice due to various uncertainties. Regarding planning ability, my opponent argues that skill and experience of the planner play an important role in 3DCRT planning, while IMRT/VMAT produces consistently better plans using inverse optimization. Although this is generally true, it should not matter for lung SBRT since the planning is relatively simple due to the small target volume and the simple geometrical relationship between target volume and critical structures. Furthermore, IMRT/VMAT planning has its own challenges, such as limited beam arrangements and couch angles for VMAT in noncoplanar planning, and interplay effects for large tumor motion and the high dose rate for FFF beams. Finally, I want to address the issue of work flow efficiency. IMRT/VMAT requires patient-specific QA, demanding substantial time for preparation, delivery, and documentation. In addition, image verification contributes significantly to the overall treatment time. 3DCRT allows for real-time verification using cine MV imaging during beam on, thus eliminating the need for pre- and post-treatment image verification that are usually required in IMRT/VMAT. Furthermore, the high dose rate with FFF beams (1400 or 2400 MU/min) can improve delivery efficiency for 3DCRT but not much for VMAT, for which treatment delivery time is largely limited by the gantry rotation speed.2 My opponent agrees with me that IMRT/VMAT SBRT plans are dosimetrically superior. He has concerns, however, about their suboptimal delivery and robustness but has not supported this with any literature. Modern linacs are pretty reliable15 and are well modeled in treatment planning systems. Their parameters are monitored every 10–50 ms for compliance with interlocks for out-of-tolerance behavior. Thousands of patients are getting IMRT/VMAT treatments every day, with satisfactory pretreatment QA. SBRT lung plans usually do not have high modulation6 and high dose/fraction allows MLC leaves and gantry to move slowly,13 thereby reducing dose delivery uncertainties. As with other radiotherapy techniques, end-to-end testing provides necessary confidence in the entire SBRT dose delivery chain. Better immobilization, coupled with gating/abdominal-compression, minimizes motion-related concerns. Cine MV-imaging for intrafraction motion is not supported on new flattening-filter-free beams which now have better alternatives which do not degrade throughput (simultaneous kV-imaging, radiofrequency tracking, etc.). Manufacturers usually cap MV-imaging to 100 MU/min, which increases treatment times substantially. Is it ethical to deny patients access to better IMRT/VMAT plans with the justification that cine MV-imaging data may pave the way for a better delivery? That 3DCRT for SBRT is inexpensive is a myth. IGRT requirements for SBRT (Ref. 16) require higher-end linacs which have IMRT and/or VMAT capabilities anyway. Slow 3DCRT delivery represents underutilization of resources, affects throughput and hence possible revenue. If the total cost of treatment delivery is calculated including the time commitment of staff-members (physician, physicist, therapists), 3DCRT might be the costliest. Eventually, everything boils down to clinical results. In a recent study comparing early results for 132 nonsmall-cell lung patients (86-3DCRT, 46-VMAT), the one-year local control rates for VMAT (100%) and 3DCRT (92.5%), favored VMAT (p = 0.03).17 These statistically significant results demonstrate robustness of the IMRT/VMAT dose delivery chain and should allay my opponentˈs concerns. I wholeheartedly support research initiatives but without sacrificing better IMRT/VMAT plans. Our patients deserve the best modality currently available.