Image‐guided radiotherapy is being overvalued as a clinical tool in radiation oncology

Arguing against the Proposition is David A. Jaffray, Ph.D. Dr. Jaffray graduated in 1988 with a B.Sc. in Physics from the University of Alberta and completed his Ph.D. in Medical Biophysics from the University of Western Ontario in 1994. The following eight years were spent as a Clinical Physicist in the Department of Radiation Oncology at the William Beaumont Hospital, Royal Oak, MI. He is presently Head of Radiation Physics at the Princess Margaret Hospital and Associate Professor in the Department of Radiation Oncology, University of Toronto, where he holds the Fidani Chair in Radiation Oncology and Medical Biophysics. His principal research interests involve the development of novel imaging systems for radiation therapy, including amorphous-silicon-based large-area detectors and cone-beam CT, image-guided radiotherapy, and integration of functional imaging into radiotherapy. Dr. Jaffray is certified in Radiation Oncology Physics by the ABMP, currently serves as a member of the AAPM Science Council, and is a frequent contributor to Medical Physics. That image-guided radiotherapy (IGRT) will improve radiotherapy is taken as axiomatic. Improved patient setup accuracy reduces probability of geographic miss and improves treatment outcome. But where are the data demonstrating that IGRT improves clinical outcome in a clinically measurable way? If treatment margins (PTV minus CTV) are already larger than setup uncertainties, and if such margins result in acceptable toxicity, what is gained with IGRT? Will IGRT enable reduction of treatment margins? If so, will this ever be testable in a controlled clinical trial? Given the uncertainties in guesstimating CTVs from GTVs, perhaps reducing field margins (naively assuming that IGRT reduces errors) will be detrimental to treatment outcome? Conversely, if one cannot reduce treatment margins what benefits are derived from IGRT? Ling et al.1 have suggested that an IGRT system should have three-dimensional (3D) imaging of soft tissues and tumors, plus an efficacious process for clinically meaningful intervention. Currently IGRT fails on both counts. Image quality, degraded by the physics of kV cone-beam scattering, is inferior to conventional CT; significant breathing artifacts result from the scan times. Marginal image quality coupled with imperfect and nonrobust image registration software both suggest questionable application of IGRT as currently practiced. Is using IGRT to move a patient around daily by a few millimeters, in ignorance of resulting clinical outcome, better than doing nothing? Current IGRT systems also do a poor job of monitoring intrafractional errors. Unless these are less than interfractional errors, using IGRT prior to but not during treatment accomplishes little. Again, where are the data? Even assuming that IGRT can detect setup errors, how should one correct for them? Is rigid-body, bony anatomy registration of IGRT and CT-Sim images sufficient, or must one consider soft tissue deformation? If the latter, then current IGRT systems are deficient. If the former, then aren't existing MV electronic portal imaging systems adequate? When is kV and∕or 3D required, and when is two-dimensional (2D) MV sufficient? Consider also the cost∕benefit ratio: 3D-kV imaging systems cost $500,000, which is nearly the cost of a linac. If one treats five patients per hour and IGRT adds three minutes this represents a 25% increase. Maintenance, physics, and physician times also increase. Is IGRT worth the price? Lamentably, its benefits will likely never be tested in a clinical trial as it is already widely used without controls or forethought. I'm not arguing that IGRT is a boondoggle. Indeed, it's counterintuitive to argue against improved accuracy. But like many technological advances IGRT has come too fast, with too little thought, and without serious discussion of which types of patients will benefit. American hospitals have already collectively spent well over $100,000,000 on IGRT without a single peer-reviewed randomized clinical trial demonstrating its benefits. This is another example of new technology searching for a problem before the need has been demonstrated. Lacking data, we've placed high expectations on an expensive, untested technology. As we should have learned from Record and Verify systems, computerized patient setup doesn't eliminate errors, it merely changes their nature. I question not IGRT, but rather its current use. To quote Walt Kelly: “We have met the enemy and they are us.” Radiation therapy is a proven therapeutic agent directed at localized targets with the explicit objectives of achieving control of the neoplasm while minimizing the risk of toxicity in surrounding normal tissues. The relationship between geometric miss and failure in outcome is one of the few relationships in radiotherapy that is not debated. The very fact that we are willing to irradiate perfectly normal tissues to ensure tumor coverage highlights the importance of succeeding in the accurate localization of dose and target. Current practice of radiation therapy devotes substantial clinical and technical effort towards this objective. The introduction of CT simulation, 3D treatment planning, portal imaging, and verification of treatment parameters are existing infrastructure investments. Ongoing operating costs are also substantial. These include daily manipulation of skin marks, checking for SSD discrepancies, applying small shifts based upon portal imaging, and the time consumed in the acquisition of port films and electronic portal images. We should also include time and effort expended in the handling of exceptions by all of the disciplines involved in the management of difficult or complex cases, so as not to have to withhold therapy for some patients due to our inability to solve the technical challenges related to targeting. The recent introduction of IMRT has heightened the concerns around geometric uncertainties as the opportunity to create highly conformal dose distributions places additional pressure on executing the delivery with a known geometric performance.2 While these advances have the best of intentions, it is remarkable to note how little quantitative validation is performed for each patient to prove that a course of conformal therapy has been executed as intended over the many fractions of radiation delivery. I would assert that there is a growing inconsistency in the amount of effort placed on the design of therapy compared with the assurance of its execution. In fact, the investments in further refinements in therapy that are occurring upstream should be questioned compared with the clear and unanswered challenge of placing radiation dose with geometric accuracy and precision within the human body. IMRT and biological targeting could be criticized as “tails wagging the dog,” however, we don't dictate the order of innovation. IGRT technologies that provide volumetric imaging of both the targeted structures and surrounding normal tissues provide the opportunity to move from an intention to deliver state-of-the-art radiation therapy, to a patient-specific verification that the intention has been satisfied. There is a need to separate the issues of IGRT technology adoption from the more fundamental underlying question: “What level of quality should be achieved in the delivery of radiation therapy to the individual patient?” Simple population-based performance metrics ring hollow. Clearly, PTV margins designed to cover the CTV in 95% of the patients would not go over well with one patient in twenty. Geometric uncertainty in the localization of therapy is a major controllable factor influencing the performance of radiation treatments, and it has been poorly addressed for too long. Recent studies based upon portal imaging and repeat CTs demonstrate the presence of patient subpopulations that are not well served by current positioning methods.3–5 It is clear that the first step toward confirming the quality for each patient is to be able to measure the geometric performance of the delivery. This requires imaging of internal structures at the time of treatment, i.e., image-guided radiation therapy. There are three ingredients to good radiotherapy: correct tumor identification, conformal treatment planning, and precise treatment delivery. With the advent of biological imaging (MR, PET, etc.) we're only now learning how to do the first part correctly. We've very nearly perfected the middle step, and as Dr. Jaffray points out, we're still relatively primitive with regard to the third step. But his arguments that IGRT will improve that third step by a clinically significant, or even measurable amount fall short of the mark. While I agree with his argument that IGRT has the potential to improve the situation and should be tested, he makes no compelling argument to justify hundreds of hospitals buying IGRT at this time to the tune of $100,000,000. For example, he quotes his own studies,2,5 both of which use conventional CT rather than linac-based cone-beam CT systems, simply to show that adaptive radiotherapy based on multiple CT scans would allow delivery of slightly higher doses to prostate (on average, 7.5% more dose in Ref. 2 and 13% in Ref. 5) without additional rectal toxicity. These are very idealized studies, however, that assume no intrafraction motion and no organ deformation. They also ignore the fact that linac-based IGRT systems have inferior image quality compared with conventional CT, thus rendering them less accurate for identifying soft tissues, positioning patients, or monitoring intrafraction motion. Furthermore, the prostate is a bad example to use to try to demonstrate the utility of IGRT. Zelefsky et al.,6 for example, have already reported an eight-year actuarial PSA relapse-free survival rate for favorable-risk prostate patients treated with IMRT to of with no grade 4 rectal toxicity and only 0.1% grade 3 toxicity, without the use of IGRT. Thus, even if IGRT could buy you a 13% dose increment, it would likely be of little clinical value. In fact, the benefits cannot be as large as Dr. Jaffray claims since he has ignored organ deformation and intrafraction motion in his analysis. His other references (Refs. 3 and 4) are little more than studies demonstrating that conventional 2D EPID improves the setup accuracy of patients—not really news and not really a justification for a $500,000 3D IGRT system. In diseases where radiotherapy does poorly, IGRT offers even less hope. In brain, for example, IGRT offers little improvement over conventional stereotactic radiosurgery. In lung and upper abdomen, IGRT will be almost useless unless it is respiratory gated, and current systems can't do that. In head and neck, 2D EPIDs seem to work fine. In short, Dr. Jaffray offers a hypothesis but no real clinical facts, and certainly no data justifying IGRT's price tag or the enormous amount of hype currently associated with it. The advent of IGRT technologies has provided interesting data to awaken the community to, and refocus our attention on, the targeting problem—something that we have neglected for too long. The recent article by de Crevoisier et al.7 highlights the influence of systematic errors on biochemical measures of outcome in radiation therapy of the prostate. This article describes a retrospective study in which a correlation between rectal-filling-induced displacement of the prostate at the time of planning is correlated with reduced biochemical control. I compliment the authors for exploring this relationship and bringing their observations to the forefront. Daily image-guidance data about the day-to-day positioning of the patient will elucidate the role that important factors such as bowel filling,7,8 respiratory-induced motion,9 lung target deformation,10 and radiation-induced changes in the geometry of normal and target tissues11,12 have in limiting the precision of radiation delivery. At this point I have not advocated for margin reduction or dose escalation as a benefit of IGRT. The arguments to date have related to quality assurance of current radiation therapy practice and how the introduction of IGRT technologies allow us to recognize the elephant in the room (“geometrical miss”) for what it is—a blind spot that limits outcomes and prevents understanding of the effects of radiation therapy. The role of image guidance as a quality assurance and education tool cannot be overstated. So what about the future? Now that we have our dose in a pile, where does IGRT let us go? I believe that the opportunities are enormous. As there is no known benefit to unnecessary irradiation of normal tissues, IGRT will permit margin reduction, hopefully applied with a strong understanding of the principles of image guidance and residual uncertainties. The frequent argument used against PTV margin reduction is that standard clinical practice is “hiding” disease in the PTV margins. This simply reflects the need for more education on this topic—the reality is that we are “hiding” disease that may not be there. The PTV is a geometric construct to assure dose to the clinical target volume, not a volume known to contain any specific tissue. Clearly, we still have some work to do in educating the community on these concepts. One side effect of applying radiation therapy without confidence in dose placement is to maintain a strict constraint on the degree of dose uniformity within the target. The emergence of stereotactic radiotherapy protocols that permit hot spots provided “they are in the target” is a harbinger of the importance of IGRT in moving away from the dogmatic and poorly supported practice of delivering uniform dose to targets in radiation therapy.13 Failure to walk away from this practice within the next decade may truly limit innovation in radiation therapy. The constraint of uniformity forces undue hardship on normal structures and will limit dose escalation for those sites with poor control. The radiation therapy community needs to remember that we are not standing by ourselves in the field of oncology. The remarkable explosion in therapies grouped under the umbrella of interventional radiology and targeted therapeutics is clearly exploiting advanced technologies such as image guidance to maximize the benefit detected in outcome studies.14 The accelerated introduction of high-intensity focused ultrasound is made possible with the integration of MR-guided placement and thermometry. It is my belief that radiation therapy wouldn't get out of the FDA gate without image guidance if it were to be proposed anew today. Failure to advance our skill in the application of radiation therapy will leave this proven therapy at the back of the pack and will limit our ability to understand how this agent actually works in vivo. Fundamentally, the adoption of IGRT is about quality in health care and our willingness to assure high-quality delivery for every patient. I would argue that 100% of the patients should achieve the best quality of radiation therapy if it is financially achievable. To bring this about we need to drive industry to provide these tools such that they operate in the current cost envelope. Early adoption is one way of achieving this objective. The answer to the question as to whether IGRT is overvalued is obviously no. The real question is how we are going to innovate our practice to make sure that it isn't overpriced.

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.