Over the next decade the success of radiation treatment planning will be judged by the immediate biological response of tumor cells rather than by surrogate measures such as dose maximization and uniformity

Arguing against the Proposition is X. Allen Li, Ph.D. Dr. Li is an Associate Professor and the Chief Physicist in the Department of Radiation Oncology, Medical College of Wisconsin. He has been a board certified medical physicist since 1994, and is currently a member of AAPM TG-74 and TG-106. Dr. Li is an Associate Editor for Medical Physics and has published 82 peer-reviewed papers. His research interests include biologically-based treatment planning and image-guided radiation therapy. Deficiencies in current radiotherapy planning, directly linked to the history of radiotherapy, are due to the paucity and/or inadequate understanding of biological and clinical data needed for optimum treatment design. Treatment planning was nonexistent or primitive prior to the application of computers in the 1960–70s, and after that, primarily by “convention” such as parallel opposed fields, four-field box or arc-therapy. As to biological and clinical issues, little was known about tumor or organ dose response, dose-volume relationships, and the heterogeneities of tumor burdens and of radiosensitivity within the tumors. In the absence of meaningful criteria, dose-based surrogates were used by default. Since the 1980s, 3-dimensional conformal radiotherapy (3D-CRT) technology has made available complete dose-volume information which, when combined with clinical outcome data, has permitted model-based analyses of tumor local control and organ toxicity. These analyses clearly reveal that dose-surrogates by themselves are insufficient to predict the consequences of a particular treatment plan. For example, the tumor control probability of prostate cancers depends on the risk groups in which the patients belong,1 lung toxicities are more likely in the lower than in the upper lobe,2 and rectal bleeding may depend on functional reserve.3 All biophysical models that attempt to describe clinical outcomes contain biological information, be it SF2 (cellular surviving fraction for a dose of 2 Gy) in the Equivalent-Uniform Dose (EUD) model,4 or the architecture of an organ's functional subunits.5 The success of the human genome project and developments in imaging strongly suggest that biological data will be increasingly available for radiotherapy planning in the future. The influence of various gene products on the radiosensitivity and chemosensitivity of tumors and normal tissues is beginning to be elucidated, leading to the suggestion that patient-specific molecular profiling will provide information for the selection of treatment modality and dose-intensity.6 Understanding of the signal transduction pathway of cellular radiation response has paved the way for combining molecular and radiation therapy with very positive results in clinical trials.7 The integration of advanced imaging techniques and knowledge of the molecular basis of cancer will be a powerful influence on cancer management in general and radiotherapy planning in particular. With this approach, one can perform radiobiological phenotyping to characterize the tumor in terms of the factors that influence its radioresponse. A good example is the use of PET for visualizing tumor hypoxia, which is known to be detrimental to the efficacy of radiation treatment in several disease sites.8 In this case, the biological response of radiation could be re-oxygenation that could be monitored by sequential PET imaging. Another promising assay of biological response, based on an increased understanding of the mechanisms of cell death, involves the use of radio-labeled Annexin V which adheres to the cellular membrane of apoptotic cells.9 Lastly, the use of various PET tracers (with FDG being the current favorite) for monitoring treatment response holds significant promise.10 Methods based on magnetic resonance are also being intensively evaluated as tools for planning and assessing radiation treatment.11 It is clear that, given the complexity of biological systems, dose-based surrogates are inadequate for describing the radiation responses of tumors and normal tissues. It is also certain that molecular profiling and imaging will provide the approach for improving treatment design and monitoring treatment response. What may be uncertain is the time course for attaining these goals. In that regard, we physicists are very familiar with error and uncertainties estimates—but whether it be a decade or two, the time is now to embrace the brave new world of biology-based treatment planning. To cling to dose-based treatment planning, given the recent findings referenced in my opening statement, is to keep “looking for the lost key under the light.” We know that using purely physical surrogates for biological endpoints is fundamentally wrong, and that dose was originally chosen as the basis for treatment planning simply because at the time it was the only thing we could quantify. But should one continue to hold to a system known to be fundamentally flawed because the scientifically correct method is as yet imperfect? Shouldn't one start looking for the key where it was lost by bringing light to that place? Clearly, radiotherapy physicists must begin learning the pertinent aspects of tumor and radiation biology to better understand the deficiencies of dose-based treatment planning vis-à-vis the potentials (and pitfalls) of biology-based treatment planning. For example, one should know that the assumption that “clonogenic-cell density and radiosensitivity distribution are constant throughout the planning target volume” is wrong, and therefore the conclusion that “a uniform dose distribution maximizes tumor control” of Dr. Li's third reference is incorrect as well. Radiotherapy planning using biological information may or may not be perfected within the next ten years, but even if not perfected it likely will surpass dose-based methods in predicting clinical outcome. It's only in the last decade or so that analysis of dose-volume histograms has led to a substantial increase in biological understanding of normal tissue complications, and that the potentials of molecular profiling and imaging have emerged relative to the radioresponse of tumors. The pace of these developments can only accelerate in providing the needed information for optimizing radiation planning. It may be a decade or even two before the goals of treatment based on biological end-points is realized, but the time is now for radiotherapy physicists to take their heads out of the warm and fuzzy sand of dose-based treatment planning, and move towards biologically-based treatment planning. The generation of treatment plans optimized to maximize tumor-cell killing and minimize normal-tissue damage (biologically-based radiation treatment planning, BBRTP), rather than to deliver a uniform dose to a specific treatment volume (physically-based RTP, PBRTP), has been proposed for many decades. With the recent surge in intensity-modulated radiotherapy and advances in biological/molecular imaging, interest in BBRTP is increasing within the radiotherapy community. I believe that BBRTP will eventually replace PBRTP. However, the central question of this Point/Counterpoint is “Will BBRTP be reliably practiced in the clinic over the next decade?” Despite many decades of effort and experience, knowledge of tumor and normal-tissue radiation responses, on which BBRTP depends, is meager, unreliable, and not well documented. Unlike PBRTP, biological models are essential to any BBRTP effort. Although there are several phenomenologic and mechanistic models available, their predicting powers are generally poor because of large uncertainties in the biological parameters. Efforts to develop such models have increased in recent years, but progress has been slow, particularly for normal structures. Many sources of uncertainty (e.g., treatment positioning, organ motion, dosimetry, the cell micro-environment, irradiation history, and chemotherapy) contribute to current deficits in information about normal tissue complications.12 In addition, unified tools to accurately document such information remain to be developed. Although in-vitro data on radiation responses are abundant, they can be applied in the clinic only with a great deal of caution.13 BBRTP requires patient-specific information on tumor/normal-tissue biology. Biological/molecular imaging technologies, although they offer great promise in eventually providing such information and even real-time monitoring of biological response, are far from mature. Moreover, tumor and normal-tissue radiosensitivities most likely change during a course of radiotherapy. The models, methods and technology needed to account for these phenomena remain to be developed. Because of uncertainties in tumor/normal-tissue biology, clinical use of BBRTP is risky for the foreseeable future. For example, the use of incomplete or inaccurate biological information to preferentially target selected regions of a tumor at the expense of other tumor regions may jeopardize the treatment outcome. It may be possible, however, to use a combined PBRTP/BBRTP approach. First, a treatment plan is generated using PBRTP. Then, BBRTP is used to selectively target regions of the tumor that are potentially radioresistant without decreasing the dose to any other region of the tumor. It is known that a uniform dose distribution maximizes tumor control so long as the clonogenic-cell density and radiosensitivity distribution are constant throughout the planning target volume.14 If the combined plan can be accomplished without appreciably increasing the dose to critical structures, the treatment outcome should be improved. As we gain additional information about the efficacy of BBRTP, more aggressive strategies may evolve. Because clinical trials require many years to complete, the replacement of PBRTP with BBRTP as the standard of practice is certainly unlikely during the next decade. I would like to acknowledge Dr. Robert Stewart for providing comments. As expected, Dr. Ling did a superb job explaining why dose-based surrogates are inadequate for treatment planning. I agree with Dr. Ling that there is a compelling need for BBRTP. However, the central issue we are debating is whether or not BBRTP will make a significant impact in the clinic within the next decade. On this issue, our opinions diverge. A major issue is that the models used to relate dose distributions to treatment outcomes are highly nonlinear, and seemingly small perturbations in physical or biological factors can have a large impact on estimates of biologic response, which is often organ/tumor and patient specific.13,15–17 Dr. Ling suggests that advances related to the human genome project and molecular imaging will provide the patient-specific information needed for BBRTP. There are many hurdles that must be overcome before functional/molecular imaging will have a substantial impact on the routine practice of radiation oncology, including imaging-pathology validation, timing of post-therapy imaging, spatial and temporal evolution of tumors, and lack of clinical outcome studies.18 For example, it is not clear whether PET or other imaging modalities can accurately depict the microscopic extension of a tumor. The spatial and temporal variation of hypoxia during radiation therapy poses a significant challenge for the use of PET or other imaging modalities. Prospective multi-institutional clinical trials to evaluate the clinical impact of biologically-based planning strategies and novel imaging techniques are almost nonexistent. Until these issues are addressed, BBRTP will need to be applied in an incremental fashion. It may be many years or even decades before BBRTP has demonstrated clinical benefits over PBRTP. Nevertheless, I agree with Dr. Ling that we should embrace the brave new world of BBRTP and actively pursue methods to speed the clinical implementation of these techniques.