• © 2004 by American Society of Clinical Oncology

Phase I's, Photons, and Philosophies: New Tactics for Exploratory Clinical Trials of Concurrent Chemoradiation

It is unusual for this journal to publish an editorial about a phase I protocol—particularly a phase I protocol that does not involve any elaborate new targeted agents. However, the article by Muler et al [1] introduces several innovative concepts in phase I clinical oncology trials that deserve some comments.

The most obvious feature of the Michigan study is the use of a novel trial methodology called time-to-event continual reassessment methodology (TITE-CRM). A detailed discussion of this design is beyond the scope of this review, but one of its major goals is to accelerate the determination of the maximum tolerated dose (MTD) of a new regimen. If properly performed, it improves the confidence of the MTD calculation when compared with conventional phase I design (3/6 design) and minimizes the number of patients subjected to doses that are likely to be subtherapeutic. This design is particularly well suited to radiotherapy trials, given that dose-limiting toxicity may occur after (sometimes long after) the treatment is finished [2]. Conventional 3/6 radiotherapy phase I trials require lengthy periods of trial suspension, which has been problematic to the Radiation Therapy Oncology Group and others. The TITE-CRM design can speed up the phase I process, avoiding such delays. This is a positive step. However, this design could also result in a higher number of patients subjected to doses above MTD, particularly if there are delays in communicating toxicity data to the medical monitoring team. Therefore, does the TITE-CRM create an ethical dilemma in phase I clinical research?

As stated in the article by Muler et al, this study was approved in its entirety by the University of Michigan institutional review board, and we must thus presume that its members closely scrutinized the TITE-CRM design and compared it with a conventional phase I design. I would concur that this particular study was designed and executed according to the Helsinki Declaration and other guidelines of human research. However, this should not be taken to mean that the TITE-CRM design should be used in all settings and without concern. Protocol features that are even more important in a TITE-CRM phase I than a conventional phase I design are as follows: (1) exceptionally strong background data to support the choice of doses; (2) a disease setting in which dose-intensity of multiple therapeutic agents is likely necessary to improve tumor control or palliation; (3) dose escalation increments that are relatively conservative; (4) an informed consent form that clearly describes the risks of a phase I study; (5) rapid, nearly real-time reporting and analysis of toxicities by physician-investigators and clinical research and nurse associates; (6) highly efficient medical monitoring, analogous to Data Monitoring Committees used in phase III studies; (7) scrutiny, approval, and frequent reapproval of the study by a very experienced academic institutional review board that often reviews early phase clinical trials; and (8) careful peer review and commentary by the journal(s) that ultimately referee this type of research.

When these criteria can be met, as seems to have been done in the study by Muler et al, the TITE-CRM design does appear to be an improvement over classical 3/6 phase I design, at least for concurrent chemoradiotherapy studies. This methodology deserves testing, not just in the tense relationship between radiotherapy and gemcitabine, nor just in pancreatic cancer. It is very reasonable to explore this design for concurrent chemoradiotherapy trials in lung cancer, malignant brain tumors, and other inoperable malignancies, and the Michigan researchers should be congratulated for introducing this to the radiation oncology research community.

The other major clinical difference between the Michigan study and most phase I chemoradiotherapy studies is the selection of radiotherapy dose. Traditionally, phase I combined modality studies use a fixed radiotherapy dose that is similar, if not identical, to radiotherapy-alone dosing. The dose of the concurrent drug(s) is then escalated in a classic phase I design until MTD of the drug is established. Unfortunately, with this design, the chemotherapy dose-intensity achieved is usually well below that which is standard when the same drug(s) is/are given alone, particularly in gemcitabine-radiotherapy studies [3]. Attempts to combine radiotherapy with concurrent gemcitabine have frequently become phase I dose de-escalation studies [4,5].

The Michigan investigators have taken a contrary and controversial approach. In a previous study, they retained maximal dose-intensity of this moderately active and clinically radiosensitizing [6] drug, gemcitabine, and determined the MTD of concurrent radiotherapy [7]. Not surprisingly, the MTD of radiotherapy was only 36 Gy in 2.4 Gy daily fractions [7]. Radiobiologically, this is about the equivalent of 40 to 45 Gy in standard (1.8 to 2 Gy) fractionation. These slightly large and accelerated fractions appear to be feasible (with conformal radiotherapy), and this improves cost and convenience—important points in the treatment of a disease with median survival of less than 1 year. However, 36 Gy is still only 36 Gy, a dose well suited to Hodgkin's disease but not generally associated with durable local control for epithelial neoplasms. The Michigan approach thus represents a compromise between two philosophies currently being tested in a randomized trial by the Eastern Cooperative Oncology Group: the usual 50+ Gy plus low-dose gemcitabine compared to standard doses of gemcitabine with no radiotherapy (0 Gy) [8]. Is 36 Gy an appropriate compromise, or is it the radiation oncologist's version of settling a divorce case by tearing the grand house into two mini-domiciles?

The approach of radiation de-escalation in the setting of concurrent or adjuvant chemotherapy is not entirely novel; it has been used with success in pediatric malignancies and lymphomas for several decades. This has allowed for significant improvement in cure rates, while minimizing some late effects of radiotherapy. Some research has also been done in adult gastrointestinal cancers. The Radiation Therapy Oncology Group showed that 50.4 Gy with concurrent chemotherapy was superior in survival and local control compared with radiotherapy alone at a higher dose (64.8 Gy) [9] Dose escalation to 64.8 Gy with the same concurrent chemotherapy did not improve the outcome [10]. In pancreatic carcinoma, the venerable Gastrointestinal Tumor Study Group study did not show an advantage to 60 Gy over 40 Gy radiotherapy with split-course radiation, in which both treatment groups also received concurrent bolus fluorouracil [11].

The Michigan experience, plus other supportive data, raises several interesting points about the treatment of locally advanced pancreatic carcinoma that may apply to other cancers: (1) There is probably little to gain from treatment of elective nodes remote from the pancreas [1,7,12]. (2) High-dose, concurrent chemoradiotherapy is tolerated poorly by the upper gastrointestinal tract [5,10], and lower doses may have a better therapeutic ratio. Even the most sophisticated forms of abdominal radiotherapy cannot avoid irradiating the gut mucosa and submucosa, so true radiotherapy dose escalation studies must await the development of more effective radioprotectant strategies. (3) More intensive chemotherapy may delay (though probably not eliminate) the appearance of distant metastases and is feasible with moderate dose radiotherapy. (4) Results with reduced radiotherapy doses plus intense chemotherapy are not worse than conventional dose radiotherapy plus reduced dose chemotherapy [1,6,7]. Unfortunately though, the outcomes (local control and survival) are still unacceptably poor.

Nonetheless, there must be some threshold dose of radiotherapy that is necessary for it to be a meaningful partner in combined-modality therapy, and it is unclear if the Michigan dose (36 Gy) meets this threshold. Arguably, the downsizing or elimination of radiotherapy for locally advanced but nonmetastatic epithelial cancers is a step backward, at least in settings when the treatment goal is curative. It may be more promising to bring the Michigan-style chemoradiotherapy regimen to the adjuvant setting, where local-regional control is a problem but is more achievable than distant control. An adjuvant dose of 36 Gy (with concurrent chemotherapy) is far more sensible than the same dose for massive unresectable disease. A large randomized trial by the European Study Group for Pancreatic Cancer, ESPAC-1, has questioned the value of postoperative chemoradiotherapy in pancreatic cancer [13]. However, an unorthodox randomization schema and suboptimal chemotherapy by modern standards hinder interpretation of the ESPAC-1 study. It would certainly be appropriate to test the Michigan regimen against standard chemoradiotherapy or chemotherapy alone in the adjuvant setting. This approach should be examined in other disease sites and stages in which the role of adjuvant irradiation is uncertain, particularly lung or esophageal cancer. If this is successful, (careful) consideration could then be given to radiation de-escalation plus concurrent chemotherapy for other sites such as head and neck, brain or gynecologic tumors, where postoperative radiotherapy has proven efficacy, but with significant subacute and late toxicities.

In summary, radiation oncologists must recognize that some classical ideas (radiation dose escalation) might be counterproductive in some settings. Radiotherapy is an important—if not the most important—component of treatment of many locally advanced carcinomas, but it must be given in a manner that minimizes toxicity and interference with systemic chemotherapy and targeted therapies. Phase I studies, such as the Michigan trials, that use novel techniques and dose fractionation of radiotherapy with exploratory systemic therapies, are needed just as much as conventional phase I trials.