Randomized adjuvant trials in oncology: a necessity or time-consuming luxury?

Unquestionably, progress in cancer management is happening in quantum leaps. Yet for many patients and their families, this progress is slow. We are extremely fortunate to have witnessed a steep decline in breast cancer mortality in the past decade, no doubt as a result of these innovations. However, this review shows that we could do much better.


Introduction
Th e question of whether randomized adjuvant trials in oncology are a necessity or a time-consuming luxury addresses one of the most complex aspects of human research and its clinical applications within public health. Randomized trials are an absolute necessity; they have resulted in hundreds of thousands of lives worldwide being saved in the decades since their inception. However, the present system of clinical trials based on an infrastructure devised over 40 years ago cannot now easily handle the rising challenges related to so many new developments occurring simultaneously. Human genome information, molecular classifi cation and techniques enabling the design of new agents based on tumor genetic characteristics will potentially result in the explosive development of new agents or diagnostic classifi ers that would accelerate the decline in mortality, especially from breast cancer. At present, however, all these require testing within the existing clinical trial infrastructure. Under the present system, excessive delays are inevitable, unless reforms are implemented that will dramatically accelerate the present process of 'bench to clinic' .
A series of proposals was recently articulated regarding qualitatively new approaches towards clinical trials in breast cancer. Th eir outline constitutes part of this invited review.

The clinical trial process
Current estimates of the interval between the fi rst evidence of eff ect of a new agent in stage IV disease and its introduction into guidelines in early breast cancer range from 15 to 20 years. Could this prohibitive delay be reduced to 5 years or less? Th at is the question we are asking. In order to substantially reduce the testing period, a system of reforms within the clinical trial process must take place, ideally involving the entire international community.
While tighter international collaboration is a fundamental prerequisite for implementing these suggestions, indivi dual aspects of the required reforms need to be articu lated clearly, with two re-assuring conditions essential they must not jeopardize: the quality of drug research leading to level I evidence of benefi t or harm and the determination of safety.
Only limited aspects of the required reforms will be discussed here, with the main objective to initiate a dialogue among scientists to identify the mounting problems as a pre-requisite to the actual reform process on a larger scale.

Trials in stage IV breast cancer versus the adjuvant setting
Is stage IV breast cancer a good environment in which to test a new agent in complex random breast cancer trials? Th is is our fi rst question, and its discussion may lead to modifi cations of the trial process leading to earlier implementation of neoadjuvant and adjuvant clinical trial testing.
Recent clinical evidence [1] has highlighted pivotal preclinical studies in bacteria and human cancer [2,3], confi rming that advanced disease is more resistant to antibiotics or oncology therapeutics than early disease. Th is is probably due to the substantially higher absolute number and proportion of resistant mutants in late disease. Not surprisingly, it is expected that agents scoring some clinical benefi t in stage IV breast cancer typically would score a much higher benefi t in the adjuvant setting where there are many fewer residual cancer cells.
A review of outcomes of breast cancer therapies [1] evaluated all therapeutic regimens known to be eff ective in randomized trials in stage IV breast cancer, and then tested in an identical manner trials in the adjuvant setting of breast cancer -the 'doubly tested trials' (Table 1). Th e review did not include incompletely tested agents or regimens, such as bevazcizumab (representing the class of angiogenesis inhibitors), due to either inconsistent eff ects in stage IV or absence of data from adjuvant trials, or both. Also, results of this review apply strictly to breast cancer, as diff erent solid tumors, such as colon cancer, may not allow this level of correlation. Analyses of all the doubly tested trials have shown that if a signifi cant eff ect was shown in stage IV, even more benefi t was seen in the adjuvant setting (Table 1). Specifi cally, even though a superior response or even prolongation of survival in stage IV is seen with an agent, this seldom if ever results in a cure. In contrast, the same agents tested in the adjuvant setting improve both longterm disease-free survival (DFS) and overall survival (OS) and thus impact cure rates signifi cantly ( Table 1).
As a corollary, meta-analyses by the Oxford Overview Early Breast Cancer Trialists' Collaborative Group [4,5] confi rmed a long-lasting 25 to 30% reduction in mortality from human breast cancer as a result of combination adjuvant chemotherapy, or tamoxifen or radiation. Yet none of these regimens or interventions produced longterm survival in advanced disease.
Th e legitimate question arises: could the early superiority of new agents identifi ed in the stage IV setting signal a future potential curative eff ect if used in the adjuvant setting? Could this information lead to a substantially faster next level of clinical trials in early disease than is presently the case? Probably yes, providing a set of reforms is implemented. If so, what would be the logistics of these reforms?
Th e crux of our proposals involves a speedier start of coordinated neoadjuvant and adjuvant randomized controlled trials as soon as new agents have been shown to be eff ective in stage IV disease, with these trials starting simultaneously in multiple centers ( Table 2). Th e prospect of saving thousands of lives by the discovery of an eff ective agent in the stage IV setting should galvanize the oncology community towards rapidly testing the same treatment in early stages of disease. Th us, evidence of superiority from the stage IV setting should be the 'tipping point' for launching a chain of multiple rapid trials in early disease, in the same way that any societal emergency should be handled when thousands of lives are at stake.
Th e new strategy would propose starting with neoadjuvant trials after benefi t in stage IV is seen. If results from the neoadjuvant trials confi rm the superiority seen in stage IV disease, multi center postoperative adjuvant trials would immediately follow. Th e advantage of this sequence is that neoadjuvant trials could generate results based on pathology responses much faster than any other approach and could provide critically needed guidance for the design of adjuvant trials.

The neoadjuvant setting: the ideal model for early randomized design
Neoadjuvant (preoperative) chemotherapy trials have changed our way of thinking about cancer treatment in a number of ways [6][7][8][9]. While they confi rmed identical DFS and OS rates compared to the postoperative adjuvant series, and have led to improved breast conservation, the real novelty of the neoadjuvant approach is the provision of a setting that allows more precise and detailed assessment of the in vivo response to new therapeutics [6,10,11].
Th e neoadjuvant setting permits repeated tissue testing, with the potential to select individualized approaches according to evolving pathology or molecular response criteria [12]. Furthermore, assessment of the clinical, pathological and molecular biological responses [13,14] can confi rm the superiority or inferiority of new agents much faster and more reliably than prolonged randomized controlled trials presently conducted in stage IV [15]. Since neoadjuvant studies show a very good correlation between pathologic complete response (pCR) and outcome, achieving a higher pCR with a new agentparticularly if associated with a superior response in stage IV -would further confi rm superiority.
Th e best example of this proof of principle is seen with the re-analysis of the National Surgical Adjuvant Breast and Bowel Project Protocol B-27 (NSABP B-27) trial: at the 68-month update, the addition of preoperative docetaxel to adriamycin and cyclophosphamide doubled the pCR rate from 13% to 26%, and improved the relapsefree survival, despite the fact that no changes in OS were seen in this particular trial [13]. The same agents and regimens were tested in identical design and dose regimens, fi rst in stage IV and then in the adjuvant setting. Comparison of fi nal outcomes is shown, with eff ect in stage IV expressed as response rates and eff ect in the adjuvant setting expressed as disease free survival (DFS) rates, with appropriate hazard ratios; curability is also indicated.
Importantly, the doubling of the pCR by docetaxel signaled a superiority of taxanes over anthracylines alone, a concept subsequently proved in the 2007 meta-analyses of all taxanes versus non-taxane regimens [13]. Th ese trials showed signifi cantly superior DFS and OS when adding taxanes over regimens with anthracylines alone.
Additional studies subsequently indicated that the superiority of taxanes is restricted to certain cohorts, such as those with negative Her2 status [16] or those with negative estrogen receptors [16], but overall, taking into account all patients, the improved pCR rates with taxanes in the B-27 neoadjuvant trial antedated by many years the full meta-analysis documenting superiority of taxanes over anthracyclines alone in the adjuvant setting.
Other neoadjuvant studies have validated the use of molecular responses in the context of neaodjuvant trials, with the regimen-specifi c genomic signatures significantly predicting pCR in patients treated with the appropriate regimens [12][13][14][15][16][17].
Th ese observations constitute one of the most important aspects of our proposed new approach: testing in the neoadjuvant setting as a fi rst step off ers the ability to more rapidly assess the tested agents -results become available within months rather than years. Secondly, rapid generation of these data would lead to a new dimension of information on new agents, with the rapidly growing fi elds of genomics, proteomics and tumor signatures also off ering a potential for substantially faster development of individualized therapies and or development of new agents.
Th us, the information gained from neoadjuvant testing is invaluable, and for the purpose of this review it represents excellent in vivo confi rmation of eff ects of new agents allowing a more rapid start of large adjuvant trials. Th is concept, while articulated already in the 1980s [6], is rapidly gaining popularity, yet is presently underutilized with regard to new trial reforms, which we propose.
As a result of these considerations, the following changes to the current randomized controlled trial process are proposed.

Step 1: advanced disease
Initiate and complete phase I to III trials in advanced stage populations within a shorter time frame to determine safety and levels of effi cacy much more expeditiously through a tighter coordination among multiple centers testing promising agents simultaneously in multiple randomized trials.

Step 2: the neoadjuvant setting
Once safety and eff ectiveness have been documented in stage IV disease, treatment modalities should be moved rapidly into randomized and non-randomized neoadjuvant trials to confi rm clinical eff ectiveness and to elucidate pathologic and molecular responses.

Step 3: adjuvant setting
At the fi rst indication of clinical response and or advantage over conventional therapy seen in the neoadjuvant setting, large, internationally coordinated, randomized trials including 15,000 to 20,000 patients per trial should be initiated. Th ese trials would be designed to also identify dose-response relationships, optimum duration of therapy and long-term safety.

Conclusion
Th e science of human randomized clinical trials has undergone a major renaissance since the introduction of empiricism in the early 1900s. Th e effi ciency and speed with which the NSABP clinical trial group has generated important and reliable results has been truly remarkable. Starting in the 1970s, tests of new concepts, agents, and regimens have been completed virtually annually and accepted into guidelines shortly thereafter.
A similar renaissance is due in 2011. As the tempo of new agent discovery accelerates, it is becoming increasingly clear that the current clinical trial process and its regulatory processes are unable to keep up. A clinical trial process is described that could facilitate more effi cient progress, with the potential for benefi ting many more patients much sooner. iii. Randomized trials of the agents against the conventional approach, with repeated needle biopsies to correlate clinical with pathologic and molecular responses Years 3 to 5: adjuvant testing (total 2 years) i. Large multicenter adjuvant trials with multiple international sites, involving more than 25,000 patients, to start as soon as fi rm responses are seen in the neoadjuvant setting ii. Objectives: dose response, therapy duration, long-term toxicity, and so on RCT, randomized controlled trial.

Summary
Unquestionably, progress in cancer management is happening in quantum leaps. Yet for many patients and their families, this progress is slow. We are extremely fortunate to have witnessed a steep decline in breast cancer mortality in the past decade, no doubt as a result of these innovations. However, this review shows that we could do much better.

Recommendations
As fi nal recommendations, we hope for an international consensus, that: the clinical trial process must be, and can be, accelerated; stage IV breast cancer is no longer the best setting to test new agents; we can do more and sooner if the principle arena of testing is shifted to the neoadjuvant and adjuvant settings; the transfer of positive results into community guidelines has to be expedited, with input by patient advocacy groups a strong consideration; the failure to apply mature results expeditiously should be considered a societal crisis.