|
|
||
 |
|
|||||
|
 |
||||||
|
||||||
|
||||||
|
Originally published as JCO Early Release 10.1200/JCO.2003.07.942 on September 8 2003 © 2003 American Society for Clinical Oncology
Hurry Up and Wait: Is Accelerated Approval of New Cancer Drugs in the Best Interests of Cancer Patients?University of Chicago, Chicago, IL IN RECENT months, the United States Food and Drug Administration (FDA) has granted marketing approval for a number of new cancer therapeutics, including oxaliplatin, imatinib, gefitinib, and bortezomib. Each of these drugs received their initial marketing approval under the terms 21 CFR 314.510 Subpart H, commonly known as the accelerated approval mechanism. Accelerated approval is a component of the Code of Federal Regulations that gives the FDA authority to grant marketing approval for a new drug product on the basis of adequate and well-controlled studies establishing that the drug has an effect on a surrogate end point that is reasonably likely to predict clinical benefit. Postmarketing studies are then required to establish clinical benefit with certainty. A surrogate end point is generally accepted to mean a laboratory measurement or physical sign used as a substitute for a clinically meaningful end point that measures directly how a patient feels, functions, or survives. For cancer therapeutics, tumor response is the most commonly used surrogate end point. Clinical benefit generally refers to improvements in survival, disease-free survival, or symptom benefit. In a recent issue of the Journal of Clinical Oncology, Johnson et al1 describe the end points that have been used successfully to obtain FDA approval under either the regular or accelerated approval mechanisms. The accelerated approval mechanism was added to the Federal Regulations in 1993 and was intended for diseases that are serious or life-threatening when no other therapy is available. Approval for marketing can be withdrawn by the FDA if postmarketing studies fail to confirm clinical benefit. At the 2003 Annual Meeting of the American Society of Clinical Oncology, Hirschfeld et al2 reviewed the FDA experience with the accelerated approval mechanism. Since 1995, there have been 16 oncologic drug products approved under these regulations. Irinotecan, docetaxel, capecitabine, temozolomide, oxaliplatin, imatinib, gefitinib, and bortezomib were all approved using response rate as the surrogate end point. Three of these agents, irinotecan, capecitabine, and docetaxel, have subsequently obtained full approval on the basis of randomized clinical trials that have confirmed clinical benefit. A cardioprotective drug, dexrazoxane, was also converted from accelerated to full approval after completion of definitive randomized trials. The accelerated approval mechanism has generally been considered an effective mechanism for speeding promising new agents to the bedside and the pharmacy. However, the validity of this regulatory strategy rests on several key assumptions: that tumor response is a reasonable surrogate for clinical benefit, that sufficient safety data exist at the time of accelerated approval to have confidence in the toxicity profile of the drug, and that confirmatory postmarketing studies will be completed expeditiously. One might reasonably ask whether the accelerated approval mechanism enhances or hinders a sponsor’s ability to complete definitive clinical trials with a new agent and, therefore, whether this regulatory strategy is in the best interests of cancer patients. Tumor response is generally accepted as a treatment effect because it is unlikely that a tumor will undergo spontaneous regression. Although considerable effort has been expended to refine response criteria,3 the type of response (complete or partial), duration of response, reproducibility of response, and site of response might all be more important than the rate of response in assessing the potential benefit of a new agent. Indeed, Hillman et al4 have presented data from cooperative group studies to suggest that measurement of 10 lesions, as required by the Response Evaluation Criteria in Solid Tumors, does little to improve the precision of the response rate estimate compared with measuring only two lesions. Although prospective definition of response criteria is critical in any clinical trial, independent confirmation of response by an expert review panel is perhaps more important in the open-label, nonrandomized studies that frequently form the basis of New Drug Applications seeking accelerated approval. Almost without exception, the response rate assessed by such independent review is lower than that reported by investigator assessments. Few studies have addressed whether response rate is truly predictive of clinical benefit. For hematologic malignancies, complete remission is generally associated with symptomatic improvement, such as reduced transfusion requirement and reduced infection rate. Thus, for these diseases, complete response is usually accepted as predictive of clinical benefit. Buyse et al5 conducted a meta-analysis of 25 colorectal cancer trials involving fluoropyrimidines and concluded that tumor response was a highly significant predictor of survival independent of performance status. By contrast, an analysis by Chen et al6 led to the conclusion that phase II response rates in patients with extensive small-cell lung cancer were not correlated with median survival in phase III trials of the same regimens. It is encouraging to note, however, that patients treated with irinotecan,7 docetaxel,8 capecitabine,9 and oxaliplatin10 all experienced improved survival in randomized clinical trials despite relatively low response rates (ranging from 9% for oxaliplatin11 to approximately 40% for docetaxel12) and few complete remissions in the studies used as the basis for accelerated approval. It seems therefore that response rate is reasonably likely to predict clinical benefit, at least for certain diseases and certain drugs. Issues that remain unsettled are whether there is a minimum response rate that is predictive of benefit and whether a different surrogate end point can be predictive of clinical benefit for those drugs that do not cause tumor regression. Time to progression could be an informative end point in this context, but only when evaluated in a randomized trial that controls for rate of disease progression in the absence of a treatment effect. Given that accelerated approval is often based on phase II studies involving small numbers of patients, it is also reasonable to ask whether there is sufficient data available at the time of accelerated approval to ensure that new agents are safe. The number of patients included in the safety databases for docetaxel, irinotecan, and capecitabine were 1,435, 305, and 570, respectively, at the time of accelerated approval of these drugs.13 At the time of full approval, these numbers had increased to 2,045, 955, and 875, respectively. It is important to note that the safety database for these drugs, although larger, did not markedly increase at the time of full approval. Indeed, 3 years after the full approval of irinotecan, two national randomized clinical trials involving this drug were interrupted because of concerns of previously unrecognized thromboembolic toxicity and early death.14 Thus, although the safety database available at the time of either accelerated or full approval is likely to permit assessment of common side effects, uncommon toxicities, rare drug interactions, unusual pharmacogenetic syndromes, and toxicities that occur only in special populations may not be recognized until the drug is more widely used in large-scale clinical trials or in the clinical practice setting.15 Earlier this year the Oncologic Drugs Advisory Committee of the FDA addressed the impact of accelerated approval on the completion of postmarketing studies.16 Studies are slow to be initiated by sponsors and accrual is often slower than expected. There are many potential reasons for this, but among them are that patients and doctors have been willing to accept that a commercially available product is proven to be safe and effective even when the definitive studies to support this conclusion have not been completed. Thus there is little incentive to enroll in trials when a drug is both available and reimbursed. As new agents enter the clinic, there is a universe of competing trials that may generate greater interest than additional studies of a recently approved drug. Indeed, after accelerated approval, increased clinical experience with the agent may diminish enthusiasm for it or, conversely, increase its use in a population different from that being studied in confirmatory trials. Increasingly, the availability of new products or advances in science may reduce the relevance or feasibility of conducting postmarketing studies. Although the FDA has the authority to withdraw marketing approval for a drug if clinical benefit cannot be confirmed, such an action has never been taken for an oncology product, which leads to concerns that drugs of unproven benefit may remain on the market for prolonged periods of time. A fundamental assumption regarding the accelerated approval mechanism is that it leads to faster drug approval. Hirschfeld et al2 have demonstrated that regulatory review represents only approximately 5% of the total clinical development time for most new oncology drug products. Nevertheless, data presented by Hirschfeld et al at the 2003 Annual Meeting of the American Society of Clinical Oncology indicate that the median time from investigational new drug filing to marketing approval is 2,000 days (5.5 years) shorter for drugs approved by the accelerated approval mechanism. Interestingly, however, there is no significant difference in time to approval between the applications with the fastest approval times by either the accelerated or standard mechanism, suggesting that a well-designed clinical development plan can lead to rapid marketing approval by either route. Sponsors and the FDA must work together to improve the process of oncology drug development. Confirmatory trials should be ongoing at the time when accelerated approval is granted. Rather than basing approval on single-arm, open-label studies, the FDA has now proposed that accelerated approval could be based on a planned interim analysis of a definitive randomized trial, as was the case with the recent accelerated approval for oxaliplatin. The advantages of this approach are obvious. The confirmatory trials are already ongoing at the time of marketing approval, accelerated approval could be applied to less refractory patient populations than has customarily been the case, accelerated approval in the context of a randomized control group gives greater confidence that the new agent is producing the observed effect, and a randomized control group may permit other end points, such as time to progression, to be used as a basis for accelerated approval. Importantly, an interim analysis of a randomized trial ensures that a larger safety database will soon be available before extensive postmarketing use of the drug occurs. In view of the relatively small safety database available for many agents at the time of accelerated approval, enhanced and simplified postmarketing surveillance for unexpectedly severe toxicities, drug interactions, and increased risk in special populations should be developed. It is clear that the accelerated approval mechanism can reduce the time to marketing approval for use of some drugs in patients with advanced, refractory cancer. This has the advantage of making the drug widely available, enabling off-label use, and facilitating investigator-initiated clinical studies that may lead to expanded indications. Thus, as currently used, accelerated approval provides a mechanism for more rapid drug approval but may, in fact, reduce the opportunity to prove that new drugs are truly safe and effective. A case in point is the recent accelerated approval of gefitinib as single-agent therapy for patients with advanced non–small-cell lung cancer. Marketing approval was based on results from two uncontrolled, randomized, phase II studies (the Iressa Dose Evaluation in Advanced Lung Cancer [IDEAL] studies 1 and 2) in patients with non–small-cell lung cancer refractory to platinum-based therapy (IDEAL-1) or platinum and docetaxel chemotherapy (IDEAL-2). Objective tumor response was observed in approximately 19% and 11% of patients in these trials, respectively.17,18 Two large, prospective, placebo-controlled trials were also conducted to examine the value of adding gefitinib to combination chemotherapy (Iressa Non–Small-Cell Lung Cancer Trial Assessing Combination Treatment studies 1 and 2). Unfortunately, addition of gefitinib to combination chemotherapy failed to improve any parameter of clinical outcome.19,20 Nevertheless, gefitinib is now marketed and its widespread availability will, undoubtedly, slow the accrual to the next generation of randomized trials necessary to demonstrate benefit, long-term safety, and optimal integration into lung cancer treatment programs. Given the experience to date, it is reasonable to conclude that the accelerated approval mechanism is in the best interests of cancer patients. However, improvements in the drug development process are clearly necessary to speed drugs to market that are proven to be safe and effective at the time when they arrive in the pharmacy or soon thereafter. AUTHOR’S DISCLOSURES OF POTENTIAL CONFLICTS OF INTEREST The author or his immediate family members have indicated a financial interest. No conflict exists for drugs or devices used in a study if they are not being evaluated as part of the investigation. Acted as consultant within the past 2 years: Richard L. Schilsky (Bristol-Meyers Squibb, Novartis, Bionumerik, NeoPharm, Eisai, Sankyo, Introgen, CominatoRx, Geron, Abbott, Baxter, and Millennium). Performed contract work within the past 2 years: Richard L. Schilsky (Novartis, Bionumerik, and Lorus). Received more than $2,000 a year from a company for either of the past 2 years: Richard L. Schilsky (Abbot and Bionumerik). REFERENCES
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
|
|||||||||||
|
|||||||||||
|
|
Copyright © 2003 by the American Society of Clinical Oncology, Online ISSN: 1527-7755. Print ISSN: 0732-183X
|
