EVIDENCE-BASED ONCOLOGY Evidence-based oncology in cancer treatment reviews

s revisited. Ann Intern Med 1990; 113: 69–76.50E V I D E N C E B A S E D O N C O L O G Y End Points and United States Food and Drug Adminis trat ionApproval of Oncology Drugs By John R. Johnson, Grant Williams, and Richard Pazdur Purpose: To summarize the end points used by theUnited States Food and Drug Administration (FDA) to ap-prove new cancer drug applications over the last 13 years.Materials and Methods: The FDA granted marketing ap-proval to 71 oncology drug applications between January1, 1990, and November 1, 2002. The end points used as theapproval basis for each application are presented, and therationale for each end point is discussed.Results: The FDA grants either regular marketing ap-proval or accelerated marketing approval for oncologydrug applications. Regular approval is based on end pointsthat demonstrate that the drug provides a longer life, abetter life, or a favorable effect on an established surrogatefor a longer life or a better life. Accelerated approval (AA) isbased on a surrogate end point that is less well establishedbut that is reasonably likely to predict a longer or a betterlife. Tumor response was the approval basis in 26 of 57regular approvals, supported by relief of tumor-specificsymptoms in nine of these 26 regular approvals. Relief oftumor-specific symptoms provided critical support for ap-proval in 13 of 57 regular approvals. Approval was basedon tumor response in 12 of 14 AAs.Conclusion: End points other than survival were the ap-proval basis for 68% (39 of 57) of oncology drug marketingapplications granted regular approval and for all 14 appli-cations granted accelerated approval from January 1,1990, to November 1, 2002.J Clin Oncol 21:1404-1411. © 2003 by AmericanSociety of Clinical Oncology. THERE IS a common misperception that the United StatesFood and Drug Administration (FDA) requires a survivalimprovement for approval of oncology drug marketing applica-tions. This article reviews the marketing applications for oncol-ogy drugs approved by the FDA’s Division of Oncology DrugProducts in the Center for Drug Evaluation and Research(CDER) from January 1, 1990, to November 1, 2002. The endpoints used as the approval basis for each application arepresented, and the rationale for end-point selection is discussed.Regular marketing approval of oncology drugs requires sub-stantial evidence of efficacy from adequate and well-controlledinvestigations. The attributes of adequate and well-controlledinvestigations are described in the regulations. Studies mustallow a valid comparison to a control and must provide aquantitative assessment of the drug’s effect. Guidance promul-gated in the 1980s indicated that efficacy should be demonstratedby prolongation of life, a better life, or an established surrogatefor at least one of these. Drugs must also be safe for theirintended use. The safety requirement comes from the FederalFood Drug and Cosmetic Act of 1938. A 1962 amendment tothat Act codifies the efficacy requirement.In 1992, Subpart H was added to the new drug application(NDA) regulations to allow accelerated approval (AA) fordiseases that are serious or life-threatening when the new drugappears to provide benefit over available therapy, but undersituations when the demonstrated benefit did not yet meet thestandard for regular approval. For instance, AA can be grantedon the basis of a surrogate end point that is reasonably likely topredict clinical benefit (Table 1) but that is not established to thelevel that would support regular approval. After AA, the appli-cant is required to perform a postmarketing study to demonstratethat treatment with the drug is indeed associated with clinicalbenefit. If the postmarketing study fails to demonstrate clinicalbenefit or if the applicant does not demonstrate due diligence inconducting the required study, the regulations describe a processfor rapidly removing the drug from the market. The AAregulations provide drugs that are promising on the basis ofsurrogates that are reasonably likely to predict clinical benefit topatients with serious or life-threatening diseases. Drugs that areonly similar to available therapy on the basis of such a surrogatewould not appear to be especially promising and would notprovide benefit over available therapy. If the drug showedclinical benefit, it would be granted regular approval and wouldnot need to provide benefit over available therapy.In the early 1980s, the FDA approved oncology drugs basedon tumor response rate alone. In the mid-1980s, on the advice ofthe Oncologic Drugs Advisory Committee (ODAC), the FDAdetermined that response rate generally should not be the solebasis for approval. The potential benefit associated with a partialresponse did not necessarily outweigh the substantial toxicity ofoncology drugs, and the correlation between response rate andsurvival or clinical benefit was not well established. The newFDA position called for an improvement in survival or patientsymptoms for regular approval.In subsequent years, the FDA stated that under selectedcircumstances, impressive tumor-related outcomes could beFrom the Division of Oncology Drug Products (HFD-150), Center forDrug Evaluation and Research, United States Food and Drug Administra-tion, Rockville, MD.Submitted August 9, 2002; accepted December 23, 2002.The views expressed are the result of independent work and do notnecessarily represent the views and findings of the United States Food andDrug Administration.Address reprint requests to John R. Johnson, MD, Food and DrugAdministration, HFD-150, 5600 Fishers Lane, Rockville, MD 20857; email:[email protected].© 2003 by American Society of Clinical Oncology.0732-183X/03/2107-1404/$20.0