|Home | About | Journals | Submit | Contact Us | Français|
Accelerated approval (AA) was initiated by the US Food and Drug Administration (FDA) to shorten development times of drugs for serious medical illnesses. Sponsors must confirm efficacy in postapproval trials. Confronted with several drugs that received AA on the basis of phase II trials and for which confirmatory trials were incomplete, FDA officials have encouraged sponsors to design AA applications on the basis of interim analyses of phase III trials.
We reviewed data on orphan drug status, development time, safety, and status of confirmatory trials of AAs and regular FDA approvals of new molecular entities (NMEs) for oncology indications since 1995.
Median development times for AA NMEs (n = 19 drugs) and regular-approval oncology NMEs (n = 32 drugs) were 7.3 and 7.2 years, respectively. Phase III trials supported efficacy for 75% of regular-approval versus 26% of AA NMEs and for 73% of non–orphan versus 45% of orphan drug approvals. AA accounted for 78% of approvals for oncology NMEs between 2001 and 2003 but accounted for 32% in more recent years. Among AA NMEs, confirmatory trials were nine-fold less likely to be completed for orphan drug versus non–orphan drug indications. Postapproval, black box warnings were added to labels for four oncology NMEs (17%) that had received AA and for two oncology NMEs (9%) that had received regular approval.
AA oncology NMEs are safe and effective, although development times are not accelerated. A return to endorsing phase II trial designs for AA for oncology NMEs, particularly for orphan drug indications, may facilitate timely FDA approval of novel cancer drugs.
Accelerated approval (AA) regulations were established by the US Food and Drug Administration (FDA) in 1992.1 These regulations are designed to improve access to therapeutics for life-threatening diseases by allowing sponsors to begin marketing relevant drugs on the basis of trials that identify improvements in surrogate outcomes, such as response rate or progression-free survival, that are reasonably likely to predict clinical benefit.1 A 1996 initiative from the Office of the President of the United States expanded use of the AA process for cancer therapies by allowing sponsors to seek AA for oncology new molecular entities (NMEs).1,2 The regulation requires sponsors to conduct confirmatory studies to verify clinical efficacy.2,3 An important feature of AA is that the registration trial can be phase II or phase III in design. AA designation of NMEs for serious medical illnesses differs in intent from the fast track designation for drugs that are first-in-class (wherein sponsors are allowed to submit the application for FDA approval in parts) and from the priority review for drugs that address unmet medical needs (wherein the FDA agrees to review the application in 6 months versus the standard 10-month time period).
In 2003, the US Food and Drug Administration requested that the Oncologic Drugs Advisory Committee (ODAC) review difficulties faced by sponsors of AA oncology NMEs who had failed to complete required confirmatory trials (Fig 1).4 Drugs discussed at this meeting had received AA on the basis of surrogate clinical outcome findings reported in phase II trials. In a journal publication, but not as formal agency guidance, FDA authors summarized recommendations from this meeting, which advised sponsors to consider adding sites to confirmatory phase III trials in countries where access to new cancer drugs is limited.5 In 2005, a second ODAC meeting reviewed continuing difficulties faced by sponsors who had attended the 2003 ODAC meeting.6 The consensus was that sponsors had conscientiously implemented recommendations from the prior meeting, yet recruitment to confirmatory phase III trials continued to be difficult. The FDA advised sponsors that future AA applications should preferentially be based on interim analyses of phase III (rather than phase II) trials, thus obviating the need for recruitment to additional confirmatory phase III trials.6–8 Also, FDA staff advised the ODAC and sponsors that single-arm, phase II studies are interpretable only for the purposes of AA applications in the setting of refractory disease. In contrast, interim analyses of phase III trials could support AA in a broad range of settings, and additional follow-up of these trials could provide evidence of clinical benefit.
Opinions on the success of the AA regulation vary. In the 2004 publication, Dagher et al5 concluded that AA had been successful in facilitating rapid evaluation of novel agents to patients with cancer.5 However, Congressman Ed Markey has voiced concern that sponsors have abused the AA process by failing to confirm efficacy or by promoting unsafe drugs. In 2007, Congress passed legislation to allow the FDA to impose fines on sponsors and to limit drug distribution if postmarketing commitments are not completed.9 Another concern is that obtaining AA, apparently, has become more difficult recently.10,11 Proposed legislation would shorten FDA review times for sponsors seeking AA to 4 months from the current 6- to 10-month time period in an effort to shorten the development time of oncology NMEs.12
We review the experience to date with AA for oncology NMEs. We evaluated whether AA for oncology drugs facilitated rapid access to effective agents; whether confirmatory trials confirming efficacy of these drugs are completed; and whether safety concerns are identified after AA is granted. We also evaluated temporal changes in the frequency with which oncology NMEs received approval with regular-approval versus AA mechanisms.
The first AA for a cancer indication was granted in May 1995. We reviewed information on new drug applications for oncology NMEs approved by the FDA between 1995 and 2008. Oncology drugs that previously received FDA approval for another indication or for supportive care indications were excluded. Data sources included clinical trial reports available under the Freedom of Information Act for new drug applications, transcripts of ODAC meetings in 2003 and 2005 on AA, efficacy and safety information available to the public on the FDA Web site, and package inserts for information on oncology NMEs that received AA and regular approval.4,6,13 Details of trials that supported AA, regular approval, or conversion from AA to regular approval were obtained from the Center for Drug Evaluation and Research database.14 Cancer indications are designated as orphan drug indications if the United States prevalence is fewer than 200,000 persons and if the FDA grants this status.15,16 Drug development times, operationally defined as the number of months between granting of an investigational new drug (IND) authorization and FDA marketing approval, were based on information obtained from pharmaceutical sponsors or the FDA. For 25% of the NMEs, the manufacturers reported that IND authorization dates were proprietary.
For analyses of FDA approvals that were based on calendar year, we defined three distinct time periods. The first time period, 1995 to 2000, encompassed applications for FDA approval of oncology NMEs in which a development time line had been established before the 1996 Presidential Initiative that established the AA process for oncology drugs. The second time period, 2001 to 2003, encompassed applications for FDA approvals of oncology NMEs in which the development time line was likely to have been established after the 1996 Presidential Initiative. The third time period, 2004 to 2008, encompassed applications for FDA approvals of oncology NMEs in which the development time line was influenced in large part by the recommendations of the FDA and the ODAC at the 2003 ODAC meeting on AA (Fig 1).
Statistical analyses involved comparisons of median values and proportions for characteristics of trials that supported AA and regular approval. We used Kaplan-Meier curves and a Cox model to evaluate time to fulfillment of subpart H requirements (ie, confirmation of clinical efficacy for drugs that initially received AA on the basis of improvements in surrogate outcome measures) through February 15, 2009. Stata version 10.1 (Stata, College Station, TX) was used.
Since the first AA for an oncology (supportive care) indication was granted in 1995, 51 NMEs have received FDA approval for cancer therapeutic indications: 32 with regular approvals and 19 with AAs (Tables 1 and and2).2). The first AA for a therapeutic oncology indication, for docetaxel for the treatment of metastatic breast cancer, was in 1996. AAs accounted for 37% of oncology NMEs that received FDA approval between 1995 and 2008 and accounted for 26% during 1995 to 2000 (ie, mean of one AA per year); they accounted for 78% during 2001 to 2003 (ie, mean of 2.3 AAs per year); and they accounted for 32% during 2004 to 2008 (ie, mean of 1.2 AAs per year; Table 3; Fig 2).
Thirty-one oncology NMEs received FDA approval for treatment of solid tumors: 66% of the regular approvals and 47% of the AAs for oncology NMEs. Twenty-three regular approvals and AAs were for treatment of common solid tumors, including cancer of the breast (n = 10 and n = 2, respectively), prostate (n = 4 and n = 0, respectively), lung (n = 1 and n = 1, respectively), and colorectum (n = 1 and n = 4, respectively). For less-common solid cancers, seven oncology NMEs received regular approval (for cancers of the kidney, cervix, esophagus, or bladder and mesothelioma) and one received AA (for renal cell cancer). Overall, 53% of oncology NMEs that received regular approval and 63% of oncology NMEs that received AA were for orphan drug indications.
Phase III trials supported efficacy findings for 75% of regular-approval versus 26% of AA oncology NMEs. Response rates or times to progression were the primary surrogate end points in registration trials for 89% and 11% of the AA oncology NMEs, respectively. Non–orphan drug indication NMEs were 1.5-fold more likely than orphan drug indication NMEs to receive initial FDA approval on the basis of phase III studies (73% v 45%, respectively). The median number of patients included in studies that supported regular versus AA approval was 1.5-fold greater for non–orphan drug indications (469 v 304 patients, respectively) and was 1.3-fold greater for orphan drug indications (201 v 152 patients, respectively).
The median development times for oncology NMEs that received AA and for those that received regular approval were 7.3 years and 7.2 years, respectively. Between 2001 and 2003, the median development time was slightly shorter for NMEs approved for non–orphan drug cancer indications (8.1 years) than for orphan drug cancer indications (8.9 years). Since 2004, median development times were slightly longer for orphan drug (7.8 years) versus non–orphan drug cancer indications (5.7 years; Table 4).
Eleven (63%) of 19 oncology NMEs granted AA have had clinical benefit confirmed in subpart H trials and have been granted regular approval. Subpart H trials were completed for 71% of oncology NMEs associated with non–orphan drug tumor indications versus 58% of NMEs associated with orphan drug cancer indications. The completed subpart H commitment trial for one drug, gefitinib, did not identify clinical benefit, and the application for marketing approval was withdrawn by the sponsor. Cox proportional hazard models indicated that, among NMEs that received AA, oncology NMEs in which FDA approvals were for orphan dug cancer indications were nine-fold less likely to confirm clinical benefit and to complete subpart H commitments than those in which FDA approval was for non–orphan drug indications (hazard ratio, 0.11; 95% CI, 0.02 to 0.63; P < .01). The median times between granting AA and granting regular approval were 2.1 years for orphan drugs indications and 3.7 years for non–orphan drug oncology indications (Table 4; Fig 3).
Of oncology NMEs that received AA, 63% are included in first-line cancer treatment regimens from the National Comprehensive Cancer Network (NCCN). Among NMEs that received FDA approval for non–orphan oncology indications, 83% that received AA and 91% that received regular approval are included in first-line regimens in NCCN clinical guidelines. Among NMEs that received FDA approval for orphan drug indications, 82% of regular-approval drugs and 33% of AA drugs are included in first-line regimens in NCCN guidelines.
After FDA approval, four black box warnings were added to package inserts for four (17%) oncology NMEs that had received AA, and three black box warnings were added to package inserts for two (9%) oncology NMEs that had received regular approval.
AA has facilitated access to many safe and highly effective NMEs that are indicated for cancer treatment. The FDA's published assessment on the first decade of the AA process for oncology drugs was that the regulation was very successful.5 In May 2008, Richard Pazdur, MD, Director of the Office of Oncology Drug Products of the FDA, publicly stated that the AA regulation continues to be highly successful in facilitating early access to large numbers of novel cancer drugs.17 The findings of this report suggest an alternative, less positive, interpretation of the recent experience with AA. Although the AA process previously facilitated early access to new oncology drugs, it is now difficult to obtain FDA approval with the AA process. Overall, fewer oncology NMEs receive AA versus regular FDA approval in the recent time period. In interpreting these findings, several factors should be considered.
Increasing the threshold for granting AA for oncology NMEs would be appropriate if large numbers of these drugs are subsequently found to be ineffective. This does not appear to be the case. The majority of AA drugs that receive FDA approval for non–orphan drug indications (including docetaxel, sunitinib, irinotecan, capecitabine, and oxaliplatin) have had clinical benefit confirmed in subpart H trials, are included in first-line chemotherapy regimens from the NCCN, and represent important therapeutic breakthroughs. Similarly, many oncology NME drugs that received AA for orphan drug indications (including imatinib, bortezomib, alemtuzumab, temozolomide, and cetuximab) also represent important therapeutic breakthroughs, have had clinical benefit confirmed in subpart H clinical trials, are included in first- or second-line chemotherapy regimens from the NCCN, and have led to improved survival for large numbers of patients with cancer. As reported here, when confirmatory phase III trials for an AA oncology NME are not completed, the approved clinical indication is almost always an orphan drug indication (eg, pediatric T-cell leukemia). Patients with cancer who have these rare diagnoses are generally unwilling to participate in a randomized, phase III trial in which there is a high likelihood that they would not receive a novel agent that recently received AA.
A second proposed rationale for increasing the threshold for granting AA is that many of these drugs are ultimately found to be unsafe. This, too, does not appear to be the case—at least with respect to serious adverse drug reactions. Of four black box warnings added to package inserts associated with AA oncology NMEs, three were added more than 2 years after approval, which suggests that safety signals could not have been recognized during clinical trials. Of three warnings added to package inserts for regular-approval oncology NMEs, one (for rituximab) was added 1 year after FDA approval; another (for trastuzumab) was added within 2.5 years of FDA approval; and an additional black box warning for rituximab was added 9 years after approval. Previous data has shown that identification of severe adverse drug events occur at approximately equal rates in phase II and phase III trials.18 We note that more rapid adoption and growth in usage for regularly approved NMEs may account for some differences.
The original motivation for extending AA from the HIV setting to the cancer setting was that this process would facilitate early access to novel cancer agents. Shortened development times would result from sponsors applying for AA on the basis of analyses of surrogate clinical outcomes. In the early 2000s, the AA process appeared to be in full gear: 78% of oncology NMEs received AA versus regular approval. However, since 2004, only 32% of oncology NMEs have received AA versus regular approval. This change coincided with the failure of gefitinib to show clinical benefit in its confirmatory phase III trial and with concern expressed by FDA officials at ODAC meetings in 2003 and 2005 that sponsors were not completing agreed-on phase III trials designed to verify improvements in clinical outcomes for AA oncology NMEs. To facilitate timely enrollment in these trials, FDA officials encouraged sponsors to base initial AA applications on surrogate clinical outcomes reported in interim analyses of phase III trials rather than on final analyses that identified improvements in surrogate clinical measures for patients enrolled on phase II trials.7 This recommendation mirrors FDA policy for HIV infection, but HIV infection is a limited, special case that does not represent the scope or complexity of the many diseases that have a malignant phenotype. The FDA approach to oncology drugs has been hampered by failure to systematically develop and analyze the range of biomarkers used in AA for validation as surrogate markers. Development of surrogate markers is a major thrust of modern therapeutic research, is the goal of several initiatives that include FDA participation, and is an area of pressing public health need to shorten time to market and to lessen costs.
Some study limitations should be noted. First, we focused on oncology NMEs intended as antitumor therapy and did not evaluate oncology drugs with AA that had previously received US Food and Drug Administration approval for other indications. These drugs differ from NMEs in that they are available, and frequently are reimbursed, for use in off-label settings; hence, AA designation is less relevant. Second, IND dates were used to derive estimates of development time. These dates were not available for 25% of the NMEs. However, characteristics of drugs for which IND dates were not available are similar to those for the entire population of oncology NMEs. Moreover, our metric of clinical development time, defined from IND date to approval, may not most accurately reflect clinical development in the human population. Pharmaceutical sponsors and the FDA should publicly disseminate information on IND dates and development times for all drugs and should allow patients and clinicians to have information on the duration of time that a drug was evaluated before it received FDA approval. Patients may be more willing to take AA medications with longer development times, and clinicians may be more willing to prescribe these medications. Third, information on new drug applications and characteristics of registration trials targeted for regular approval versus AA of oncology NMEs are not readily available. The possibility remains that, with maturation of AA experiences, regular FDA approval may be based on improvements in surrogate end points previously evaluated in AA applications. Finally, an alternative explanation for our findings is that fewer applications for AA have been made. Raw data on the number of applications for AA are not available. However, start-up pharmaceutical manufacturers have expressed their support for basing AAs for novel oncology NMEs on results of phase II trials, as the opportunity to achieve positive revenue in a short time frame is essential for their financial viability. They also note that the phase III trial design is particularly difficult for evaluating NMEs designed to treat orphan drug cancer indications, as the trials are expensive, require several-fold more patients with rare cancer diagnoses than phase II trials, involve larger numbers of clinical trial sites, require extensive collaboration of large numbers of physicians, and take longer to complete.
We conclude that the promise of AA (ie, shortening the time to approval and decreasing the resource burden of novel cancer drugs) is not being met, as evidenced by the similar development times, emphasis on phase III trial designs for both regular approval and AA, and the decreased number and percentage of oncology NMEs that receive AA since 2003. Going forward, for oncology NMEs associated with orphan drug indications, the phase II trial design may be optimal for identifying improvements in surrogate clinical outcomes, shortening development times, and lowering the bar for achieving AA. AA oncology NMEs in which clinical efficacy is not confirmed in phase III trials are usually associated with orphan drug indications, for which accrual to confirmatory phase III trials is particularly difficult. Punitive actions included in recent legislation, such as assessment of fines for failing to complete postapproval commitment studies, should be undertaken with caution for sponsors of drugs that receive AA for oncology indications. Although this approach also is proposed in the European Union, such an approach would represent a disincentive for sponsors to seek AA in the United States and ultimately would present a disservice to patients who have cancers associated with orphan drug indications.
Supported in part by Grants No. 1R01CA 102713-01 and P 30 CA60553 from the National Cancer Institute (C.L.B. and J.M.M.).
Presented in part at the 2008 Annual Meeting of the American Society of Hematology, December 6-9, 2008, San Francisco, CA; and at the Oncology Drug Development Conference at the American Enterprise Institute, March 13-14, 2008, Washington, DC.
Authors' disclosures of potential conflicts of interest and author contributions are found at the end of this article.
The author(s) indicated no potential conflicts of interest.
Conception and design: E. Alison Lyons, June M. McKoy, Steven Trifilio, Oliver Sartor, Al B. Benson III, Kenneth R. Carson, Beatrice J. Edwards, Timothy M. Kuzel, Dennis W. Raisch, Martin S. Tallman, Dennis P. West, Antonio J. Grillo-Lopez, Charles Bennett
Financial support: June M. McKoy, Charles Bennett
Administrative support: Elizabeth A. Richey, Thanh Ha Luu
Provision of study materials or patients: Jonathan R. Nebeker
Collection and assembly of data: Elizabeth A. Richey, E. Alison Lyons, Jonathan R. Nebeker, Thanh Ha Luu, Narissa Nonzee, Steven Trifilio, Douglas Gilchrist-Scott, Steven Hirschfeld
Data analysis and interpretation: Elizabeth A. Richey, E. Alison Lyons, Jonathan R. Nebeker, Veena Shankaran, June M. McKoy, Steven Trifilio, Kenneth R. Carson, Beatrice J. Edwards, Douglas Gilchrist-Scott, Dennis P. West, Steven Hirschfeld, Antonio J. Grillo-Lopez, Charles Bennett
Manuscript writing: Elizabeth A. Richey, Jonathan R. Nebeker, Veena Shankaran, June M. McKoy, Narissa Nonzee, Oliver Sartor, Al B. Benson III, Beatrice J. Edwards, Douglas Gilchrist-Scott, Dennis P. West, Charles Bennett
Final approval of manuscript: Elizabeth A. Richey, E. Alison Lyons, Jonathan R. Nebeker, Veena Shankaran, June M. McKoy, Thanh Ha Luu, Narissa Nonzee, Steven Trifilio, Oliver Sartor, Al B. Benson III, Kenneth R. Carson, Beatrice J. Edwards, Douglas Gilchrist-Scott, Timothy M. Kuzel, Dennis W. Raisch, Martin S. Tallman, Dennis P. West, Steven Hirschfeld, Antonio J. Grillo-Lopez, Charles Bennett