Kerin B. Adelson, MD

: To characterize clinical trials supporting FDA approval of new cancer therapeutics and of supplementary indications and off-label indication inclusions on DRUGDEX, a Medicare-referenced compendium, each of which can guarantee reimbursement by Medicare.

: Cross-sectional study using publicly available documents for cancer therapeutics initially approved by the FDA from 2005-2012. Trials supporting approval were identified, characterizing randomization, blinding, comparator, end point, number of patients, and duration.

: Between 2005-2012, FDA approved 37 new cancer therapeutics for 39 indications, based on 50 supportive trials. These therapeutics subsequently received 21 FDA supplementary indication approvals and 16 DRUGDEX off-label indication inclusions, based on 22 and 37 supportive trials, respectively. Of 109 total trials, 53.2% (95% CI, 43.9%-62.3%), 28.4% (95% CI, 20.8%-37.5%), and 53.2% (95% CI, 43.9%-62.3%) were randomized, double-blinded, and used a comparator, respectively. There was a median of 383 (interquartile range, 178 to 623) patients among aggregated supportive trials, whereas 38.2% (95% CI, 28.1%-50.6%), 69.7% (95% CI, 58.7%-78.9%), and 18.4% (95% CI, 11.3%-28.6%) were supported by at least 1 trial that lasted ≥6 months, used a comparator group, or used overall survival as a primary end point, respectively. There were few substantive differences in the aggregated clinical trial evidence supporting FDA new drug and supplementary indication approvals and DRUGDEX off-label indication inclusions.

: The evidence supporting DRUGDEX off-label indications was similar to that used for FDA approval of new and supplementary cancer therapeutic indications, all of which had limitations for informing clinical decision making.

Understanding the evidence supporting new drug and supplementary indication approvals by the FDA is important for cancer care. Because many cancer patients face life-threatening disease for which there is not necessarily an effective treatment, anecdotal evidence can exert a strong influence on clinical decisions,

 making rigorous and objective evaluation by the FDA essential. The FDA approves the use of cancer therapeutics for specific medical indications only when “adequate and well-controlled investigations” can demonstrate the therapeutic’s safety and efficacy.

 Although physicians can use therapeutics “off-label”—that is, for an indication not specifically approved by the FDA&mdash;FDA approval guarantees an indication’s reimbursement by CMS.

 Reimbursement, in turn, may influence how cancer therapeutics are used in an era of increasingly expensive cancer treatment.

For cancer therapeutics, there are 2 ways by which pharmaceutical manufacturers can ensure reimbursement for indications beyond those initially approved by the FDA. First, manufacturers can apply to the FDA for a second (or third, and so on) “supplementary indication” by providing supportive clinical evidence of a therapeutic’s efficacy and safety when used for the new indication. Second, manufacturers can seek the inclusion of off-label indications on Medicare-referenced compendia, which are comprehensive drug guides that list off-label indications deemed appropriate based on expert review of biomedical literature. The Omnibus Budget Reconciliation Act of 1993 required that CMS reimburse off-label cancer therapeutic use endorsed by these compendia.6 Because some private insurers are required to follow the same compendia per state legislation,

 inclusion of an off-label indication on compendia can secure coverage without FDA indication approval.

Concerns have been raised regarding the quality of clinical evidence supporting the FDA approval of new cancer therapeutics,

 particularly the reliance on surrogate markers of disease as opposed to clinical outcomes.

 One prior study found that new drug and supplementary indication approvals for cancer drugs were supported by trials focused on clinical outcomes with similar frequency, but that trials supporting supplementary indication approvals more frequently used active comparators (ie, compared the new therapeutic with an available therapeutic alternative).

 However, no studies have comprehensively assessed the quality of evidence supporting the inclusion of off-label indications on compendia for cancer therapeutics, even though a 2009 review by the Agency of Healthcare Research and Quality raised concerns regarding whether compendia used consistent standards and methodologies.

 Our objective was to characterize and compare the clinical trials supporting new drug and supplementary indication approvals by the FDA and off-label indication inclusion on DRUGDEX, a Medicare-referenced compendium, focusing on trial number, size, design, duration, and end points. We used DRUGDEX, the compendium managed by Micromedex, because all of its supporting documentation is publicly available, it may list the greatest number of off-label indications among Medicare-referenced compendia,

 and its classification of indications as on-label versus off-label facilitated comparison with the FDA approval process.

Drugs@FDA is a publicly accessible database available through the FDA’s website that lists regulatory actions, such as approvals and drug labeling changes, for all currently approved prescription therapeutics.

 The database also contains documents relevant to each therapeutic’s regulatory history, including approval letters and other correspondence, medical reviews for new drug applications and biologic license applications, and current and historical drug labels. Clinical evidence supporting each FDA-approved indication is summarized on these labels. The Drugs@FDA database was downloaded on March 6, 2015, and October 12, 2015.

Micromedex DRUGDEX® (Truven Health Analytics) is 1 of 5 compendia currently recognized by the CMS under the Omnibus Budget Reconciliation Act.

 The most recent version of the compendium is available online by subscription. Data regarding revision history were not available on request. Each therapeutic’s monograph in DRUGDEX contains information on dosing, pharmacokinetics, safety, and clinical applications. Clinical applications include information on both FDA-approved and off-label indications, with a description of supporting evidence reviewed. DRUGDEX monographs were downloaded on March 9, 2015, and October 12, 2015.

We limited our sample to novel therapeutics originally approved by the FDA between January 1, 2005, and December 31, 2012, for the treatment of cancer, as identified in a previous study (Figure).

 This time period allows at least 2 (and up to 10) years of follow-up for therapies to acquire supplementary indications and DRUGDEX off-label indications. Our sample excluded generic drugs, reformulations, and nontherapeutic agents, as well as non-anticancer agents (eg, plerixafor or tbo-filgrastim). Each therapeutic was categorized by original FDA approval year and as a pharmacologic or biologic agent.

Identifying Supplementary FDA Indications

We searched Drugs@FDA to identify supplementary indications for adults approved through March 6, 2015, and updated our search on October 12, 2015. Supplementary indications were identified by searching for regulatory decisions modifying the “Indication and Usage” section of the drug label. We included both supplementary indications approved through an additional new drug application or biologic license application, as well as those approved through an “efficacy supplement” to the therapeutic’s original new drug application or biologic license application.

 We excluded regulatory decisions that limited the scope of prior indications, or qualified or reworded a prior indication without significantly changing its scope. We excluded supplementary indications that only expanded a therapeutic’s use to pediatric patients, to facilitate comparison against off-label indications reimbursed by Medicare due to their inclusion on DRUGDEX.

Identifying Off-Label Indications in DRUGDEX

We last searched DRUGDEX on October 12, 2015, to identify off-label indications currently listed with a Class I, IIa, or IIb strength of recommendation or efficacy, per CMS criteria for what constitutes a “medically accepted indication.”

We characterized each identified indication by approval type (FDA new drug, FDA supplementary, or DRUGDEX off-label). Indications were classified into “hematologic,” “solid tumor, curative intent,” and “solid tumor, noncurative intent.” “Noncurative intent” refers to indications that explicitly mention metastatic or advanced disease, salvage therapy, second-line use or use after other therapies, or failure of prior therapies (eg, disease that is refractory, resistant, relapsed, or progressive after prior therapy). We also identified whether FDA supplementary and DRUGDEX off-label indications treated the same target cancer type as the original FDA indication.

Identification of Evidence Supporting Therapeutic Efficacy

We identified the clinical evidence of therapeutic efficacy supporting FDA new drug and supplementary approvals and DRUGDEX off-label indications. For FDA new drug approvals, we used medical reviews to identify “pivotal efficacy trials” supporting the indication’s approval, as described in a previous study.

 For FDA supplementary indications, medical review documents were generally not available. Instead, we included all trials listed on FDA drug labels issued at the time of an indication’s approval as supportive evidence for that indication. Per FDA guidance, labels should include “clinical studies that provide primary support for effectiveness…provide other important information about a drug’s effectiveness…and prospectively evaluate important safety end points.”

 To obtain further details for trials listed on labels, we performed literature searches to identify associated publications. For DRUGDEX off-label indications, supportive evidence and associated publications are summarized online on each therapeutic’s monograph. We included all trials listed on the monograph as evidence of clinical efficacy, but excluded retrospective studies, review papers, and trials focused only on safety. When multiple cited publications referred to the same trial, we counted it as only 1 trial.

Each supportive trial was categorized by its use of randomization, double-blinding, and comparator group. Trials using “active comparators” compared the efficacy of the study therapeutic against an available therapeutic alternative; trials using “placebo comparators” compared the study therapeutic against placebo; and trials with “no comparator” included single-arm trials and multi-arm trials comparing the study therapeutic to itself only. We classified primary end points as “overall survival” or “surrogate marker of disease.” Using a framework based on prior literature and an Institute of Medicine report,

 surrogate markers of disease included measures of disease progression and other radiologic or pathologic markers of response. We also categorized trials by the number of overall patients and those assigned to the intervention group. We recorded the median duration of treatment for each trial, calculated as the time between when the first dose of the drug was administered and when the primary end point was measured. Median duration of exposure was used as a proxy for trial duration in trials which measured time to outcomes (such as overall survival or time to progression). When the trial had multiple arms, we calculated a weighted mean of the median durations of each arm. Two investigators (NSD, JSR) performed the abstraction for FDA new drug approvals, while a third (KWS) performed the abstraction for supplementary FDA approvals and DRUGDEX off-label indications. Any abstraction for which there was uncertainty was reviewed with 2 investigators (NSD, JSR) to establish consensus.

Using descriptive statistics, we characterized the overall sample of supportive trials identified in our study. Next, we used descriptive statistics to characterize features of trials aggregated at the indication level (ie, summarizing all trials used to support each indication). We then used χ2 or Kruskal-Wallis tests as appropriate to examine differences among supportive trials and indications. In addition, we calculated the median time from the original FDA new drug approval to the approval of the first FDA supplementary indication; dates for off-label indication endorsement were not available within DRUGDEX. Analyses were performed using Microsoft Excel 2011 and JMP version 12 (SAS Institute, Inc; Cary, North Carolina). All statistical tests were 2-tailed and used a type I error rate of 0.01 to account for multiple comparisons across 5 therapeutic agent and indication characteristics.

Between 2005 and 2012, the FDA approved 37 new cancer therapeutics, of which 21 (56.8%) were pharmacologic agents and 16 (43.2%) were biologic agents (Table 1). These 37 therapeutics were originally approved for 39 indications. Fifteen (38.5%) and 24 (61.5%) of the indications were for hematologic malignancies and solid tumors, respectively, and all 24 solid tumor indications were classified as noncurative intent.

Of the 37 novel therapeutics, 17 (45.9%) subsequently received an FDA supplementary indication approval, for a total of 21 supplementary indications (Supplementary Table). Seven supplementary indications (33.3%) were for hematologic malignancies, while 14 (66.7%) were for solid tumors, and 11 of these solid tumor indications were classified as noncurative intent. Ten (47.6%) treated the same cancer type as the original FDA indication. Of these, 8 (80.0%) expanded treatment to an earlier line of therapy or changed the accompanying regimen. One of the 10 (10.0%) expanded the use of pertuzumab to the neoadjuvant setting for breast cancer, while another 1 (10.0%) expanded the use of romidepsin from cutaneous to peripheral T-cell lymphoma. A median of 920 days (interquartile range [IQR], 583-1521) passed between the original FDA new drug approval and the approval of the first supplementary indication.

Inclusion of Off-Label Indications in DRUGDEX

Of the 37 novel therapeutics, 10 (27.0%) had off-label indications included on DRUGDEX, for a total of 16 off-label indications (Supplementary Table). Six off-label indications (37.5%) were for hematologic malignancies, while 10 (62.5%) were for solid tumors, and all 10 solid tumor indications were classified as noncurative intent. Six (37.5%) treated the same cancer type as the original FDA indication. Of these 6 indications, 4 (66.6%) changed the line of therapy and/or accompanying regimen, 1 (16.6%) expanded the use of nilotinib to blastic phase chronic myelogenous leukemia (in addition to chronic and accelerated phase disease), and 1 (16.6%) expanded the use of lapatinib to inflammatory breast cancer.

Characteristics of Supportive Trials by Approval Type

In total, 50 trials supported new drug approvals, 22 trials supported FDA supplementary indication approvals, and 37 trials supported DRUGDEX off-label indications (Table 2). Of the 109 total trials identified, 53.2% (95% CI, 43.9%-62.3%) were randomized, 28.4% (95% CI, 20.8%-37.5%) were double-blinded, 53.2% (95% CI, 43.9%-62.3%) used any comparator (either active or placebo), and 19.3% (95% CI, 12.9%-27.7%) used an active comparator. Use of randomization differed by approval type (P = .01), with fewer trials supporting FDA new drug approvals and DRUGDEX inclusions using randomization than trials supporting FDA supplementary indication approvals. Similarly, use of double-blinding and any comparator differed by approval type (P = .009 and P = .01, respectively), with fewer trials supporting FDA new drug approvals and DRUGDEX inclusions using double-blinding and any comparator than trials supporting FDA supplementary indication approvals, although there was no difference in use of an active comparator (P = .57). Surrogate markers were used in 87.2% (95% CI, 79.6%-92.2%) of trials, which did not vary by approval type (P = .22).

There were significant differences in the number of total patients and intervention patients enrolled in trials supporting FDA new drug approvals, FDA supplementary drug approvals, and DRUGDEX off-label indications (total patients, 264, 414, and 86, respectively; intervention patients, 191, 214, and 63, respectively; both P <.001; Table 3). Overall, trials had a median length of 18.4 weeks (IQR, 12.4-31.6), and 30.3% of trials (95% CI, 22.4%-39.5%) lasted more than 6 months. Neither measure of trial duration varied by approval type (P = .32 and P = .06, respectively).

Characterization of Aggregate Trial Features Supporting Approved Indications, by Approval Type

FDA new drug approvals, FDA supplementary indication approvals, and DRUGDEX off-label indication inclusions were supported by a median of 1.0 trial (IQR, 1.0-1.0), 1.0 trial (IQR, 1.0-1.0), and 2.0 trials (IQR, 1.0-3.0) per indication, respectively (Table 4). All indications had at least 1 supportive trial. FDA new drug and supplementary indication approvals cited 1 or 2 trials each, with the exception of the original approval of dasatanib, which cited 6. DRUGDEX off-label indication inclusions cited a median of 2 trials (IQR, 1-3), with a maximum of 7 trials for the first-line use of bendamustine for non-Hodgkin lymphoma.

Indications were supported by total and intervention group patient populations of 383 (IQR, 178-623) and 239 (IQR, 153-397) across aggregated supportive trials, neither of which varied by approval type (P = .47 and .31, respectively). Only 38.2% (95% CI, 28.1%-50.6%) and 6.6% of indications (95% CI, 2.8%-14.5%) were supported by trials lasting more than 6 and 12 months, respectively. This did not vary by approval type (P = .95 and .81, respectively). Patient-time exposure among aggregated supportive trials was 8516 patient-weeks (IQR, 2645-15,744), which did not vary by approval type, either (P = .14).

While 69.7% of indications (95% CI, 58.7%-78.9%) were supported by a trial with any comparator (either active or placebo), only 26.3% of indications (95% CI, 17.7%-37.2%) were supported by a trial with an active comparator. Neither use of any nor an active comparator varied by approval type (P = .09 and .04, respectively). Only 18.4% of indications (95% CI, 11.3%-28.6%) were supported by a trial using overall survival as a primary end point: 23.1% (95% CI, 12.6%-38.3%) of FDA new drug approvals, 14.3% (95% CI, 5.0%-34.6%) of FDA supplementary indication approvals, and 12.5% (95% CI, 3.5%-36.0%) of DRUGDEX off-label indications. This did not vary by approval type (P = .56). No hematologic indications were supported by a trial using overall survival as a primary end point.

Our study characterized and compared the clinical trial evidence underlying the original FDA approval of new cancer therapeutics, the subsequent FDA approval of supplementary indications for these therapeutics, and the inclusion of off-label indications for these therapeutics on the Medicare-referenced compendium, DRUGDEX. Some may consider the quality of evidence supporting cancer therapeutic approvals to have limitations for informing clinical decision making, as only 70% of indications were supported by a trial using an active or placebo comparator, 30% by a trial lasting more than 6 months, and 18% by a trial using overall survival as a primary end point. However, when viewed in aggregate, new drug and supplementary indication approvals by the FDA and off-label indication inclusions by DRUGDEX were generally supported by similar bodies of clinical trial evidence.

At the individual trial level, we found some differences across approval types, potentially suggesting that the trials supporting FDA supplementary indication approvals were of higher quality, as they had the highest rates of randomization, double-blinding, and use of any comparator, and had the largest patient populations and duration of study. While these differences were not significant on aggregate, it does suggest that once a cancer therapeutic is originally approved for use, the FDA evidentiary standards for supplementary indication approvals are higher than those used by DRUGDEX for off-label indication inclusions. Further studies should examine the standards used by other compendia. If their standards are lower, from an evidentiary standpoint, the decision to allow experts outside of FDA to determine indications for which Medicare will pay for therapy may be eroding incentives to produce rigorous evidence that supports FDA supplementary indication approvals.

The strength of clinical evidence supporting these indications has important implications for informing cancer care clinical decision making. Although oncologists routinely use cancer therapeutics for indications outside those approved by the FDA or included on compendia,

 the formal endorsement of new indications through these pathways provides an additional expert review of safety and efficacy, and defines the prevailing standard of cancer care by helping secure reimbursement. The vast majority of indications in our study were only supported by trials using surrogate markers of disease, as opposed to overall survival, even though recent research has cast doubt on the mortality benefit of cancer therapeutics approved on the basis of surrogate markers.

 Moreover, no trial in our sample used patient-reported outcomes as a primary end point. Our finding that nearly one-third of indications lacked support by a trial using a comparator group further highlights weaknesses in the quality of trials used as supportive evidence. Nearly three-fourths of indications also lacked active comparator data, raising questions regarding the comparative effectiveness of these treatments, though some may have targeted disease for which no therapeutic alternative exists. FDA and Medicare should consider whether supplementary indication approvals and compendia off-label indication inclusions should be permitted on the basis of evidence that is focused on surrogate markers of disease or without active or placebo comparator; given that these medications are already approved for use by patients, higher standards could be required.

The development of new indications for already-approved therapeutics, whether it be through the approval of supplementary indications by the FDA or inclusion of off-label indications on compendia, may complement the costly process of new drug development, particularly for advanced or rare diseases. Indeed, more than half of the new indications in our study expanded therapeutics’ use to a different target cancer type, and most of them treated advanced disease. As clinical trials begin to systematically match already-approved therapeutics to different target cancers based on tumor genomic abnormalities,

 supplementary indications and compendia off-label indications will likely play an increasingly important role in defining the standard of care. However, both the FDA and compendia review processes, as well as the reimbursement policies referencing their decisions, must balance the urgency of identifying new treatments with the need to ensure drug safety and efficacy. Prior work has highlighted the FDA’s flexible standards for the approval of new cancer therapeutics.

 This flexibility may be beneficial when the cancers being treated are life-threatening and lack other effective treatment options, but it also invites criticism that the evidentiary threshold for approval is too low, especially given concerns that some new and expensive treatments offer only marginal benefits over existing therapies.

 Our findings suggest that similar uncertainties may apply to FDA supplementary indications and compendia off-label indications.

Ultimately, it is vital that the pharmaceutical industry produces data that matter for patients. Stringent postmarketing requirements could theoretically ensure that the industry conducts rigorous studies evaluating whether FDA-approved indications improve survival.

 Unfortunately, recent studies suggest that these requirements are not consistently enforced and completed on a timely basis,

 that one-third of FDA cancer therapeutic approvals based on surrogate markers have not been followed up with a study examining overall survival,

 and that less than 5% of FDA approvals based on surrogate markers, across all indications, have not been followed up with a randomized, double-blind study demonstrating improved clinical outcomes.

 Furthermore, no external regulatory mechanisms currently exist to evaluate off-label therapeutic use endorsed by compendia. Some authors have proposed expanded use of “coverage-under-evidence development” as a solution,

 in which CMS reimburses off-label therapeutic use on the condition that treated patients are enrolled into a clinical trial evaluating the therapeutic’s safety and efficacy.

Our study has several limitations. First, we only characterized the inclusion of off-label indications on DRUGDEX, 1 of several CMS-referenced compendia. While the National Cancer Comprehensive Network (NCCN) compendium probably exerts greater influence on clinical decision-making, it does not clearly label endorsed indications as off-label or list any supportive evidence, making direct comparison against FDA approvals difficult. Although the NCCN guidelines (upon which the compendium is based) may cite relevant studies in its discussion section, study citation is not uniform and systematic, precluding its use for this study.

Second, we identified supportive evidence differently for each approval type, due to the varying regulatory information available: We used medical review documents for FDA new drug approvals, drug labels for FDA supplementary approvals (as has been done previously

), and trials cited by DRUGDEX online for endorsed off-label indications. We likely identified the strongest evidence considered for FDA approvals, as we found a median of 1 supportive trial for both new drug and supplementary indications. However, we may have overstated DRUGDEX’s evidentiary standards, as we had access to the supportive evidence currently listed online, but not the evidence listed at the time an off-label indication was first endorsed.

Third, we did not assess measures of evidentiary quality other than trial design, such as effect sizes or safety, nor did we assess factors that might reasonably change the evidentiary threshold for approval, such as the severity of the target cancer and the availability of other effective treatments.

Articles

Evidentiary standards for new or supplementary cancer therapeutic indication approvals by the FDA are consistent with off-label indication inclusions on Medicare-referenced compendia.