Evaluating Off-Label Uses of Anticancer Drugs: Time for a Change FREE

In the United States, the term off-label use means using approved drugs for clinical indications that the U.S. Food and Drug Administration (FDA) has not approved. Once the FDA has approved a drug, physicians have few constraints on how they use it. In 1996 and 1997, the FDA issued guidance documents that forbade drug companies from advertising off-label uses of drugs. Shortly thereafter, the Food and Drug Administration Modernization Act loosened the strict prohibition, and case law has further freed drug companies to promote off-label indications (1). Now, the major restraint is payers' willingness to pay for the drug. In 1993, the U.S. Congress gave Medicare legal guidance for approving off-label uses of 1 class of drugs—cancer chemotherapy—within the Omnibus Budget Reconciliation Act. That system is now under intense scrutiny, which continues with a key report in this issue. This editorial will contrast the present system for off-label indications with existing alternatives and end with several proposals.

In the United States, policy constrains medical prescribing practices. States can suspend physicians' licenses when they indiscriminately prescribe addictive drugs. The FDA controls access to the market by insisting on evidence from randomized trials that a drug is more effective than placebo for at least 1 indication; it does not regulate other uses for an approved drug. Practice guidelines, which represent a professional consensus about the best drugs in specific clinical situations, often shape policies with real clout: medical insurance coverage and measures of performance and quality. Insurance coverage decisions determine what insurance will pay for. Performance and quality measures have become the standard for evaluating physician and health system performance, which can influence payment. Given their importance, do policies based on practice guidelines use an unbiased selection of the evidence? With guideline programs that use best practices, the answer is yes.

The process for assembling the evidence for practice guidelines had a strong beginning in the early 1980s. Since then, guideline development has evolved in several ways. First, many guideline programs use systematic reviews and meta-analyses, which increases confidence that they have considered all of the evidence. Second, some programs are adopting common language to describe the strength of the evidence and recommendations (2 - 3). Third, the reasoning behind the recommendations has become more transparent as more organizations publish the background review and guidance statement. Fourth, guideline programs ask outside experts to review their guidelines. Finally, programs are starting to disclose panelists' conflicts of interest. Guideline users can now be prudent shoppers, by placing greater weight on guidelines from guideline organizations that adopt these practices (4).

This background positions us to judge how Medicare evaluates off-label uses of cancer drugs. The article by Tillman and colleagues (5) provides the background to evaluate Medicare's processes (6). In 1993, Congress directed Medicare to use specific medical compendia as its source for reasonable practices in cancer chemotherapy. A medical compendium is a comprehensive listing of drugs, their clinical properties, and recommended uses (6). Under the law, Medicare could pay if one of its designated compendia listed the indication and should not pay for unlisted indications. In 2008, Congress required these compendia to describe their methods and to publish their conflict of interest policies. Recently, Centers for Medicare & Medicaid Services (CMS) officials evaluated the procedures and performance of their designated compendia. The Agency for Healthcare Research and Quality commissioned researchers at Duke University and the Evidence-based Practice Center at Tufts University to do the study. The article by Abernethy and colleagues (6) is the result.

The purpose of Abernethy and colleagues' study was to see whether the compendia provided “comprehensive, research-based, and timely information for off-label prescribing in oncology.” They found that the compendia listed widely differing numbers of off-label uses; only 1 compendium listed each off-label indication covered by the reviewers' own 2006 systematic review of 14 off-label indications. Only 2 indications were listed by all compendia. The centerpiece of the study is an evaluation of how thoroughly the compendia cited the published evidence. For each indication, the reviewers compared the number of articles cited by a compendium with the number that they found in their own systematic review. The compendia cited different articles for the same off-label indication, and their supporting materials cited far fewer articles than the reviewers' 2006 systematic review. Of 43 articles identified by the reviewers' systematic review, the compendia cited 0 to 7. The reviewers' 2008 re-review of gemcitabine found another 25 reports; the 2008 version of the compendia added few or none of these citations. In short, the compendia were inconsistent, incomplete, and out-of-date. Abernethy and colleagues' findings speak for themselves in making the case for change.

Gillick's article (7) answers the call for change. She calls for an “ethically sound, logistically efficient, and financially prudent” approach to deciding which off-label indications Medicare should pay for. An agreed-on decision-making process would be important. Gillick thinks that a strengthened decision-making process should focus attention on the most expensive and toxic cancer drugs, of which an increasing proportion are biotechnology drugs. These hard-to-manufacture agents are likely to remain expensive indefinitely, in part owing to lack of “generic” substitutes (biosimilars). Gillick points to the national coverage decision-making process at Medicare as particularly noteworthy for its transparency and rigor.

A salient feature of today's world of practice is the role of evidence-driven practice guidelines and their offspring, quality and performance measures and coverage decisions. Systematic reviews have become the linchpin of this process. In this context, clinical policies for off-label uses of drugs are a throwback to earlier, less rigorous times. This anomalous situation is in part a consequence of a system in which drugs enter the market after a costly and difficult path that provides solid evidence that a drug offers benefit for 1 indication. Manufacturers have scant motivation to shoulder the enormous costs of randomized trials to test other effects of approved drugs that have entered the market. And so evidence accumulates slowly as individual investigators—sometimes supported by pharmaceutical firms—launch trials of drugs for off-label indications. This accumulating body of evidence need not meet the FDA's high standards of clinical investigation. Our only system for evaluating off-label indications is the one described in these 3 articles: Medicare cannot pay for 1 class of drugs—cancer therapy—unless it is for an FDA-approved indication or it is listed in a drug compendium. The first of these hurdles is difficult and costly to surmount. The second seems far too low and its results are too inconsistent, as witnessed by Abernethy and colleagues.

For off-label indications, an intermediate-level hurdle seems wisest; witness the efforts to strengthen the compendia. Abernethy and colleagues' review strongly suggests that even an enhanced compendia-based system is too weak to deal with the methodological challenges of the evidence base for off-label indications. Compendia play a useful role, but Congress has assigned them a task that requires considerably greater rigor than they are accustomed to using for their main purpose. From these considerations, I conclude that we should adopt something like Gillick's proposal to use the CMS national coverage procedures, of which the salient feature is a systematic review of the evidence by experts in doing such reviews. The CMS procedures are remarkably transparent. They include posting background material on the CMS Web site, recommendations by a panel of experts who deliberate in a public meeting, and good management of potential conflicts of interest. To implement Gillick's proposal, CMS would probably need to add institutional expertise in oncology, as the Agency currently employs no oncologists.

The CMS system has many desirable attributes but, unlike the FDA, it does not shape the quality of the evidence by requiring trials that meet high standards. The CMS could use coverage under evidence development, a tool that the agency played a major role in developing (8). Under this arrangement, CMS pays for an as-yet unproven intervention—but only if the patient participates in some form of evaluative mechanism, ideally a clinical trial but sometimes an observational study with appropriate controls. Coverage under evidence development may partially offset the additional financial burden on product developers that is associated with conducting new studies for these additional indications. For this approach to work, it would also be important to develop methodological standards for evaluating off-label indications—is the same level of evidence required as for the primary indication, or might less rigorous studies be acceptable?

The articles in this issue shine a bright light on a weak point in our efforts to inform clinical practice by the best possible evidence. The present system seems wanting. We have the tools to do much better.

Harold C. Sox, MD

Editor