FEATURES:
The FDA’s Risky Risk-Aversion

By Henry I. Miller and David R. Henderson

Misplaced priorities for drug approval

During the past half-century, the stunning pace of scientific progress has yielded powerful tools for pharmaceutical development, including rational drug design, combinatorial chemistry, gene-splicing, dnasequencing, gene therapy, and genomics. These technologies have wrought a creative revolution fueled by massive public- and private-sector investment, enabling drug and biotechnology companies to produce innovative life-saving and life-enhancing medicines. Since 1950, new vaccines have virtually eliminated measles, rubella, poliomyelitis, mumps, and diphtheria from the United States and other industrialized countries. Because of new antibiotics, the child who once would have died from bacterial meningitis or a ruptured appendix now survives. Because of genetically engineered alfa interferon, the 50-year-old with a rare type of leukemia can be successfully treated. Thanks to new drugs based on monoclonal antibodies, cancer is now increasingly considered a chronic disease rather than a death sentence. And to patients with all kinds of potentially crippling health problems — from diabetes to multiple sclerosis to congestive heart failure — drugs can offer the gift of an all-but-normal daily life. The list of diseases and conditions that can be conquered or ameliorated by new drug therapies is long and growing.

Drugs often improve the span and quality of life in a remarkably cost-effective way, a fact of crucial significance not only to the individual patient but also to society as a whole: The responsible use of drug therapies lowers the total cost of health care. A study done by three leading health economists for the National Bureau of Economic Research found, for example, that the overall cost of therapy for heart attacks and depression — both of which are commonly treated with drugs — actually declined by an average of 1 percent each year from 1984 through 1991. Similarly, the costs of treatment per episode of major depression fell by 25 percent from 1991 to 1995, and studies of the impacts of a thrombolytic drug in stroke patients¹ and a new drug for migraine headache show that these treatments, too, are highly cost-effective.²

Another sign of progress is that in general new drugs confer an advantage over older ones in reducing mortality. In a study of patients who took drugs between January and June 2000, those who took newer medications were less likely to die by the end of 2002. The estimated mortality rates were directly related to time that had elapsed since approval of the drugs: For pre- 1970 drugs, the estimated mortality rate was 4.4 percent, while the mortality rates for drugs approved during the 1970s, 1980s, and 1990s were 3.6 percent, 3.0 percent, and 2.5 percent, respectively.³ Not surprisingly, perhaps, drugs are getting better all the time.

The same cannot be said about the regulatory climate, however. The potential of pharmaceutical development is profoundly affected by government interventions, particularly regulation. In this article, we focus on regulation in the United States, partly because the policies, procedures, and decisions of the U.S. Food and Drug Administration ( fda) are widely regarded as the “gold standard” (which means little more than they are the most stringent and risk-averse in the world) and because the United States is the leader in both the development of new drugs and in volume of sales.

Government giveth, taketh away

Governmental influences on drug development can be positive. In the United States, the research funded by the federal government (primarily the National Institutes of Health [ nih] but also the National Science Foundation [nsf] and several others) that attempts to better understand fundamental processes and to establish “proof of concept” often provides the scientific substrate that is essential to downstream product development. For example, recombinant dna technology, which is now ubiquitous in both basic research and drug discovery, was the product of synergistic advances in several areas of government-funded research, including enzymology, microbial genetics, and physiology. The nih also carries out some clinical trials, and even occasionally undertakes early research and development on a drug — examples of which include the anti-cancer drug Taxol and erythropoietin, which stimulates the bone marrow to produce red blood cells and is used to treat certain anemias. And as a repository of safety data on all drugs, the Food and Drug Administration possesses information that no other organization commands, which it can make available to health practitioners and the public. In addition, recent research-oriented initiatives from the fda, including the Critical Path Initiative, have the potential to help address industry-wide issues, such as the need for better animal models of human disease and for biomarkers of drug efficacy and toxicity.

However, by imposing policies that hamper innovation, government actions can exert a significantly negative influence on drug development. Policies that will affect drug pricing and drug regulation are presently the subject of considerable debate in the United States and elsewhere, and there is growing momentum for governmental actions to address the perceived problems in these areas. On one hand, as a result of concerns about the rapid growth in healthcare spending, efforts to contain the prices of new drugs (which in many countries are set by the government) are attracting increasing attention in the United States, which has long resisted such controls. On the other hand, high-profile problems (or alleged problems) of drug safety have led to calls for increased stringency of drug regulation, which would likely increase the already astronomical costs to drug developers to bring a drug to market; and that would, in turn, reduce further the number of new drugs coming on the market.

Both aspects of public policy are the focus of bills at present under discussion by the U.S. Congress (see table 1). As discussed below, some previous congressional interventions have impeded new drug discovery and development, and have created uncertainty and confusion — a situation that is anathema to corporate planning for this lengthy and risky process. The trend in regulatory policy currently has more to do with public relations than the public interest, and we believe the near-term prospects for avoiding further deterioration, let alone for meaningful reform, are dismal.

The costs of price controls

Despite the evident health benefits and cost-effectiveness of the introduction of new drug therapies, there are now more and more efforts to control the prices of drugs in the United States which, unlike Japan and Europe, has for the most part eschewed such controls.

It is therefore notable that some researchers have argued that even in the U.S., the impact of price control efforts has been significant. For example, a group at the Center for Healthcare and Insurance Studies at the University of Connecticut reported that prices fall as the government ’s share of spending on drugs increases, and that this exerts a negative effect on innovation and, ultimately, on public health. They studied U.S. data from 1960 to 2001 and found that due to the influence of laws intended to curb drug prices under government programs, “from 1992 to 2001 a 10 percent increase in the growth of government’s share of total spending on pharmaceuticals was associated with a 6.7 percent annual reduction in the growth of pharmaceutical prices.”⁴ When the government increases its share of spending, argued these researchers, pharmaceutical companies considering an investment in the development of new drugs can look forward to lower revenues, and this reduces their incentive to innovate.

Using regression analysis, the group concluded that in the absence of any government influence on drug prices, they would have been about 35 percent higher, and that the “government-induced loss of capitalized pharmaceuticalr&d expenditures was $188 billion (in 2000 dollars) from 1960 to 2001.” Applying econometric models on the productivity of pharmaceutical r&d in the United States over the same period, this translated into 140 million human life-years lost as the result of increased pain and suffering and lives shortened by the absence of new medicines. This was equivalent to more than half a year of life lost per person in America at the time.

Notwithstanding such findings, the Democratic congressional leadership now wants to change the Medicare drug benefit to require government officials to negotiate drug prices with the pharmaceutical companies — which amounts to price controls. (Under the current program, competing insurance companies individually negotiate the deals and offer coverage to the retired and disabled.) But surely no one believes that the government ’s use of its monopsonistic muscle to force drug prices to submarket levels will stimulate drug companies to develop more life-saving drugs.

The precedent of the Veterans Administration health care system suggests that another likely outcome of such compulsory negotiations will be that the government will decide not to cover certain drugs under Medicare at all. Only 19 percent of drugs approved by the fda since 2000 are listed on the va formulary, and less than 40 percent of drugs approved in the 1990s are listed.

The critical gatekeeper

The fda is arguably the most omnipresent regulatory authority in the United States. It has responsibility for more than a trillion dollars ’ worth of consumer products annually, ranging from condoms and x-ray machines to drugs, vaccines, pregnancy home-testing kits, and artificial sweeteners. Its role as the nation ’s regulator of drugs makes the agency the primary gatekeeper between the developer and the marketplace. The agency ’s enabling statutes — the Federal Food, Drug and Cosmetic Act and the Public Health Service Act (which mandates regulation of biological drugs) — are not highly prescriptive or detailed. They permit government regulators great latitude to apply scientific knowledge and common sense to oversight, but this latitude also frees regulators to decide how much power and discretion they should exercise. Not surprisingly, their tendency has been consistently toward more power and discretion.

The fda evaluates and approves drugs, but it does not discover or test them. That is done by a sponsor, usually a private pharmaceutical company. This point cannot be overemphasized: It is not bureaucrats who bring life-saving new therapies to consumers, but private companies and research institutions. The process of research and development is difficult and uncertain in the best of circumstances, so the relationship between sponsor and regulator is critical to ensure that consumers reap all the health benefits of drugs in a timely manner. The objective of regulation should always be to impose only that degree of scrutiny and constraint that is necessary and sufficient.

Development of a new drug begins with preclinical investigations: in vitro screening for a desired chemical or biological activity or characteristic, followed by testing in laboratory animals to determine therapeutic activity and possible toxicity. These preclinical investigations generate preliminary knowledge about the pharmacological and toxicological properties of the agent. If they yield promising results, they are followed by clinical testing in humans over a period of years. Two obligatory applications to the fda are made as part of the process of drug development. Before embarking on the first phase of clinical testing, the sponsor must submit an application called an Investigational New Drug, or ind, filing. The agency then monitors the testing through periodic reports, inspections, and audits. When clinical testing has progressed to a point where the drug sponsor is satisfied that the drug meets the regulatory standards of safety and effectiveness for a specific use, it submits the second mandatory application, the New Drug Application, or nda, seeking approval to market the drug.

The fda often promulgates new requirements — or new, seemingly arbitrary, interpretations of old requirements — with little regard for the costs to patients and regulated industry. One measure of these increasingly rigorous and expansive (and expensive) requirements is the growing number of clinical trials during drug development. Since 1980, the average number of clinical trials conducted to support an nda has more than doubled, from 30 to about 70. Likewise, the average number of patients required to support an nda has almost tripled, from 1,576 in the late 1970s to 4,237 in the mid- 1990s, and the number of medical procedures performed during the clinical testing rose 61 percent from 1992 to 1997. ⁵ These costs add up. Had they increased at the pace of inflation, the average costs of new drug development would have risen from $ 231 million in 1987 (in 1987 dollars) to $318 million in 2000 dollars 13 years later. But by 2000, the average cost of development for a new drug had increased to $802 million (in 2000 dollars), and to over a billion dollars by 2006, according to Dr. Joseph A. DiMasi, director of economic analysis at the Tufts University Center for the Study of Drug Development. These estimates are “average capitalized cost” per product. (The capitalized cost calculates both the cash outlay to develop a drug and the opportunity costs to investors of having funds expended for years before they realize any return.) That research group also found that while costs have increased in inflation-adjusted terms for all phases of research and development, the increases were particularly large for clinical trials. The inflation-adjusted annual growth rate for capitalized clinical costs ( 11.8 percent) was more than five times as high as that for pre-clinical research and development.

Chasing these skyrocketing costs, between 1995 and 2006 U.S. research-based drug companies’ expenditures on clinical trials almost tripled, to $43 billion. A 1994 study at Duke University revealed that during the period 1980–84 — when r&d was far less expensive than currently — only three of every ten marketed drugs produced revenues that covered their development costs, and that the 20 percent of products with the highest revenues generated 70 percent of total revenues.⁶  Because the expenses of regulation have so inflated the cost to bring a drug to market, companies have a strong incentive to focus their research on products that will likely be blockbusters, while they are discouraged from researching drugs that could be of great benefit to small populations. These two factors reduce the robustness of the development pipeline.

Why so expensive and slow?

The fda constantly raises the bar for the initiation and progress of clinical testing of new drugs. For example, in the past few years fda officials have arbitrarily and unexpectedly directed clinical investigators to begin trials at inappropriately low dosages; limited approval of Phase i studies only to single-dose, instead of dose-ranging, studies; demanded unnecessary, invasive procedures on patients; and even required that foreign trials be completed and the results submitted before the U.S. trials could begin.

The 1999 death of a patient in a gene-therapy trial at the University of Pennsylvania offers an example of overreaction by the fda and its consequences. Although the cause of the multi-organ failure in the teenaged patient had not been determined, the fda tightened manufacturing standards for academic investigators, inaccurately but publicly accused the clinical investigators of various kinds of mistakes and misconduct, shut down all of the university ’s gene-therapy trials, and even halted unrelated trials being performed by a drug company using a similar preparation. The fda’s precipitous actions, following a single unexplained death during a trial, cast a pall on gene-therapy research throughout the country and discouraged commercial support for the field. In 2002, just as the field was beginning to recover, the fda struck again, precipitously halting approximately 20 percent of all ongoing gene-therapy trials — many for life-threatening diseases — after two children successfully treated in a French study for a potentially lethal immunodeficiency developed leukemia.

The fda’s constant raising of the bar for approval, tendency to overreact, and anxiety about new technologies has made the drug development process in the United States the lengthiest in the world, and it has grown longer over time. According to the Tufts Center, since the 1960s the total time required for drug development — from synthesis or discovery in the laboratory to delivery to the patient — has almost doubled, from 8.1 years to 15.2 years.⁷

In April 2007, the FDA announced what appears to be a landmark decision. Although the law requires that in order to be marketed, a drug must be shown to be safe and effective, by denying approval of Merck ’s new drug, Arcoxia, a cox-2 inhibitor for the relief of arthritis pain, the fda appeared to introduce a third criterion: that Arcoxia was not sufficiently superior to existing drugs to merit approval. Robert Meyer, director of the fda office that oversees arthritis drugs, said the agency’s advisory committee had sent a clear message that “simply having another drug on the market . . . didn’t seem to be sufficient reason” for approval.⁸But that reasoning ignores the question whether the drug is simply “another drug on the market” that is no better than many alternatives; in fact, cox-2 inhibitors have been shown to have less gastrointestinal toxicity than many other commonly used pain relievers, and there is preliminary evidence that members of this drug class may prevent malignancies or delay their recurrence.

Proving that a drug is better than existing drugs is harder than proving that it is simply safe and effective, and many drugs useful for some patients will founder if this new third criterion is widely implemented. That would reduce competition in the drug market, with correspondingly higher prices. Whatever one thinks of regulation for safety and efficacy, surely we should not have an fda that purposely discourages competition.

Slow equals safe?

The reputation of the fda’s regulatory regimes as the gold standard might imply to some that fda-approved products are the safest anywhere and that a ponderous, expensive regulatory regime is the price that must be paid for a high level of safety. However, that conviction hardly justifies a system that ignores the costs, both in money and lives lost, of inefficient, unnecessarily burdensome regulation. Public health is harmed when potentially beneficial products are delayed, abandoned, or never tested at all.

Another myth about the supposedly inevitable link between slowness and safety in drug regulation is that the more stringent regulatory policies of the United States afford greater protection from drug-related adverse effects, such as toxicities that are unknown at the time of marketing approval but emerge after extensive use of the drug. Although data are difficult to obtain for comparisons of safety discontinuations (removals from the market) between the fda and its European equivalent, the European Agency for the Evaluation of Medicinal Products ( emea), there have been comparisons between the United States and the United Kingdom (whose approach is similar to that of the emea). Norwegian researcher Olav M. Bakke and his colleagues found that while the number of safety discontinuations in the United Kingdom was larger than that of the United States between 1964 and 1983, the overall number of drugs approved in the United Kingdom was also larger. As a result, the numbers of safety discontinuations as a percentage of total new drug introductions in each country were similar — approximately 4 percent in the United Kingdom versus about 3 percent in the United States. They came to similar conclusions in a more recent study of safety discontinuations in the United Kingdom, the United States, and Spain between 1974 and 1994.⁹

Risk-averse

Over the past several decades, the pendulum has swung concerning what ails pharmaceutical regulation. Thirty years ago, the concerns were primarily about “drug lag” — slow reviews and approvals by the fda that put Americans at a disadvantage to consumers in other countries; but in recent years, concern has shifted primarily to what might be called “drug leap” — allegations of hurried approvals, insufficient attention paid to drug safety, and too close a relationship between regulators and industry. Several highly publicized events have heightened public concern about drug safety during the past few years: inadequate warnings on the labels of anti-depressants, the discovery of previously unknown adverse reactions to nonsteroidal anti-inflammatory drugs ( nsaids) and the multiple sclerosis drug Tysabri, and the life-threatening, multi-organ failure in volunteers almost immediately after receiving the first dose of a drug ( tgn1412) in a Phase i trial in England.

However, contrary to these perceptions, drug regulation in the United States in recent years has actually become progressively more risk-averse, as the fda has steadily raised the bar for the initiation and progress of clinical testing of new drugs. Recent criticism from Congress, the media, and others regarding drug safety has caused an already risk-averse agency to become even more conservative and defensive in its decision-making. In September 2006, Genentech announced that approval of its colon cancer drug Avastin for the treatment of breast cancer would be delayed at least a year because of requests from the fda for additional data. The company said that regulators appeared to be increasing the stringency of requirements for certain types of clinical trials and had arbitrarily demanded that its trials be “audited and summarized” in a way different from that earlier agreed upon.

Regulators’ moving the goalposts in the middle of the game is particularly vexing for drug developers, and Avastin is not an isolated case. Another recent and particularly problematic example involves Somaxon Pharmaceuticals ’ testing of an already-approved drug, doxepin, for a new indication. The drug, approved for the treatment of depression since 1969, is being tested in very low doses for use as a sleeping pill. The fda initially assured the company that it could begin human clinical trials without first doing animal tests because of doxepin ’s long history of use in people and because Somaxon was using a dose smaller than one-tenth that used to treat depression. However, in May 2006, after having completed several clinical trials, Somaxon was meeting with the fda to discuss the submission of a New Drug Application, when regulators unexpectedly asked for a full battery of testing in animals — a step that is usually considered to be pre-clinical. What is the logic of animal testing for an almost 40-year-old drug that is undergoing trials for a new indication, and at a far lower dose than is normally used?

In addition, a number of drugs previously granted marketing approval in Europe have received “approvable” — instead of “approval” — letters from the fda, meaning that additional data are required before the drug can be marketed. These include Sanofi-Aventis ’s Acomplia for weight loss and cessation of smoking, nps Pharmaceuticals’ Preos for osteoporosis, and Encysive Pharmaceuticals’ Thelin for pulmonary hypertension.

A further sign of greater risk aversion is the increasingly aggressive use by regulators of post-marketing “Risk Minimization Action Plans.” These Riskmaps can include the submission of additional safety information, including larger studies to screen earlier for relatively rare potential adverse reactions, greater restrictions on distribution and advertising, and so on. In March 2005, for example, the Riskmap that accompanied the fda’s approval of a diabetes drug called Symlin prohibited the company from conducting any direct-to-consumer advertising or journal advertising for one year following approval, and also restricted promotion primarily to physicians who specialize in diabetes management and who are supported by certified diabetes educators. A 2006 Washington Legal Foundation Legal Backgrounder offered persuasive arguments that this kind of regulatory expansionism is unwise, outside the fda’s statutory authority, and unconstitutional. It characterized the fda’s ban on advertising as “nothing more than an effort that seeks to keep people in the dark for what the government perceives to be their own good — a concept the Supreme Court has warned courts to view with skepticism.” Moreover, the backgrounder points out that thefda’s ban on the advertising of Symlin conflicts with the agency’s own guidance to industry, which cites “promotional techniques such as direct-to-consumer advertising highlighting appropriate patient use or product risks ” as an example of how to use [a Riskmap].¹⁰

This sort of inconsistency from the fda, which is particularly problematic for companies that have invested substantial sums in research and development, is not uncommon. Regulators ’ recent actions on the post-approval risk management of two drugs, Tysabri and Rituxan, are illustrative.

In late 2004, Tysabri was approved for multiple sclerosis, a common and debilitating autoimmune disease that affects the central nervous system. The drug ’s testing in clinical trials yielded impressive results — the frequency of clinical relapses fell by more than half — and induced the fda to grant accelerated approval. In early 2005, however, with several thousand patients already being treated with Tysabri, three patients were found to have contracted progressive multifocal leukoencephalopathy ( pml), a rare and often fatal neurological disorder caused by a virus. (Because the drug suppresses certain components of the immune response, regulators, clinicians, and the product ’s developers had from the beginning been sensitive to the possibility of infections as a side effect.) Immediately, the manufacturers voluntarily halted production and distribution and withdrew Tysabri from the market. After the analysis of new safety data, an fda advisory committee recommended Tysabri’s return to the market with revised labeling, but the fda went far beyond adding to the labeling more prominent warnings about the drug’s side effects (which would arguably have been sufficient) and insisted instead on a complex Risk map that imposes onerous restrictions on the use of Tysabri. They include limited distribution and additional education and monitoring requirements for patients, prescribers, pharmacies, and infusion centers.

That brings us to Rituxan, a treatment for rheumatoid arthritis and certain kinds of lymphomas. Like Tysabri, it acts by suppressing elements of the immune system and also has been linked to pml; there have been 23 confirmed cases of pml in patients receiving Rituxan for the approved indication of non-Hodgkin’s lymphoma and, most recently, two in patients being treated experimentally for systemic lupus erythematosus (lupus). But in contrast to Tysabri, Rituxan has never been subject to a Risk map. And in spite of the new cases of pml in patients with lupus — and the fact that Rituxan also is under consideration for treatment of ms — the fda was content merely to update the package insert for Rituxan.

Leaving aside the question of whether Rituxan should be subject to a more restrictive Risk map or whether Tysabri deserves a less restrictive one, the fda’s inconsistency sends mixed signals and creates uncertainty, the bane of patients, physicians, and drug companies alike.

Whose life is it, anyway?

Statistical significance, often expressed as a “p-value,” is an estimate of the level of certainty of a set of observations. Thus, a p-value of 0.05 means that there is a likelihood of five percent that the observed finding is due to chance rather than to a genuine effect. Although arbitrary, the bar is typically set at 5 percent — in other words, p must be less than 0.05 (p<0.05) — as “proof” for most phenomena in the realms of medicine and science. However, for a patient about to succumb to an illness or to lose his brain or kidney function, a different level of certainty, or proof, may be more appropriate. Also factored into patients ’ and physicians’ decisions will be the magnitude of potential benefit, quality of life, safety, convenience, and perhaps cost. Putting it another way, a patient might be willing to accept a 20 percent risk that the observed benefit from a treatment is not genuine but due to chance (p= 0.20) if that treatment was (relatively) safe and could potentially prolong his life. In fact, patients make such judgments all the time when they decide whether to have angioplasty or open-heart surgery for coronary artery disease; lumpectomy or more radical surgery for breast cancer; or whether to opt for radiation therapy, surgery, or drug treatment for prostate cancer. In these cases, patients make decisions based on advice from their physicians and their own analysis and worldview. Why shouldn ’t federal regulators grant patients and their physicians greater autonomy to decide the appropriate levels of risk and uncertainty for investigational drugs?

Advances in molecular biology, immunology, chemistry, nuclear medicine, and other disciplines have ushered in a new era of potential diagnostic and therapeutic products for a wide range of diseases. “Targeted therapies” have now touched many areas of medicine including oncology, infectious diseases, and metabolic disorders, among others. In some cases, the use of molecular diagnostics, improved imaging, and other approaches has led to a re-examination of conventional clinical trial paradigms. The potential for new technologies to enhance drug development and clinical research programs spurred the fda Critical Path Initiative and the nih Roadmap for Medical Research.

But too often a regulator unfamiliar with a new technology is a fearful regulator; and a fearful regulator tends to: ( 1) doggedly apply old paradigms to novel situations; (2) slow down every phase of clinical development; and (3) require unnecessary testing, in order to provide himself with cover should anything go wrong. Such inflexibility drives biotech and pharma companies to stick with the established methods and approaches and discourages innovation. It is a particular burden on small and midsize biotech companies, the principal sites of innovation. In spite of the obvious need for regulators to adapt to new and changing technologies, often the fda has forced biotechnology and pharmaceutical companies to fit their product development efforts into existing, well-established regulatory paradigms. (An apt, classic cartoon depicts two scientists at the lab bench. One of them, swirling a conical flask, says to the other, “Well, it looks as though we’ve finally done it — developed a drug that will confer immortality! The trouble is, it will take forever to test it.”

Although the cytostatic drug Nexavar, which is now approved for metastatic renal cell cancer (a rapidly lethal disease), had a clinical tumor response rate of only 2 percent (compared to zero for placebos) in its initial randomized trial, it delayed tumor growth as measured by an improvement of a parameter called “progression-free survival.” But because a statistically significant improvement for the pre-specified criterion of survival was not observed, the fda required an additional randomized clinical trial, which delayed approval by more than a year. (Ironically, during the delay, as additional data became available from the initial trial, it became evident that Nexavar did, indeed, improve survival. However, this made it unethical to continue the second trial as planned, and patients who experienced tumor progression while on placebo were permitted to “cross over” and receive the drug.)

Many new targeted therapies are difficult to assess with traditional clinical trial designs or endpoints. For example, a drug intended to arrest or eliminate cancer by boosting the patient ’s immune system might be most efficacious in early-stage cancers when the tumor load is low and the patient is not debilitated. Thus, if the usual fda-preferred clinical design called for recruiting and treating only patients with advanced cancers, one would expect to see reduced efficacy. Such a possibility argues for greater flexibility in both clinical trial design and selection of the criteria for “efficacy,” but instead the fda has been risk-averse and defensive. Many developers of cancer drugs, in particular, characterize fda reviewers and managers as “lazy” and inadequately prepared for meetings with industry scientists and insensitive to the needs of dying patients.

Although many of the new targeted therapies are difficult to assess with traditional clinical endpoints — especially if they act through novel mechanisms of action — they may produce unequivocal benefits to patients. One that came to light in May 2007 is a new immunotherapeutic drug, Provenge, which, in spite of apparent clinical benefit, was not approved because of its failure to meet pre-specified endpoints at conventional measures of statistical significance.

The case of Xcytrin, an investigational drug with a novel chemical structure and mechanism of action evaluated in patients with lung cancer that had metastasized to the brain, also is illustrative. Brain metastases, which affect 100,000 patients annually in the U.S., can cause horrific neurological problems, and there is no approved drug therapy. The standard treatment is whole brain radiation, which has been in use for over 50 years and is toxic and of only marginal benefit. In order to test Xcytrin, its manufacturer developed and validated a novel and innovative endpoint that measures neurological and neurocognitive function. The pivotal clinical trial failed to demonstrate statistically significant improvement in the primary endpoint (median time to progression of neurological signs and symptoms, which was 15.4 months on drug vs. 10.0 months for the control), although there was statistically significant improvement in several secondary endpoints and in a pooled analysis involving 805 patients. (The failure to reach significance on the primary endpoint was largely because of noncompliance with the clinical protocol by two foreign centers, of the dozens that were involved with the study.) The fda’s response to the company’s New Drug Application was to dismiss it with a “refusal to file.” In other words, regulators declined to evaluate the arguments for clinical benefit altogether, eliminating even the possibility that they could have judged the drug “approvable.” (Approvable means that safety and efficacy would be demonstrated with the submission of certain additional data, or that the drug would be approved, but with a requirement for certain post-marketing clinical trials.)

Yet another example is Eloxitan (oxaliplatin), a drug widely used for the treatment of metastatic colorectal cancer, the third most common cause of death from cancer in the United States. An nda for this drug was filed in February 1999, with two large randomized trials which showed that the drug, when added to standard therapy, had clear anti-tumor activity (improved tumor response and progression-free survival). However, the fda refused to approve the nda because the survival difference between the treatment arms, although superior for Eloxitan, was not statistically significant (p= 0.12). The fda required the sponsor to conduct another study, which led to accelerated approval of the drug in August of 2002. Subsequent clinical studies have conclusively established that Eloxitan saves lives, but the three-and-a-half year delay made the drug unavailable to almost 200,000 patients with a lethal disease: Approximately 52,000 U.S. patients are diagnosed annually with metastatic colorectal cancer. ¹¹Sam Kazman of the Competitive Enterprise Institute has calculated the lives lost because of regulatory delays for other drugs. ¹²

The Institute of Medicine’s role

A report¹³focused on drug safety released in September     2006 by the U.S. Institute of Medicine (iom) of the National Academies — a nonprofit organization that provides analysis on matters of biomedical science, medicine, and health — represents the culmination of perceptions that systems for drug safety needed to be fundamentally reconsidered. The report makes sweeping, radical recommendations for change that are not only unlikely to remedy the fda’s current shortcomings, but will, if instituted, actually make the agency even more risk-averse, further inflate the costs of drug development, discourage innovation, reduce the number of drugs emerging from the r&d pipeline, and have the net effect of compromising public health.

The proposed recommendations include what would amount to the introduction of only limited approvals for new drugs during a lengthy period after approval (with special warning labels and restricted direct-to-consumer advertising), greater use of advisory committees, a new registry of clinical trials, and additional resources for the fda.

Furthermore, under proposals for the disclosure of information about clinical trials that have come from both the iom report and Congress, the fda would force drug developers to include details such as timelines, milestones, and endpoints, which drug companies (and current federal law) consider to be proprietary information. Such mandatory disclosure would give competitors an early look at sensitive data, thus expropriating from drug developers some of their proprietary information. The prospect of “competition via forced disclosure” would further reduce the incentive to develop new drugs.

The fda anticipated the calls for reform with initiatives of its own to increase the surveillance and reporting on the safety of drugs. These include the creation of a Drug Safety Board, whose objectives are “to provide oversight and advice to cder leadership on the management of important drug safety issues and to manage the flow of emerging safety information to healthcare professionals and patients ” and a number of projects under the rubric of the Critical Path Initiative. The latter include much needed improvements in the fda’s Adverse Event Reporting System (aers) and research on animal models for human disease, cardiovascular biomarkers, and the genetic basis of adverse events.

Other initiatives, such as the agency’s Drug Watch program, appear to have more to do with public relations than public health. In May 2005, the fda published draft guidance on the Drug Watch program, which will make “emerging safety information” publicly available. According to the fda, this program

The fda  further notes in the same document that Drug Watch is intended

It is difficult to predict what physicians and other health-care providers — let alone members of the public — will do with such preliminary data, which are available on the agency’s website. There is a difference between indiscriminate data and useful information, and Drug Watch seems destined to provide far more of the former than the latter. Then- fda Deputy Commissioner Scott Gottlieb addressed that point: “Information that could influence clinical medical practice needs to be made available more quickly, and more widely, after it has gone through a deliberative scientific process that firms up its meaning and the magnitude and the veracity of its conclusions. ”¹⁵ But Gottlieb himself characterized the information that would appear on Drug Watch as data “still un-scrubbed by scientific rigor.”

Moreover, given the current desire at the fda for ways to demonstrate a commitment to drug safety, and the difficulty of proving a negative, how will a “suspect” drug ever be able to clear its name and get off the Drug Watch list? Surely, it would be far more constructive to update product labeling continually and rapidly — which can now be accomplished using the fda’s website and e-mails to healthcare providers and consumers — once regulators are past the stage of merely “attempting to assess the meaning and potential consequences of emerging safety information ” and have actually “determined the significance” of the information.

In January 2007, the fda announced a plan to perform a comprehensive assessment of the safety of some new drugs within 18 months after their introduction, and to issue a “report card” on their performance. Although this may sound plausible, it is inconsistent with the data, cited above, showing that in fact newer drugs confer an advantage over older ones in reducing mortality. And in February 2007, in reaction — many observers would say overreaction — to proposed legislation, the fda introduced new restrictions on members of advisory committees, which are comprised of outside experts. Committee members who receive money from a drug or device maker would be barred for the first time from voting on whether to approve that company ’s products; and if they receive more than $50,000 from a company or a competitor whose product is being discussed, they would no longer be allowed to serve on the committees. This eliminates those experts who are likely to possess the greatest expertise about the subjects under discussion; disclosure of potential conflicts of interests and recusal when appropriate would have been far better than the new, rigid, one-size-fits-all automatic exclusions and disqualifications.

Congressional Confusion

During the past few years, Congress has weighed in with a variety of proposals ostensibly intended to enhance drug safety, including the creation of an independent agency concerned narrowly with drug safety. Most of these proposals are ill-conceived.

An agency concerned only narrowly with drug safety would place us in the realm of regulation according to the flawed “precautionary principle,” which erects high barriers to new products, processes, and technologies, whatever their potential benefits. Were this new agency to be created, it would effectively create within the fda an anti-drug entity with strong incentives to argue for the nonapproval or withdrawal from the market of drugs that have significant side effects even if they offer huge net benefits. (We have seen this already from certain factions within the agency who have publicly expressed anti-drug views that conflict with fda policies and decisions.) At the least, the net effect would be to make the agency ’s drug evaluators even more defensive and risk-averse. Fewer drugs would be approved and more would be subject to dubious withdrawals from the market; and with these increases in development “failures,” the average cost to bring a new drug to market would rise even more, diminishing further the number of new drugs developed.

Congressional meddling in the conduct of clinical trials has been particularly ill-advised. For example, in response to a long history of lower representation of women and minorities in clinical trials (often for good reason, such as reluctance to expose women of childbearing age to experimental drugs), in 1993 Congress passed a law requiring their inclusion in federally funded clinical trials in numbers “sufficient to provide for a valid analysis of any differences . . . in response to drugs, therapies and treatments. ” If this were enforced strictly, it could raise the numbers of subjects in clinical trials — and the attendant costs — substantially. Moreover, although one-size-fits-all requirements of that kind might seem to offer clarity, they are fraught with all sorts of challenges: Must every minority group be represented? Eskimos? Pubescent girls? Pregnant women?

Although regulators have not enforced such quotas, their enshrinement in law remains a concern: The strict interpretation of legislation does not always permit the interposition of common sense, and the requirements could easily be enforced in the future as a result of litigation or congressional oversight hearings.

Some congressional machinations have been remarkably subtle. An example is the reauthorization of the State Children ’s Health Insurance Program, created in 1997 to cover children from lower-income families who make too much to qualify for Medicaid, which is up for renewal in late 2007. Scott Gottlieb raises the specter of a land-mine buried in the legislation: “Tucked into page 414, section 904 of the House bill is a provision to spend more than $300 million to establish a new federal ‘Center for Comparative Effectiveness’ to conduct government-run studies of the economic considerations that go into drug choices. ” Gottlieb fears that its mission will be to “arm government actuaries with data that proponents hope will provide ‘scientific’ proof that expensive new drugs are no better than their older alternatives,” and that it will strive “to maintain just enough credibility around the conduct of these trials to justify unpopular decisions not to pay for newer medicines. ” He concludes: “The political cover offered by government-directed research will surely help when it comes time to impose unpopular limits on prescribing. ” ¹⁶

What is the price of safety? Defenders of the present risk-averse system argue that it is lower efficiency. But this is a false tradeoff. High standards of safety and greater efficiency could be achieved if we were to reform fundamentally the way drugs are regulated. If we could end regulatory excesses (especially those that are politically motivated) and introduce competition into regulatory oversight, more patients would benefit from a greater number of drugs made available to them in a timelier way. One model for certification is Underwriters Laboratories, which evaluates thousands of categories of consumer products from lighting fixtures and flame retardant chemicals to bulletproof glass: A private, nonprofit entity crafts standards and certifies compliance.

A recent study of the effects of the Prescription Drug User Free Act (pdufa) — which requires drug sponsors to pay fees to the fda for the timely review of New Drug Applications (ndas) — not only concludes that expeditious reviews do not compromise drug safety, but also finds evidence of net benefits to public health of rapid reviews. The study assessed the tradeoff between the benefits of getting “good” drugs on the market more quickly and the advantages of preventing “bad” drugs from reaching the market. The researchers estimate that by shortening the time the fda took to approve drugs (by providing more reviewers and committing the agency to certain statistical milestones), pdufa significantly improved the situation of both patients and drug producers. According to their calculations, this benefit, or “social surplus,” was in the range of $18 billion to $31 billion. Converting these economic gains into equivalent health benefits, they found that the faster access of drugs to the market made possible by pdufa’s additional resources and reforms saved the equivalent of 180,000–310,000 life-years.

The researchers also took into account the possibility that the faster approval times mean more unsafe drugs on the market. However, they found no statistically significant difference in the proportion ( 2 to 3 percent) and timing of withdrawal of drugs that were approved pre- and post-pdufa. To investigate further the relationship between faster approval times and drug safety, they assumed that without pdufa, none of the drugs actually withdrawn for safety reasons would ever have been allowed in the first place — a 100 percent ability of regulators to detect what will turn out to be hazardous products. Even with this extremely unrealistic assumption, they found that only 18 to 31 percent of the benefits — 55,600 life-years — were lost due to speedier approval.

We would offer several caveats about the presumptive benefits of pdufa, which is still in effect and is widely expected to be reauthorized, with higher user fees, sometime in 2007. First, and most important, the speed of approval measures only the relatively brief timespan between the submission for review of a New Drug Application ( nda) or a Biologics License Application (bla) and the fda’s approval. However, the majority of the pre-approval time during which a drug is under the fda’s jurisdiction is the Investigational New Drug (ind) phase, the much longer interval between the initiation of clinical testing and the submission of the nda or bla; the ind phase may last a decade or more. Second, given that the benefits of speed in the above study were measured relatively accurately but the costs of speed (that is, the benefits of slowing down) were explicitly overstated, a reasonable conclusion is that even with pdufa, the time between submission of the application and its approval still is too long. Finally, the same benefits of more rapid approvals might well have been obtained if the additional resources had been provided to the fda by congressionally appropriated funds instead of by pdufa user fees, which are nothing more than a discriminatory tax on a single industrial sector.

Potential for reform

Although any system of drug oversight should preserve a reasonable degree of confidence of product safety and efficacy for the indications listed on the label, such assurance can obtain not only through federal regulation but also from the evolving interplay among industry, government, academia, medical practice, insurers, and the courts. Since the current framework for the regulation of new drug development was put in place more than four decades ago, basic and clinical research techniques have advanced, the public ’s sophistication and awareness about drugs have grown, and the media have become both more aggressive and more attuned to health issues. In addition, pharmaceutical marketing, tort case law, and the system for cost reimbursement for medical treatments all have evolved. These factors have altered the nature of manufacturer-physician-insurer-patient relationships profoundly and superimposed new layers of scrutiny and control upon the fda’s evaluation, approval, and monitoring activities.

There are many more professional, full-time, career clinical researchers now than in the 1960s, and also more proficient for-profit clinical research organizations that design and perform clinical studies for drug sponsors. Moreover, there are additional institutional safeguards to patients, such as corporate procedures for carrying out and overseeing clinical trials and conducting post-marketing drug surveillance, and government-mandated institutional review boards ( irbs) in clinical research institutions. The reporting of adverse events has been vastly improved, and the science of pharmacoepidemiology systematically examines adverse events in exposed populations.

But the most profound changes have resulted from the evolution of various nongovernmental entities into de facto drug-vetting, standard-setting organizations. The newest and most potent of these are managed-care organizations, which exercise their influence through large-scale purchasing, construction of formularies, monitoring, and drug utilization review. The ability of physicians who practice within organizations like hmos to prescribe is increasingly affected and constrained by computerized systems that perform overall integration of the medical record for case management. A physician can be prevented from prescribing medication if, for example, according to computerized monitoring of his decisions, the drug is inconsistent with a patient ’s listed diagnosis; excessive in dose, frequency, or length of administration; likely to interact dangerously with another medication the patient is taking; or even judged not to be cost-effective compared to alternative drugs. Drugs can be omitted or removed from formularies if cheaper alternatives are available or if they are deemed to be nonessential because they treat “nondisease” conditions like baldness or skin wrinkles. In short, the hmo or insurer has become a third gatekeeper — along with the fda and the prescribing physician — between the manufacturer and the patient.

These various influences work in concert to protect the integrity of the clinical trials and, after approval, to assure product safety, efficacy, and effective post-marketing surveillance. The operation of these factors diminishes the relative importance of the fda as the protector of the patient. But instead of a diminution of the fda’s power, responsibilities, and requirements, we have seen quite the opposite; and, as discussed in the previous section, the pendulum seems currently to be swinging even farther in that direction. Regulators now make decisions defensively; in other words, in order to avoid approvals of harmful products at any cost, they tend to delay or reject new products. That is bad for public health, for drug developers, for consumers ’ freedom to choose, and for patients’ well-being. The fda is not unique in this regard. All regulatory agencies that perform pre-market evaluations are subject to criticism if dangerous or questionable products make it to market (often even for products that offer net benefits), but actions that keep beneficial products from reaching consumers seldom receive attention, let alone condemnation.

Routes to reform

Meaningful change in the performance of the fda will require legislative action, but recently Congress’s interest in drug regulation has taken the form of politically motivated investigations of supposed under-regulation or insufficient attention to product safety. Ironically, it is Congress ’s failure to carry out its oversight and legislative role responsibly that has permitted the risk-averse culture at the fda to become progressively worse and more entrenched. So, what could be done to address this situation?

As discussed above, the fda has already begun to improve the post-marketing surveillance of adverse reactions to drugs but has yet to address the culture of risk aversion that unnecessarily delays product approvals. The fda could contract out product reviews — which has been highly successful in pilot programs — as well as the accumulation and analysis of safety data, and Congress could create extra-governmental mechanisms for product oversight. For example, the regulation of medical devices (and many other consumer products) in the European Union relies heavily on product standards but normally does not involve government regulators directly in product review. For low-risk devices, manufacturers themselves are allowed to certify that their products meet the necessary standards. For higher-risk products, manufacturers must obtain third-party review from private-sector, profit-making entities — “notified bodies” — that test products, inspect manufacturing systems, and ultimately verify that eu standards have been met. Another apposite model is the Nationally Recognized Testing Laboratories in the United States, the prototype of which is Underwriters Laboratories.

In addition, the fda’s senior and mid-level managers must be made more accountable — especially for scientifically dubious policies and needless delays in getting new drugs, vaccines, and medical devices to patients who need them. One way to achieve that would be to create an independent, strong ombudsman mechanism that could impose negative sanctions on civil servants who are incompetent, indolent, or insubordinate. However, all of the newly introduced checks on the fda’s drug approvals — such as the Drug Safety Board and the Drug Watch program — and more recent proposals along these lines are asymmetrical, in the sense that they address primarily concerns about safety, narrowly defined, but not the lost benefits of drugs needlessly delayed or abandoned.

Conclusions

By means of policies that include the funding of research, protection of intellectual property, establishment of price controls, and pre-market and post-approval regulation, governmental influences on the discovery, development, and marketing of new diagnostics and drugs are profound. Regulatory policies and decisions are especially potent, spelling life and death for patients and companies alike. Those who disagree with regulators ’ actions are largely without recourse; the courts consistently defer to the presumed disinterested expertise of the regulatory agencies.

Many lives would be improved and saved by a more efficient system of oversight. Why, then, is there no sense of urgency, no lobbying for regulatory reform from any prominent quarter or interest group? The reasons are complex. Would-be reformers often are accused of being beholden to drug manufacturers and of plotting to deregulate, which is seen as a conspiracy among political reactionaries and free-market fanatics who favor commerce over the protection of public health. Paradoxically, even the largest pharmaceutical companies fail to lobby for reform, either individually or through their trade associations. How can that be?

First, drug companies continue to be profitable. For them, the vast expense of regulation is simply part of the cost of doing business. Their own massive regulatory-affairs bureaucracies are, to some extent, special interests convinced that they would not be well served by less regulation. Moreover, up to a certain point, excessive, expensive, inflexible regulation is advantageous to larger companies because, due to regulatory affairs economies of scale and greater experience with compliance, it discriminates against smaller companies, which increasingly are the principal innovators in drug development. Smaller companies feel the burden of excessive regulation (including user fees) and shifting goalposts disproportionately and are more eager for regulatory change. But even for them regulatory reform is, at most, a long-term strategic goal, and they tend to be more concerned with day-to-day technical and financial crises. Individually they are ill-equipped — and probably would be ill-advised — to criticize and antagonize the regulators who have so much discretion over the testing and marketing of their products. Trade associations are in a better position to be aggressive, to pressure regulators for reforms or other concessions (inasmuch as, unlike individual companies, they are largely immune from retribution), but the two major U.S.-based trade associations that represent pharmaceutical companies — the Pharmaceutical Research and Manufacturers of America ( phrma) and the Biotechnology Industry Organization (bio) — are dominated by large companies that are relatively content with the status quo. Characterizing them as apathetic about regulatory reform would be an understatement.

Second, in spite of all the obstacles, entrepreneurial ingenuity combined with technological innovation continues to spawn new companies. Even as the industry has carried out a series of mergers to create mega-companies during the past decade, a wave of new start-ups has been launched with each new technology or scientific breakthrough. Entrepreneurs created new companies to exploit recombinant dna technology and hybridoma technology in the 1970s; and more recently, to take advantage of genomic, proteomic, and metabolomic data; anti-sense technology; human gene therapy, and interfering rnas (rnai); and to develop “individualized therapies.” Because the ultimate rewards are potentially great, even the imposition of regulatory obstacles cannot extinguish the drive to create, compete, and succeed in this field. However, there would be more numerous and great successes if the regulatory barriers were less imposing and more conscientiously manned.

For reasons that are equally complex, the American public is as passive as the drug companies about regulatory reform. Although literally dying for it, the aging American population is not clamoring for a more streamlined, responsive system that will offer more new drugs sooner and at lower cost. For one thing, few Americans now pay the full costs of pharmaceuticals out of pocket. A significant fraction of prescription drug costs is defrayed by some kind of third-party payer, usually a managed-care organization, insurance company, or the government. Therefore, people who need expensive new drugs seldom experience the full impact of inflated prices at the pharmacy.

A third factor, which influences the public’s view of regulation and regulatory reform, harks back to the asymmetry described above: The public is made aware by the media of approved drugs that manifest problems (or alleged problems) but does not know about — and therefore is not concerned about — drugs that have never been developed at all because of regulatory barriers to innovation. A drug that is not developed is out of sight and out of mind, and the public is simply not aware of what it has been deprived of.

Fourth, as far as most Americans are concerned, the nuances of drug development and its regulation are quite arcane and obscure. For example, many people assume that the fda actually carries out the testing of pharmaceuticals, but it does not: Regulators only evaluate data generated and submitted by industry. And the media have done little to educate them, too often evincing interest in regulatory affairs only during a crisis — or what it can portray as one.

Most important, at any given time, the majority of Americans are healthy. They tend not, therefore, to pay attention to the fact that the quest for zero risk systematically prevents them from getting drugs that they might need in the event of injury or illness. And even if they were aware of this regulatory bottleneck, the lack of fda and congressional accountability means that they can do little about it. Such impotence, in turn, offers them little incentive to become informed and involved. Even the few consumers who have a rudimentary understanding of who does what, when, and to whom during drug development tend to be fearful about new products. Many Americans have become conditioned to seek technological innovation that is completely risk-free, and they seek someone to blame when it is not. Their innumeracy and lack of understanding about tradeoffs among different sources of risk make them highly susceptible to misleading information from those who regularly raise false alarms and demand that regulators ban, withdraw, limit, and restrict many useful products.

These factors combine to confuse consumers and make them hesitant even to endorse, let alone demand, significant regulatory reform. Not realizing that there is a point of vanishing returns that we have passed, they believe that more regulation must be synonymous with more protection, and that more government scrutiny will make us safer and move us ever closer to the Holy Grail of zero risk. In addition, because of the lack of sound, objective, easily accessible information about the potential harmfulness of the regulatory status quo, they exhibit “rational apathy,” which simply means that in the absence of any obvious and proximate threat to their well-being — and of any likelihood that their actions will change the course of events — it is reasonable for them to remain unconcerned.

Another related phenomenon is that significant societal change seldom occurs except at times of crisis, when ideas have a greater likelihood of consequences. As Nobelist Milton Friedman and co-author Rose Friedman observe, in a kind of economist ’s equivalent of the physical property of inertia, “once a tide in opinion or in affairs is strongly set, it tends to overwhelm counter-currents and to keep going for a long time in the same direction. The tides are capable of ignoring geography, political labels, and other hindrances to their continuance. ” Yet, the Friedmans continue, the very success of these tides “tends to create conditions that may ultimately reverse them.” ¹⁷

No matter how profound the economic and public health costs of over-regulation, no matter how obvious and persuasive the arguments for regulatory reform, the fundamental changes needed to turn the tide will occur only if the public demands them.

Henry I. Miller, a former official at the National Institutes of Health and the Food and Drug Administration, focuses on government regulation of science and technology as a research fellow at Stanford University’s Hoover Institution. David R. Henderson is a research fellow at the Hoover Institution and an associate professor of economics at the Naval Postgraduate School in Monterey, California. He was the senior economist for health policy on President Reagan’s Council of Economic Advisers.

¹Richard G. Frank, Ernst Berndt, and Susan H. Busch, Price Indexes for the Treatment of Depression(aei/Brookings Institution, December 1997).

² Susan C. Fagan et al, “Cost-effectiveness of Tissue Plasminogen Activator for Acute Ischemic Stroke,” Neurology50 (1998).

³National Bureau of Economic Research Working Paper No. 10884(November 2004), http://www.nber.org/papers/w10884(accessed February 4, 2005).

⁴ <http://www.manhattan-institute.org/html/mpr_01.htm> (accessed February 1, 2006).

⁵ Pharmaceutical Research and Manufacturers of America, 1999 Industry Profile, 32.

⁶ Henry G. Grabowski and John M. Vernon, “Returns to r&d on New Drug Introductions in the 1980s,” Journal of Health Economics 13 (1994).

⁷ Henry G. Grabowski and John M. Vernon, “Returns to r&d on New Drug Introductions in the 1980s,” Journal of Health Economics 13 (1994).

⁸ Healthy Skepticism, http://www.healthyskepticism.org/library/ref.php?id=10064 (accessed May 15, 2007).

⁹ O.M Bakke et al. “Drug safety discontinuations in the United Kingdom, the United States, and Spain from 1974  through 1993: A regulatory perspective,” Clinical Pharmacology and Therapeutics   58  (1995).

¹⁰ C.A. Brown and T.C. Johnson, “Conditioning fda Approval on Agreement Not to Advertise Violates Law and Constitution,” Washington Legal Foundation Legal Backgrounder 20   (July 15, 2005).

¹¹ A. de Gramont, A. Figer, M. Seymour et al. “Leucovorin and fluorouracil with or without oxaliplatin as first-line treatment in advanced colorectal cancer, ” Journal of Clinical Oncology   18  (2000).

¹² Summarized at http://www.smcma.org/Bulletin/BulletinIssues/March06 issue/Is%20 the
%20fda%20 safe.html.

¹³ <http://www.iom.edu/cms/3793/26341/37329.aspx>  (accessed October 16, 2006).

¹⁴ <http://www.fda.gov/cder/guidance/6657dft.htm>  (accessed September 1, 2006).

¹⁵ <http://www.fda.gov/oc/speeches/2005/npc0928.html>  (accessed October  16, 2006).

¹⁶ Scott Gottleib, “The War on (Expensive) Drugs,” Wall Street Journal  (August  30, 2007).

¹⁷ Milton Friedman and Rose D. Friedman, “The Tide in the Affairs of Men,” Annelise Anderson and Dennis L. Bark, eds., Thinking About America: the United States in the 1990s  (Hoover Institution Press,  1998).