ImClone Systems

Bungled relations with the FDA cost this company dearly.

In early December 2001, investors, researchers, doctors, and patients alike were anxiously awaiting FDA approval of ImClone Systems’ first product, a new cancer drug called Erbitux (cetuximab). In some patients, it dramatically shrank tumors that had been unresponsive to other treatments. Doctors and patients had already made more than 10,000 “compassionate use” requests for the drug, and ImClone investors were banking on its success.

But on December 28th, the Food and Drug Administration faxed ImClone headquarters with bad news: the agency refused to consider ImClone’s submission for market approval.¹

The insider trading scandals that followed are well documented. However, except to scapegoat the agency, no one has analyzed ImClone’s communications with the FDA. The two had been working together on Erbitux for years (see “The Road to Perdition,” in the sidebar). Since the agency issues a refusal to file (RTF) only as a last resort, had the FDA failed to communicate its concerns during that time, or had ImClone failed to get the message?

In August 2000, ImClone asked the FDA whether data from an ongoing Phase II study (the 9923 study) in colorectal cancer patients could be used to fast-track Erbitux’s review.² ImClone’s strategy for accelerated approval was ambitious. It was seeking approval based on a small trial that used a combination therapy of Erbitux and irinotecan (Campto, Aventis Pharmaceuticals; Camptosar, Pharmacia), a chemotherapeutic. According to FDA records, the agency “advised other options” and warned ImClone that relying on results from the proposed study “would be more rapid, but higher risk.”³

But ImClone had submitted, and the FDA evaluated, a protocol different from the one the company actually followed. Though ImClone must have known that the FDA’s decisions were based on an outdated protocol, the company never explicitly informed the agency. According to the subsequent RTF, this was one of several factors making the data “uninterpretable.” The agency also enumerated requirements necessary for approval using the fast-track approach; much of the RTF was based on ImClone’s failure to meet these requirements.

While the scale and consequences of ImClone’s development blunders may be extreme, these missteps illustrate common mistakes that many drug sponsors, even those with more experience, make. Thus, ImClone serves as a useful case study.

Irrational optimism
Though essential to prove a drug is safe and effective, clinical trials are expensive. Sometimes, drug sponsors see trials as impediments to marketing their products, and a sponsor’s faith in the efficacy of its drug can lead researchers to believe that fewer trials are necessary. Even established drug companies underestimate the number of clinical trials ultimately required for approval.

Putting market valuation first
ImClone’s aggressive self-promotion meant that typical clinical development setbacks could be fatal to the company’s market capitalization. Not surprisingly, ImClone was reluctant to add trials that the FDA repeatedly advised, betting instead that the clinical performance of Erbitux would overcome procedural deficiencies. On January 19th, 2001, the FDA told ImClone to conduct an additional trial comparing Erbitux alone to Erbitux combined with irinotecan. ImClone completed this trial October 12th; that same day, according to congressional testimony, then-CEO Sam Waksal told Bristol-Myers Squibb that results were promising and that he expected Erbitux to be on the market by March. Two weeks later, Bristol-Myers paid a premium to acquire almost 20% of ImClone’s stock. The difference between the results of the two trials, however, was not statistically significant, so the FDA was unlikely to approve the combination therapy.

Keeping silent
ImClone avoided mentioning to the FDA certain information. When the agency granted Erbitux an accelerated review process, it did so based on an outdated protocol. In the trial ImClone actually used, patients had renal cancer, thus making the results less relevant clinically. According to congressional testimony, the FDA learned crucial details of the clinical trials only after ImClone filed its biologics license application (BLA) at the end of October 2001.

The FDA repeatedly asked ImClone for data that were never supplied. For example, at a meeting in January 1999, the FDA questioned the high dose level for the drug. ImClone defended its dosing, and the FDA requested data to support this defense. It requested these data again at the August 2000 meeting, in a letter of January 2001, and finally once more in March 2001. ImClone never supplied them. The FDA also repeatedly requested data showing that patients receiving combination therapy hadn’t responded to irinotecan alone; ImClone never supplied these to the FDA’s satisfaction.

Not seeking impartial advice
In the RTF, the FDA recapped its very specific advice about the kind of data it required. ImClone would have been better off listening and responding earlier, but evidently no one at the company wanted to hear the message. This was ImClone’s first BLA, and there is no record of an external review. Even Bristol-Myers, which has brought many drugs to market, apparently was not consulted for advice; ImClone seemed more interested in sharing good news with its potential business partner than in seeking its input.

ImClone’s overarching mistake may have been putting a good spin on all results, even to the point of deluding itself. While the FDA wanted to evaluate the drug quickly, clearing Erbitux for accelerated review did not imply tacit approval. According to congressional testimony, not until December 4th did ImClone’s vice president of regulatory affairs realize that the possibility of an RTF was real. Throughout December, ImClone staff members called the agency in an attempt to stop the RTF. But at that point, the agency needed data (much of which did not exist)—not phone calls.

Case Studies

» ImClone Systems

VaxGen

Merck-Medco

1  FDA. Refusal to File letter to ImClone Systems. Published in The Cancer Letter. Special Report: The C225 Controversy. (March 8, 2002) 1,23–7.
2  Potential therapies for life-threatening diseases with few treatments often go through a faster approval process that allows smaller trials.
3  Center for Biologics Evaluation & Research. (October 10, 2002) Cetuximab (Erbitux)—Slide Presentation.

VaxGen

Spinning results deflated a firm’s reputation.

VaxGen, a pharmaceutical company in Brisbane, California, is currently facing class actions and possible bankruptcy following the disappointing results earlier this year in its first large trial of an AIDS vaccine. The overall results showed that HIV infection rates between the placebo and vaccine groups were identical. However, the company claimed that the vaccine did prevent infection in some minorities, notably Blacks and Asians, subgroups that together made up only 10% of the subjects. Although most of the criticism of VaxGen has fallen on the data analysis, many of the company’s problems stem from how it monitored and announced its trial results.

Investigators often present trial results at scientific conferences, which provide an opportunity for peer review, and this particular trial was widely anticipated by the media, activists, and investors. But instead of objectively reporting the results at a professional meeting of this sort, VaxGen announced its results on a conference call and Web cast, in which the company claimed the findings were “pretty incredible,” touting efficacy in predefined subgroups of subjects, but not releasing the detailed data behind this assertion.

For example, a closer look at the numbers reveals that among the 314 Black subjects out of more than 5,000 trial enrollees, the infection rate was 8.1% in the placebo group and 2.0% in the vaccine group. Unfortunately, reanalysis has called into question the significance of this result. Statisticians generally distrust subgroup comparisons, especially in the absence of an overall trend: unexpected results are rarely reproducible or medically explicable, and the small sizes of subgroups make such results notoriously imprecise. Although VaxGen was right to probe its data, the company should have realized that subgroup analysis alone is not persuasive and an additional trial was necessary to demonstrate definitively the protection in Blacks.

VaxGen’s negative results and consequent drop in share price led to class actions claiming that top company executives, including CEO Lance Gordon and President Donald Francis, purposely misled investors to expect a positive outcome. On this point, the company deserves the benefit of the doubt. As is typical in trials with outcomes of serious morbidity or mortality, VaxGen used an independent data safety and monitoring board (DSMB) to ensure patient safety. During the trial, the DSMB examined the data once for overwhelming efficacy and at several points to check for subject safety. Interim data was known only to the committee. As part of accepted practice, VaxGen executives did not receive “unblinded” data until the final analysis.

At some point during the trial, a statistical analysis would have demonstrated that there would be no benefit from the treatment. But because the trial was monitored to completion, it raised expectations unrealistically. If VaxGen had included such an analysis in its trial design, the DSMB could have included a so-called futility analysis and advised ending the trial early. Had the trial been stopped for this reason, VaxGen would have avoided raising false hopes, conserved resources (trials are expensive), and faced less criticism for spinning results. While companies, of course, are loath to yield this kind of control to an independent board, the practice can prevent turmoil from occurring in trials for treatments that lack strong pretrial evidence of efficacy.

The organization of a high-profile clinical trial—its leadership, independent data safety and monitoring committee, and statistical analysis—all contribute to the trial’s credibility. The VaxGen trial had all these elements in play, but its execution fell short; the DSMB had too narrow a mandate, and VaxGen officials were too eager to put a good face on results. Now that the results have been announced, peer review must referee the data controversies.

VaxGen may not have the chance to finish the game. In a March 31st filing of the U.S. Securities and Exchange Commission, VaxGen’s independent auditor, KPMG, warned that the company has insufficient funds to continue operations through 2003 without obtaining further funding. The future of this HIV-vaccine pioneer remains in doubt.

Case Studies

ImClone Systems

» VaxGen

Merck-Medco

Merck-Medco

The merger of this pharmaceutical giant and a pharmacy benefits firm amounted to an intolerable conflict of interest.

Merck was the darling of Wall Street from the mid-’80s till the early ’90s. For seven years running, it was the most admired company in the United States, according to Fortune. But after years of robust growth—between 24% and 34% annually from 1985 to 1992—Merck, like other large pharmaceutical companies, began to feel the pinch as health management organizations cut costs. By 1993, Merck’s yearly growth had slipped to around 10%, and the value of its shares had fallen 38% from July 1992 to July 1993.

Meanwhile, Medco Containment Services, which helped medical plan sponsors lower costs of prescription drugs, was growing at a rate of 35% a year, with 1,450 clients serving 33 million patients—26% of all Americans covered by drug benefit plans in 1992. Medco’s purchasing clout was painfully apparent to Merck: when Merck refused to discount its products to Medco, the pharmacy benefits management (PBM) firm persuaded its network of doctors to favor Bristol-Myers Squibb’s Pravachol (pravastin), a cheaper cholesterol-lowering drug than Merck’s relatively expensive Mevacor (lovastatin).

Suddenly warming to the PBM, Merck began to talk to Medco about how its drugs might be better placed on the organization’s list of approved therapies. Medco’s chairman and founder, Martin Wygod, and Merck’s then-CEO, Roy Vagelos, hit it off so well that Merck acquired Medco for $6 billion. The acquisition proceeded swiftly, despite whispers of conflict of interest. In 1993, Merck-Medco became the official PBM subsidiary of one of the world’s biggest drug companies.

Nine years later, Merck-Medco’s revenue had soared from $2.2 billion in 1992 to $29 billion in 2001; the group managed more than 537 million prescriptions annually, serving more than 65 million patients. But despite Medco’s business success, in January 2002, Merck’s management announced plans to spin off Merck-Medco, renamed Medco Health Solutions, and establish the unit as a separate publicly traded company worth an estimated $6 billion. In July 2002, Merck postponed the proposed initial public offering—a decision that was “due solely to [poor] market conditions” according to a company source who chose not to be named.

Merck needed to shed this valuable asset because the conflict of interest had become too damaging to ignore. By January of this year, class actions encouraged attorneys general from at least 25 states to begin inquiries into Merck-Medco’s operations, to determine whether the company violated state antitrust, consumer protection, or licensing laws.

In March, court documents revealed that the PBM received more than $3.5 billion in rebates in the late ’90s from pharmaceutical manufacturers that wanted, as Merck originally had, to be placed on Medco’s recommendation list. Moreover, according to the documents, in a three-month period, Medco persuaded doctors to switch more than 71,000 prescriptions from Lipitor (atorvastatin calcium), Pfizer’s cholesterol medication, to Zocor (simvastatin), a more expensive Merck treatment. Medco also promoted other products by Merck over those of competitors.

In mid-March, a Medco spokesperson, Anita Kawatra, insisted that the company’s strategy of persuading doctors to switch medications had “been approved by an independent committee of doctors and pharmacists and by our clients, who save money on the whole.” But criticism of the alliance is undiminished, and other PBMs, like Advance PCS and Express Scripts, which are not owned by pharmaceutical companies, have come under scrutiny as well. All are accused of violating fiduciary duties by failing to disclose the extent of their financial ties.

Now Medco is a thorn in Merck’s side, and while the PBM’s acquisition began a beneficial relationship that has lasted nine years, it may not be in Merck’s ultimate favor. Other large pharmaceutical companies may be well advised not to mix drug development with drug dispensing, even if doing so promises to bring short-term financial gains.

Case Studies

ImClone Systems

VaxGen

» Merck-Medco