Hidden Risks, Lethal Truths

Warner-Lambert won approval for Rezulin after masking the number of liver injuries in clinical studies

By DAVID WILLMAN, Los Angeles Times Staff Writer

SUNDAY REPORT: June 30, 2002

WASHINGTON -- Newly obtained internal documents show that Warner-Lambert Co. executives who promoted the diabetes pill Rezulin masked early indications of the drug's danger to the liver from federal regulators and later delayed sharing information about its lethal toxicity with family doctors.

The newly acquired materials show that company management rebuffed employees who questioned liver-injury totals from clinical studies that excluded 38% of the cases. At the time, Warner-Lambert was assuring doctors nationwide that the drug was as safe as a placebo, the harmless pill used as a control in medical testing.

The story of the Food and Drug Administration's handling of Rezulin is well known. After the agency gave the drug "fast-track" approval in early 1997, scores of liver-related deaths were linked to the pill. Rezulin was withdrawn from the U.S. market in March 2000. By then, it had generated $2.1 billion in sales for Warner-Lambert.

But only now is a picture emerging of Warner-Lambert's actions, showing how a major drug maker can obscure risks and then win blockbuster sales. Pfizer Inc., which acquired Warner-Lambert in mid-2000, has turned over voluminous documents to plaintiffs' lawyers in response to more than 2,000 lawsuits filed on behalf of approximately 5,100 Rezulin users or their survivors. Many of the materials have been kept out of public view by judges' orders. The Times obtained copies of the materials, along with transcripts of recent sworn statements by company executives and others.

One of the central figures in the documents is Dr. Randall W. Whitcomb, Warner-Lambert's vice president for diabetes research. Without telling the FDA, Whitcomb used an unorthodox method to count the number of liver injuries among patients who used Rezulin while using different criteria to count liver injuries among patients taking a placebo. The result trimmed the number of cases the company reported to the regulators and to doctors nationwide, giving a markedly more benign view of the drug's risks.

Pfizer said in a recent financial statement that a federal grand jury based in Greenbelt, Md., has sought documents and the sworn testimony of certain former Warner-Lambert employees regarding Rezulin.

"We are cooperating with this investigation," Pfizer said in the financial report, filed March 28 with the Securities and Exchange Commission. The investigation is being led by the U.S. attorney's office in Greenbelt and by agents from the FDA's Office of Criminal Investigations.

It is a crime in the U.S. to deliberately conceal an important fact about the safety of a prescription drug.

A spokesman for Pfizer, Robert J. Fauteux, said in a written statement that his company "believes that, in developing Rezulin, Warner-Lambert made a valuable contribution to the effort to control diabetes." Fauteux also said that, in Pfizer's view, the product labeling for Rezulin and other warnings, approved by the FDA, "appropriately communicated the risks associated with the medication, including the risk of liver injury."

Rezulin, he said, "offered an innovative treatment option to nearly 2 million patients." Citing the pending civil lawsuits, Fauteux declined to respond to specific questions posed by The Times.

The newly obtained materials also reveal that:

Company representatives discussed in 1998 how best to persuade doctors treating Latinos "to take the risk of prescribing Rezulin." The same internal company document described the Latino patients as "easy to intimidate."

When an Arcadia woman, Rosa Delia Valenzuela, suffered liver failure and died in December 1998 while taking the drug in a Warner-Lambert study, company representatives argued against promptly alerting the hundreds of doctors nationwide who continued to conduct the research on more than 2,000 other patients.

As the deaths of U.S. patients mounted, Warner-Lambert offered to--and did--reimburse doctors if they got sued for prescribing Rezulin. The American Medical Assn. has warned doctors against such reimbursements, terming them "unacceptable gifts."

The FDA now attributes 94 liver failures, 66 of them fatal, to Rezulin. A review by The Times of reports filed with the FDA found that Rezulin was cited by doctors and others as a suspect in 556 deaths, including but not limited to the liver failures. Because reactions to prescription drugs are reported voluntarily in the U.S., epidemiologists suspect that as few as 1% to 10% of the actual events are made known to the FDA.

The hospitalizations and deaths stand in contrast to the benefits offered by Rezulin. The drug lowered blood-sugar levels but was not proved to offer any benefits that would either save a patient's life or reduce the serious complications of adult-onset diabetes, including stroke, amputation or blindness.

People with adult-onset, or Type 2, diabetes typically can lower their blood sugar with diet or exercise, oral medications, or insulin. The disease is distinguished from juvenile-onset, Type 1 diabetes, which kills patients unless they receive daily infusions of insulin.

A Perfect Champion for Diabetes Drug

Earnest but affable, scientific yet plain-spoken, Randy Whitcomb proved the perfect promoter for Rezulin.

Whitcomb graduated from medical school at the University of Kansas in 1981 and worked at the National Institutes of Health and at Harvard Medical School before joining Warner-Lambert's Parke-Davis drug unit in 1992. The next year, he took charge of shepherding Rezulin to market.

Whitcomb's presentation to an FDA advisory committee on Dec. 11, 1996, marked a watershed for the drug: If he could help persuade the committee to endorse Rezulin, the agency staff, in all likelihood, would approve it.

Whitcomb provided the panel a seamless summary:

He told the advisors that clinical studies had proved Rezulin to be remarkably safe for the liver, in his words, "comparable to placebo." Whitcomb said only 1.1% of the study patients treated with Rezulin had detectable liver injury. He assured the committee that additional liver-related data that he had examined from the clinical studies, to be furnished within days to the FDA staff, were "very, very similar."

Later that day, the advisory committee unanimously endorsed the pill.

When Warner-Lambert provided the additional information to the FDA eight days later, it was not similar: The total fraction of liver-injured patients who took Rezulin doubled. A scant 0.6% of those given a placebo were found to be injured.

But based on the advisory committee's recommendation, the FDA approved Rezulin on Jan. 29, 1997 and it arrived on the American market two months later.

On May 23, 1997, Warner-Lambert again provided the FDA with an inaccurate description of liver injuries that were found in the clinical studies. This time the company assured the agency that, while four newly identified patients from the studies had elevated liver enzymes in their blood, "none" exceeded the standard for clinically significant liver injury. In fact, two of the patients' enzymes well exceeded the standard; the enzymes in one of them was measured at 30 times normal--a magnitude reflecting potentially life-threatening toxicity.

The erroneous company submission to the FDA was signed by eight Warner-Lambert representatives, including Whitcomb and his boss, Dr. Robert L. Zerbe, the senior vice president for worldwide clinical research and development.

By the fall of 1997, details of the first liver failures among patients who were prescribed Rezulin were about to spill into public view.

With sales of their new drug already approaching $250 million, Warner-Lambert executives in Ann Arbor, Mich., and in Morris Plains, N.J., began drafting language for a revised product label that would address the risk and recommend short-term liver monitoring to doctors as the appropriate safeguard.

In preparing for the new label, Whitcomb in September 1997 started a reexamination of the number of liver injuries that were detected in the earlier clinical studies.

A computer printout titled "Job 105" and dated Thursday, Oct. 23, 1997, provided him with this information:

By the time Rezulin came onto the U.S. market in March 1997, 78 patients in the clinical studies had been found by laboratory tests to have liver injury.

But when Whitcomb and eight of his colleagues met with the FDA to discuss Rezulin's liver toxicity on Oct. 24, 1997, the executives said nothing about 78 patients.

In a written submission to the FDA and in conversations that fall and thereafter with the agency, the company representatives said repeatedly that only 48, or 38% fewer than 78, patients had been detected with liver injury.

If Warner-Lambert had revealed that 78 Rezulin patients experienced liver injury, this would have corresponded to 3.1% of the 2,510 total patients given the drug in the clinical studies. At that percentage, Rezulin likely would have been flagged as the most potentially liver-toxic diabetes drug on the market. This would have depressed sales and heightened concern at the FDA.

An FDA summary of the Oct. 24, 1997, meeting between the agency and Warner-Lambert demonstrates the success of the company's approach. At the outset of the session, held in Rockville, Md., a senior FDA official, Dr. G. Alexander Fleming, "thanked Parke Davis for coming in and being proactive on the issue once the problem was discovered."

Relying on the company's data, the FDA accepted Warner-Lambert's recommendation to place language on the label reflecting 48 cases; the new label would call the incidence of liver failure "very rare."

The FDA officials who conferred with Whitcomb had no idea that the number of acknowledged liver-injury cases had been shrunk, according to interviews and the newly obtained documents.

Dr. Robert I. Misbin, an FDA medical officer who for years supported Rezulin but who ultimately fought for its withdrawal, testified recently that Warner-Lambert "never" informed him of the higher number of cases.

Misbin testified that if the company had conveyed more accurate and complete information about the severity of liver injuries in the clinical studies, he would have opposed approving the drug, in August 1997, as a stand-alone pill for adult-onset diabetes.

According to company documents and sworn testimony this year, even specialists within Warner-Lambert were unable to discern Whitcomb's rationale for reporting 48 cases rather than 78.

Beth Ann Baron, a manager on Warner-Lambert's diabetes team in Ann Arbor, was one of the specialists who tried to obtain documentation for Whitcomb's rationale, starting in October 1997. "I think it's fair to say that everyone that was involved ... would have liked to have seen that," Baron testified in a deposition on March 12. Her colleague, Donald Sizemore, testified in January that he also raised the issue with Warner-Lambert's director of worldwide regulatory affairs, who, he said, "never got back to me on that." Sizemore testified that he was unaware of any other instance when Warner-Lambert did not internally "validate" safety data submitted to the FDA.

A request by The Times to interview Whitcomb last week was declined by Pfizer. In recent trials of civil lawsuits, company lawyers have defended as proper his handling of the liver-injury cases.

Whitcomb remains active in the pharmaceutical industry. In November, he joined the board of directors of Insmed Inc., a publicly traded firm that is developing another drug for adult-onset diabetes. In January, Whitcomb, 48, was one of four former Warner-Lambert executives who launched a privately held drug-development company based in Ann Arbor called QUATRx.

Striving for Clinical Studies Free of Bias

Controlled clinical studies are the gold standard of pharmaceutical research. The studies can help measure the effectiveness and the safety of a drug. The goal is to insulate the research from the subjective judgment, or bias, of the people conducting the experiment.

For this reason, the ground rules for clinical studies are written and discussed in advance, typically in painstaking detail.

This scientific approach helps the FDA and prescribing physicians sort viable therapies from unproved hype. Companies also conduct clinical studies of medicines that have already won government approval, often in an effort to find expanded uses for them.

Dr. Curt D. Furberg, a clinical studies expert who is a professor of public health at Wake Forest University, said that in his view Warner-Lambert may have inappropriately reduced the count of liver injuries from its clinical studies of Rezulin. "It appears that they made up rules as they went along," said Furberg, who also serves on an FDA drug-safety advisory committee.

Whitcomb, in sworn testimony given March 6 to 8, said he did not count cases if he concluded that a study patient's liver injury resulted from what he suspected was a cause other than Rezulin. Yet, according to clinical studies experts, all cases of injury should have been counted and reported. Theorizing about the cause of the events, they say, improperly infused Whitcomb's opinion into the research.

John W. Hornbeck, a Los Angeles-based lawyer for plaintiffs suing the manufacturer, asked Whitcomb in March whether the rejection of a liver-injury case on the basis of what Whitcomb suspected was a "clear alternative cause" could introduce bias into the study.

Whitcomb responded: "It is a possibility, yes."

Hornbeck then asked Whitcomb whether placebo patients showing liver injury were counted, regardless of his opinion about the cause.

Whitcomb responded: "Right."

When Hornbeck asked whether he applied all of the same rules for counting liver injury to placebo and Rezulin patients alike, Whitcomb responded: "I would need to go back and look at each of the placebo patients. Again, I don't recall exactly."

Whitcomb testified that he conferred with a liver specialist retained by Warner-Lambert about shrinking the total of liver-injury patients from the clinical studies from 78 to 48. Whitcomb said he could not remember whether the rules for excluding cases were in writing when he met with the liver specialist, Dr. Paul Watkins, in the fall of 1997.

"There may have been some handwritten things that were put together in the fall of 1997 as we were working on this," Whitcomb said. " ... I can't tell you for sure."

Whitcomb said he was unable to specify exactly how he went about reducing the 78 cases to 48. "I don't have a specific recollection of how we got down to the 48," Whitcomb testified.

Whitcomb and company lawyers have said that he subtracted the greatest number of cases by varying the definition of what constituted liver injury, depending on which laboratory tested each patient's blood. The uniform standard for liver injury was a liver enzyme count of 102--three times the boundary of the upper limit of normal, or 34. By using a variable standard, Whitcomb could have excluded cases with enzyme counts as high as 135.

The use of a variable standard for liver injury ran counter to the FDA's understanding of the ground rules for the Rezulin studies, according to internal company documents, the testimony of FDA medical officer Misbin and statements that Whitcomb himself made under oath more than a year ago.

Indeed, an aide to Whitcomb, Thomas Valiquett, wrote in an April 30, 1999, e-mail to colleagues, "The value of 34 was decided by Randy" Whitcomb. Valiquett pointed out that 34 was the value that Warner-Lambert used when addressing the FDA's earlier concern about liver toxicity.

When Watkins, the company-hired liver specialist, was asked whether the value for liver enzymes that he and Whitcomb used in fall 1997 was 102, or triple the upper limit of 34, he testified: "That is correct."

When Whitcomb was asked in his first deposition, on Dec. 13, 2000, whether the upper limit of normal used for the Rezulin studies was 34, he testified: "Uh-huh.... That would be about right, yeah."

Whitcomb then was asked by Zoe B. Littlepage, a plaintiff's lawyer from Houston, "Do you recall ever being asked by anybody at Parke-Davis to go back and look at the clinical trial data and use any other upper limits of normal number other than 34?"

Whitcomb responded: "I don't ever remember being asked that, no."

In testimony in March of this year, Whitcomb defended his count of the cases. Asked by Hornbeck, "Sir, the label does not reflect that there were 78 cases and then a post-hoc analysis reduced that 78 to 48, does it?"

Whitcomb countered: "Seventy-eight is a scientifically invalid number."

When Hornbeck asked whether "there were really two sets of books for the data," Whitcomb replied: "I disagree with that completely."

A More Risky Course on Rezulin in the U.S.

When it came to dealing with the concerns of regulators, Whitcomb was Warner-Lambert's go-to executive. Late in the fall of 1997, Warner-Lambert dispatched him to Britain, where alarm about Rezulin was cresting. The drug was marketed in the United Kingdom by another company, Glaxo-Wellcome.

But while he was en route, executives at Glaxo, in consultation with British health authorities, decided to withdraw Rezulin. The decision was a stinging setback to Warner-Lambert, prompting a senior executive to suggest a link between sales and the degree to which a company would tolerate a safety risk.

"[T]he product was looking to be a commercial disappointment in the U.K. due to spotty commitment within [Glaxo] and poor preparation. Hence it was easy for them to 'take the high road' on safety," wrote pharmaceuticals division President Anthony H. Wild in a Nov. 30, 1997, letter to his Warner-Lambert colleagues.

Left unsaid in Wild's letter was that, in the U.S., Rezulin was among the biggest-selling new drugs on the market.

While Glaxo chose the "high road," Warner-Lambert stayed a more risky course.

Despite the mounting liver failures, the company aggressively recruited doctors with general and family practices, the new documents show.

Warner-Lambert's marketing executives helped devise a program that would pay such doctors up to $350 for every patient they enrolled in a study called REACH.

Some of the physicians stood to make thousands of dollars by shifting their patients from other diabetes drugs to Rezulin.

The REACH study was planned for well more than a year. And unlike the earlier clinical studies, REACH could have no bearing on the FDA's original decision to approve Rezulin.

But by early 1998, word of Rezulin's liver toxicity was complicating the recruitment of doctors and patients. Concerns deepened with the liver failure and death, on May 17, 1998, of Audrey LaRue Jones, an East St. Louis, Ill., woman given Rezulin in a diabetes-prevention study conducted by the National Institutes of Health. Because doctors had strictly monitored her liver functions, which Warner Lambert and the FDA had relied on to catch liver problems in time, the death raised significant new doubt about Rezulin. On June 4, 1998, officials at NIH announced they had excluded Rezulin from the government's ongoing study.

Company executives sought to preempt any regulatory move to force the pill off the market.

By late July 1998, Whitcomb and other company executives convinced the FDA that only a slight change on the drug's label was needed. The new label made no reference to the death in the NIH study. The FDA made no public statement about the event, and the FDA did nothing to deter use of Rezulin in other studies, including REACH.

A panel of experts at NIH found that Rezulin "probably" caused the woman's liver failure, and the coroner's death certificate attributed the "underlying cause" to Rezulin. But Warner-Lambert issued a news release that blamed the death on factors "unrelated" to Rezulin.

Moreover, the newly obtained documents show that Warner-Lambert's senior director of medical and scientific affairs, Dr. Robert G. Thompson, wrote to doctors who had signed up for the REACH study, assuring them of the drug's safety.

But many remained anxious.

"There is a lot of concern from the physicians about the safety of Rezulin," wrote Jenny Chin, a scientist under Thompson, in an e-mail to him on June 17, 1998. "I truly believe that it would be very helpful if we sent out another Parke-Davis letter from you (doctor to doctor) reassuring the physicians."

There also was concern internally at Warner-Lambert, the new documents show.

After reviewing adverse reactions reported among patients who were prescribed Rezulin, another company scientist, Alexandra Pearce, wrote in an e-mail that she sent to a colleague on Nov. 1, 1998:

"When I was entering the data yesterday, I was thinking I really wouldn't like to be a woman on 400 [milligrams] of Rezulin." Pearce termed the adverse reactions "statistically significant."

Yet publicly, company representatives betrayed no such concerns. They continued to extol Rezulin as safe when used in accordance with the recommended liver monitoring. And Thompson kept reassuring the REACH doctors, also called investigators.

"Bob Thompson is planning teleconference with the REACH investigators to update them on the status of the trial and to reassure them that Liver function monitoring is going well," Chin wrote in a Nov. 9, 1998, e-mail. "We need to find a way to motivate the doctors.... They want to be reassured that Rezulin therapy is safe as long as liver function monitoring is being conducted according to labeling guidelines."

On Dec. 8, 1998, Thompson wrote again to the REACH doctors. He claimed that the incidence of liver-related deaths or organ transplants among Rezulin patients who were "not adequately monitored" was just 1 in 60,000.

"The company has always been forthcoming with information regarding the safety and efficacy of this therapy," Thompson wrote.

Ten days later, Rosa Delia Valenzuela died at USC-University Hospital in Los Angeles.

Medical records and recent sworn testimony show that her doctor enrolled the Arcadia woman in the REACH study on Oct. 9, 1998, shifting her from Glucophage to 400 milligrams of Rezulin daily. Valenzuela's liver functions were monitored at the outset and were found to be normal.

A month later, her liver functions were again monitored and were found to be normal. But 13 days later, on Nov. 19, 1998, her liver enzymes were more than 30 times normal. She descended into liver failure.

Valenzuela, 63, the mother of five adult children and the wife of a renowned, retired horse-racing jockey, Ismael "Milo" Valenzuela, died Dec. 18, 1998. The Los Angeles County death certificate attributed it to her body's reaction to Rezulin. It cited as contributing factors weight, alcoholism and diabetes.

At the time of her death, Valenzuela was one of 2,607 patients in the REACH study who were scheduled to take Rezulin for six months.

Her death posed a clear threat to the viability of Rezulin, because it was the second time in just seven months that a patient had developed liver failure and died, despite having been closely monitored in a clinical study.

Warner-Lambert complied with the law by including the death among cases that it reported to the FDA. But the company executives balked at promptly alerting the 400 or more other doctors who were conducting the REACH study.

"Do we have to send a letter out to all of the REACH investigators to inform them of this event?" Chin asked in an e-mail that she sent to Thompson and others.

Said a follow-up internal e-mail: "We have NO REGULATORY OBLIGATION to send a letter to the REACH Physicians." The e-mail added, "The Liver Experts we have spoken to were not convinced that this incident was related to the study drug. So for Bob [Thompson] to send out a letter now would be misleading because we cannot fully explain the case and it would be unnecessarily frightening."

Another memo cited similar advice from Whitcomb's boss, Dr. Zerbe, the company's senior vice president for worldwide clinical research and development.

According to records and interviews, the physician who placed Valenzuela on Rezulin and the liver specialist who treated her at USC-University Hospital concluded that the drug probably did cause her liver failure. The death certificate implicated Rezulin. And specialists consulted by the company, including Watkins, advised as of late December 1998 that the drug could not be excluded as having caused the liver failure.

Yet more than a month after Valenzuela's death, when Warner-Lambert's Thompson informed the other REACH study physicians of the fatality, he pointed to factors other than Rezulin. And he did not mention that monitoring the patient's liver functions had, once again, failed to prevent organ failure and death. Pfizer declined to make Thompson available for an interview for this article.

"I want to update you with available information on a 64-year-old [sic] Hispanic female who died of liver failure while enrolled in the REACH trial," Thompson wrote the doctors on Jan. 21, 1999. "In addition to Type 2 diabetes, this patient had a history of gallstones, cardiac problems and a long history of alcohol abuse.... She was admitted to the hospital but was not eligible for liver transplantation because of other illnesses. She died of liver failure approximately 5 weeks later."

About the same time, Warner-Lambert took the extraordinary step of offering to indemnify doctors nationwide if they were sued for prescribing Rezulin. The company also offered to provide the doctors with experienced lawyers and to cover the physicians' other lawsuit-related expenses.

The American Medical Assn., in a formal opinion first issued in 1992, advised doctors against entering such arrangements:

"Physicians should prescribe drugs, devices and other treatments based solely upon medical considerations and patient needs. A third party's offer to indemnify a physician for lawsuits arising from the physician's prescription or use of the third party's drug ... introduces inappropriate factors into medical decision making. Such offers, regardless of their limitations, therefore constitute unacceptable gifts."

Fauteux, the Pfizer spokesman, said the indemnity policy was made available "only to doctors who prescribed Rezulin in accordance with federal, state and local laws, and in compliance with the" product labeling. He said that "Warner-Lambert believed that doctors should not be held liable for properly prescribing a drug that was an appropriate treatment for their patient."

The Valenzuela family is suing the company and the doctor who shifted Rosa Valenzuela onto Rezulin. A trial is scheduled to begin in September in Los Angeles Superior Court.

The first six trials of civil lawsuits alleging wrongdoing by the company in its handling of the drug unfolded over the last few months. In two cases, tried in Houston and in Rockville, Md., juries returned verdicts in favor of the company. Those suing Warner-Lambert won multimillion-dollar verdicts or monetary settlements in the other four cases, in Jackson, Miss.; Corpus Christi, Texas; Liberty, Mo.; and Tulsa, Okla.

The jury foreman for the Missouri case, Clark Lamoreux of Kansas City, said that in his view Warner-Lambert broke the law.

"Warner-Lambert, Parke-Davis, showed blatant disregard for public safety by misleading the FDA, doctors and patients," Lamoreux said in an interview. "If I had a family member that was on Rezulin, by God I would definitely want somebody prosecuted to the fullest extent."

* * *

Los Angeles Times researchers Janet Lundblad in Los Angeles and Sunny Kaplan in Washington contributed to this report.