Chinese Trial Misconduct Delayed Bristol-Myers MedicineDrew Armstrong
Drugmakers have increasingly been turning to China for large clinical trials because they’re cheaper and there’s a bigger population of subjects to draw on.
Now U.S. regulators have stepped in, questioning sloppy data and irregularities from the world’s most populous country.
Bristol-Myers Squibb Co. and Pfizer Inc.’s blood thinner Eliquis, approved in December, was stalled for nine months because of misconduct, errors and an alleged cover-up attempt at a Chinese trial site overseen by Bristol-Myers, according to documents posted by the U.S. Food and Drug Administration. The delay came after the company told the FDA that patients got the wrong medicine, records were secretly changed and “serious adverse events” went unreported, the documents show.
The errors led to a lengthy reanalysis of the data and spurred a debate within the agency on what the drug’s label should say about its effectiveness. An agency official also questioned whether large trials in countries like China with similar data shortfalls were a viable basis for approving treatments, according to the documents.
The mistakes showed a “pattern of inadequate trial conduct and oversight,” according to minutes of a Feb. 9 agency meeting involving the two New York-based companies and the FDA, posted on the agency’s website.
Sales for Eliquis may one day reach $10 billion a year, according to analysts. The delay, though, may cut the time Eliquis is protected by patent, reducing revenue by billions of dollars. The Eliquis incident is an example of the increasing scrutiny the pharmaceutical industry is facing on its research in China, which offers a huge base of test subjects and costs that the Tufts Center for the Study of Drug Development says can be half that in the U.S.
Drugmakers will keep having problems with sloppy data and misconduct as long as they keep doing trials in places like China without providing better oversight, said Thomas Marciniak, an FDA medical team leader who wasn’t directly involved in the Eliquis application but reviewed the trial independently.
“What we need is high-quality trials. If we’re not getting them in the low-cost areas, either fix the low-cost areas, or stop doing them,” Marciniak said in an interview, emphasizing that he was speaking for himself and not the agency.
Last month, London-based GlaxoSmithKline Plc said it fired its head of Chinese research after the scientist allegedly misrepresented data that was published in a medical journal.
Bristol-Myers, which ran the Eliquis trial known as Aristotle, responded appropriately once the mistakes became known, said Elliott Levy, the company’s executive who oversaw the research. A reanalysis done by the company and the FDA deleted the questionable data and found it didn’t substantially affect the final, positive result, he said.
The mistakes “were not exceptional,” Levy said in a telephone interview. “The issues they raised required recourse to the primary source data and some months to fully evaluate, but they’re not exceptional issues.”
Asked whether the issues in China created concerns that other misconduct or bad data may have occurred in the trial, Levy said Bristol-Myers was confident they hadn’t. “I don’t think there’s anything unique about China in this regard,” Levy said. “We’ve looked closely at the quality of the data and reliability and it’s not distinguishable from the United States and Europe.”
Pfizer is confident in the trial results, said Mackay Jimeson, a spokesman for the company.
Christopher Granger, a professor of medicine at Duke University in Durham, North Carolina who was the lead outside researcher on the trial, disagreed with Levy.
“There is a greater likelihood of some of this impropriety in certain regions,” Granger said in a telephone interview. “We’ve had experiences in India and China where we’ve had more than we would have expected.”
Eliquis was developed as a safer and easier-to-take replacement for warfarin, a half-century-old blood thinner widely used to combat blood clotting and strokes. Pfizer, the world’s biggest drugmaker, and Bristol-Myers share sales on Eliquis, which competes with Boehringer Ingelheim GmbH’s Pradaxa, and Bayer AG and Johnson & Johnson’s Xarelto.
The final-stage trial of Eliquis began in 2006, and eventually grew to more than 1,000 sites in 40 countries, according to the FDA. About 16 percent of the 18,000 patients were in Asia, with three dozen sites located in China.
Doctors and hospitals who sign on as investigators are typically paid for getting patients to enroll in the trial. They’re overseen by the drug companies, which monitor the patients in coordination with the physicians. Much of that work is done by contract research organizations.
In the Eliquis trial, Bristol-Myers hired Pharmaceutical Product Development Inc., a closely held, Wilmington, North Carolina, company known as PPD, to help oversee it.
The Eliquis trial was questioned on two issues, according to the FDA documents first cited by the journal Pharmaceutical Approvals Monthly. One was the improper manipulation of records at a study site for 35 patients at the Shanghai 9th Peoples Hospital in China. The second involved the high percentage of the 9,000 patients who were supposed to be getting Eliquis, and instead were either given the wrong drug, or the wrong dose.
There was a broad list of issues at the Shanghai hospital, according to FDA documents. They included failure to report four potential adverse medical events, late reports on three others and three medical outcomes that weren’t included in the data. Additionally, some patient names and dates were wrong, and Chinese and English records didn’t match in some cases. The FDA also reported that some patient records disappeared just ahead of a site visit by agency inspectors.
“The records were altered in order to cover-up GCP violations which had occurred at the site,” the FDA said in its report. GCP stands for “good clinical practice.”
Levy disputed some aspects of the report. The company “examined the trial data at that site and found that all the primary endpoints and the key secondary endpoints were appropriately documented and reported,” he said.
While the Shanghai site had only 35 of 18,000 patients in the trial, the FDA raised the possibility that problems there may have extended well beyond one hospital. A female Bristol-Myers employee who ordered patient records changed ahead of the FDA’s inspection worked at 18 other trial locations in China, according to the documents, and she helped prepare at least one other site for an FDA visit.
The employee, whose name has been redacted from the FDA documents, told contract research staff at the site to change documents and data on a portable computer drive ahead of an FDA inspection after the trial ended, according to the documents.
Afterward, a contract worker with PPD, who also isn’t identified, told supervisors about the cover-up, and Bristol-Myers fired the female employee who ordered the data changed. Two others were also fired, including a senior management employee, the records show.
“This was an attempt by a single individual to fill in gaps in documentation that might be noted by an inspector,” Levy said. “That obviously should never happen, but the fact is that in enterprises as large and complex as an 18,000-patient trial, where tens of thousands of people are at work, there can be instances where one individual misinterprets their work.”
Lee Kennedy, a PPD spokesman, said the company had no comment on the report about the trial.
At a Feb. 9, 2012, meeting at the FDA’s sprawling Silver Spring, Maryland campus, the agency told Bristol-Myers and Pfizer executives the mistakes showed a “pattern” of inadequate oversight,’’ according to minutes from the session posted on the agency’s website.
In a summary, the agency said it was “uncertain whether it was reasonable to assume that the problems in trial conduct identified by the review team were all or even most of the significant problems.”
The companies and the FDA also looked into whether a large number of patients received the wrong medicine, after the finding that six times as many patients in the Eliquis arm of the trial either got the wrong dose or the wrong drug.
After Bristol-Myers told the FDA about the medication problems in January 2012, the agency said the drugmakers “could not provide a rationale” why so many more patients in the Eliquis arm of the trial were taking the wrong drug.
In fact, the discrepancy was a problem with the study’s record-keeping methods, not the trial itself, according to Granger and Levy. “It certainly raised concerns at the FDA that had to be thoroughly investigated,” Levy said.
When the companies announced in March 2012 that the FDA was delaying approval by three months, they didn’t publicly provide details about the problems they’d uncovered. All they said then was that they “submitted additional information about the Eliquis clinical program to the FDA, which constitutes a major amendment to the application and will require additional time.”
Marciniak has been critical of the design and execution of past large trials, such as for GlaxoSmithKline’s diabetes pill Avandia. In a Dec. 11, 2012, “special clinical review” that he conducted on his own, Marciniak said that missing data in the Eliquis trial meant that Bristol-Myers and Pfizer shouldn’t be able to make certain claims about the drug’s efficacy and safety because the statistically significant benefit in death outcomes might not be valid, according to minutes of a meeting.
His report also attacked the documentation around when deaths occurred. In a number of cases, he wrote, clinical records from the trial showed that patients had visits with doctors after they were already dead.
In an interview, Marciniak said sloppy trial data is a growing problem.
“You can have a bad trial, and it doesn’t necessarily mean a bad drug,” Marciniak said.
The FDA is “comfortable with the results” of the trial, Sandy Walsh, an agency spokeswoman, said in an e-mail. It also allowed the companies to include information about the death benefit on the label, over Marciniak’s objection, without saying anything about the data problems.