Why did Merck and Schering-Plough take 20 months to analyze a key study in which their cholesterol blockbuster Vytorin failed to best a cheaper generic?
The companies say they were stalled by data problems. Congress is investigating whether they're telling the whole truth. And James Stein, a University of Wisconsin researcher, finds himself the man in the middle of the fracas.
The reason: Stein he says he told Merck/Schering-Plough representatives on two occasions that there were problems with specific images from the delayed study. But now he says the final data presented last month looked "quite good." That contradicts the idea that the data needed so many months of cleanup.
Stein was already front and center in the investigation by the House of Representatives' Committee on Energy and Commerce. When he felt the companies were misrepresenting the views of an expert panel on which he served, he called them out on it, documents released April 11 show.
"I'm an honest guy, and my first obligation is to tell the truth," says Stein. "Even if it means I'll probably never get to do a scientific investigation with Schering-Plough again."
Critics of Merck (nyse: MRK - news - people ) and Schering-Plough (nyse: SGP - news - people ) worry that technical problems became an excuse to avoid facing a result the companies knew might hurt their $5 billion-a-year cholesterol franchise. Merck and Schering-Plough say that the new revelations from Congress are consistent with their previous accounts of the study and that they have acted properly.
Vytorin is a combination of two cholesterol drugs: Schering's Zetia and Merck's Zocor. The panel on which Stein served was one event on the labyrinthine path between the end of a study called ENHANCE and the final release of its shocking result: the Vytorin combo did no better at preventing heart-attack-causing atherosclerosis than Zocor alone, even though it lowered cholesterol more. Zocor, now generic, is available for a third of Vytorin's price. Although the study is far from definitive, it sent sales of Zetia and Vytorin into decline, and the stock prices of Merck and Schering-Plough into a tailspin.
Stein got his first call from Schering in April 2006, as the last of 720 patients with a rare genetic disease that causes high cholesterol left the ENHANCE study. Ultrasound sensors had been used to take pictures of arteries in the patients' necks and legs. The idea is that noninvasive artery pictures can tell how well a heart drug works faster than the traditional method, counting heart attacks.
Schering researchers showed Stein four images produced for ENHANCE, but they wouldn't tell him what study they were from or what labs produced them. He thought they might be testing him to see if they wanted to use his imaging lab in a future study. "I was honored," he says. He pointed to some problems with the images, and didn't hear back for more than a year.
In January 2007, Michiel Bots, an artery-imaging expert from Utrecht, Germany, was asked to give an independent opinion on the quality of the ENHANCE data. He compared it to other studies and said in a 13-page report the data were "no better, no worse than previously reported in the literature." Schering and Merck still weren't satisfied, and continued to analyze the data to make sure they were correct.
The outside academic running ENHANCE, John Kastelein of Academic Medical Center in the Netherlands, was chafing at the delays. In July, after learning of yet another setback, he wrote in an e-mail to Schering-Plough Vice President Enrico Veltri, "You will be seen as a company that tries to hide something, and I will be perceived as being in bed with you!"
On Aug. 20, Kastelein, Merck and Schering-Plough agreed to convene an expert panel to decide what to do about the study. In September, a Merck executive complained that Schering was not moving fast enough to set up that panel. Veltri told a colleague to "tell him to f' off" in an email.
Stein was asked to serve on that expert panel at the very end of the summer of 2007. He says Merck and Schering initially wanted to hold the meeting that September, but the panelists' packed calendars pushed it back until Nov. 16.
The meeting would be anonymous, and no minutes would be taken. The panelists were all close colleagues of Kastelein's; anonymity would ensure they could speak their minds. Kastelein recused himself. Michiel Bots, who had prepared the 13-page report a year before, was one of the panelists, but Stein doesn't remember being told anything about Bots' previous and more thorough analysis.
Only 75 of the 40,000 or so images collected were shown to the panelists, who agreed that there were problems with them. Stein also pointed to potential problems with the study design and the way the data were gathered. Quality control fail-safes were missing, and there were some questions about the accuracy of measurements.
Two statisticians from Schering-Plough argued to the panel that whatever answer the study gave might well be wrong. Stein says that their argument was technically difficult and that they had trouble articulating their points because they were not native English speakers, but the prospect worried him. Stein remembers that he and the other panelists all gave a wide range of opinions, and no vote was taken on any matter.
On November 19, ENHANCE exploded. Forbes published a story about the delay (see "The Vytorin Question"). Merck and Schering-Plough issued a press release saying that a panel of experts had advised them to change the study's primary endpoint, or main goal--an irregular move. Stein says that was never an accurate characterization of the panel's discussion, although all five panelists did say such a change was permissible.
"We never made a recommendation to change any endpoints. We simply gave our opinion," says Stein. "At the end of the day, most people thought it was reasonable to change the endpoint. But we had no power to change the endpoint. That was in the hands of the principal investigator and the company. We never made that recommendation."
The change was technical--using images from one place in the neck artery instead of averaging several--but so unusual it drew more criticism. On Dec. 15, U.S. Reps. John Dingell and Bart Stupak, both Michigan Democrats, began their investigation into whether ENHANCE was delayed for commercial reasons. The same day, Merck and Schering said they would not change the endpoint after all.
Four days later, Stein received an e-mail from a Schering scientist who works with Veltri containing "minutes" to the anonymous expert panel. Despite the timing, Merck and Schering say they began putting the minutes together not because of Congress but because of a November request by the U.S. Food and Drug Administration.
According to e-mails released April 11 by Dingell and Stupak, Stein objected that the panel was supposed to be anonymous, and he sent back a flurry of revisions, including some related to the endpoint change. "The tone here implies we strongly recommended this," he wrote on Dec. 21, "when in reality, we just advised you on what the scientifically valid approaches would be."
The endpoint change wouldn't have mattered--ENHANCE showed no benefit for Vytorin over Zocor in any set of artery images. A bigger question is whether the images really needed more than a year of cleanup.
Stein had theoretical problems with the study's design, he says, and with the small number of images he saw. But the expected problems didn't materialize in the final results, as presented two weeks in The New England Journal of Medicine and at the recent meeting of the American College of Cardiology (ACC).
"For anyone to say the data in the study are bad, they either need to say that John Kastelein is a liar, which I'm certain he is not, or they need to talk to the [Schering-Plough] statisticians for the study and ask them if there are any problems with the data," Stein says.
In an interview at the ACC meeting, Veltri said he had run through a series of analyses that proved Schering couldn't have known the results of ENHANCE before January. Veltri also argued that the reason the data look so good now is because Schering had put so much effort into cleaning it up. Robert Spiegel, Schering's chief medical officer, said the company had a regulatory mind-set aimed at making sure the data would be up to snuff.
In a prepared statement, Stupak said his energy and commerce committee's investigation is "far from over." He promised to continue to explore "the 20 months between when the trial ended and when the results were released.”
Stein, for his part, no longer knows how to answer questions about Schering's intentions. "I just don't know," he says. "All along I thought I was being asked for my honest scientific opinion. I never saw any signs anyone was trying to delay release of any data. But I only worked with a select group of people from the company."
These days, Stein's big projects are for the National Institutes of Health. He's studying the artery effects of smoking cessation and the heart effects of various HIV drugs, using the same type of technology used in ENHANCE.
"I'm not concerned with what Merck or Schering think of me," he says, "so long as they don't call me dishonest."
