Suit Alleges Pfizer Spun Unfavorable Drug Studies

Pfizer Inc. marketers urged the suppression of medical studies that reached unfavorable conclusions about the effectiveness of the company's big-selling drug Neurontin, according to internal Pfizer documents submitted in a lawsuit against the company.

In 2004, Pfizer's Warner-Lambert unit pleaded guilty to felony charges that it promoted Neurontin for uses not approved by the Food and Drug Administration, including bipolar disorder and chronic nerve pain. The FDA originally approved the drug as an antiseizure treatment for epilepsy and in 2002 for one kind of pain related to shingles.

Pfizer paid $430 million to resolve the charges and reimburse state Medicaid programs for unapproved, or off-label, uses of Neurontin. Pfizer said it made sure there was no improper marketing after it purchased Warner-Lambert in 2000. Pfizer has booked about $12 billion in Neurontin sales since then and, though the drug is now subject to generic competition, it remains a strong seller.

Documents and emails released this week in the case in U.S. District Court in Boston suggest Pfizer's marketers influenced the drug's scientific record to boost sales at least until 2003 by declining to release or altering the conclusions of studies that found no beneficial effect from Neurontin for various off-label conditions. The case consolidates lawsuits by health insurers and consumers seeking refunds for their Neurontin expenditures, based on alleged civil fraud. Plaintiffs are seeking $4.9 billion.

In a statement, Pfizer said it was "committed to the communication of medically or scientifically significant results of all studies, regardless of outcome." The company pointed to examples where it published studies of Neurontin with negative outcomes.

According to documents in the Boston case, a European study done in the late 1990s by Warner-Lambert to measure Neurontin's use for diabetic nerve pain produced consternation at Pfizer after it failed to find a significant effect. "I think we can limit the potential downsides of the ... study by delaying the publication for as long as possible," wrote Michael Rowbothan, then Neurontin's marketing team leader, in a 2000 email sent after Pfizer bought Warner-Lambert. He added that "it will be more important to how WE write up the study."

The study's scientific manager, Beate Roder, wrote in an email to employees after Pfizer agreed to buy Warner-Lambert that she had been instructed "that we should take care not to publish anything that damages Neurontin's marketing success." Neurontin has been widely prescribed for diabetic nerve pain, according to market researchers.

In 2002, Angela Crespo, then Neurontin's senior marketing manager, emailed an outside firm that was contracted to write up the study's results: "We are not interested at all in having this paper published because it is negative!!" Pfizer declined to make the three employees in the emails available for interviews.

The company's writeup of the study was subsequently rejected by two medical journals. Some of the reviewers for these journals said the company was putting too rosy a spin on the results, the court documents show. The paper was never published, although Pfizer later summarized some of the study's results in a broad review published in 2003 in an obscure medical journal, Clinical Therapeutics.

That review excluded another study that Pfizer conducted in Scandinavia, which examined Neurontin's effectiveness for post-operative nerve pain. The study found that Neurontin didn't work measurably better than a sugar pill, according to a Pfizer internal report. The study's finding has never been published, though the analysis was completed in 2003. Pfizer presented a summary of the study at a 2002 conference.

By declining to publish negative studies, "Pfizer created the illusion of Neurontin's efficacy in the scientific record" that encouraged doctors to use the drug even where it doesn't work, said Tom Greene, a Boston attorney heading the litigation. U.S. District Judge Patti Saris in Boston last year denied Mr. Greene's motion for class-action status; the plaintiffs are seeking a second try in her court this week.

"Ethically, one is compelled to publish results," said Kay Dickersin, a professor of epidemiology at Johns Hopkins University in Baltimore, who wrote a report for plaintiffs that was submitted in the case. Ms. Dickersin said her fee will be paid to Johns Hopkins to set up a fund to distribute Pfizer's documents.

Ms Dickersin said that in exchange for being experimented upon in trials, patients are told they are contributing to human knowledge. To withhold negative results from the public breaks that ethical obligation to such patients, she argued. Pfizer says it has committed to publish all medically significant results.

In 2004, Pfizer completed a five-year study of Neurontin on patients suffering from bipolar disorder in Spain. A 2004 internal Pfizer report summarized the finding: The study "did not show statistically differences between Gabapentin and Placebo." (Gabapentin is the generic name for Neurontin's chemical formula.)

But in 2006, a Spanish psychiatrist, Eduard Vieta, who has been a paid consultant to Pfizer, published a report on the same study in the Journal of Clinical Psychiatry along with a Pfizer employee, Katia Verger, and other scientists. Their report claimed a statistically significant benefit. One difference: Dr. Vieta modified the scale used to assess how severely patients exhibited the symptoms of bipolar disorder, also known as manic-depressive disorder. That change increased Neurontin's score in the study. The change wasn't disclosed in the journal. Making such a change after a study's parameters were set, especially without disclosing the change, is generally considered improper by scientists.

Dr. Vieta's article also said it included results from all patients the doctors in the study intended to treat, a principle known as "intention to treat." But in reality, Dr. Vieta's article excluded 16 patients who hadn't complied with the study protocol, such as by not taking all the pills they were given.

The two modifications had the effect of turning a failure into a success.

In an interview, Dr. Vieta, of the University of Barcelona, said the modifications to the formula were requested by the journal and were a more-conservative technique. "Nobody in Pfizer wrote the manuscript," he said. He said the company's 2004 report of the study "takes a broader interpretation of the intention-to-treat concept" but that it was appropriate to exclude some patients from the analysis.

Pfizer said that "an implication that Pfizer improperly influenced any aspect of the published Vieta study for commercial reasons is without basis," adding that the drug lost its patent protection and top-selling status in 2004.

Write to Keith J. Winstein at keith.winstein@wsj.com