A prominent researcher who reviewed a critical study on the diabetes drug Avandia for a major medical journal leaked the findings before publication to the drug's maker, GlaxoSmithKline P.L.C., according to the journal Nature.

The reviewer, Steven M. Haffner, a professor of internal medicine at the University of Texas Health Science Center at San Antonio, breached confidentiality rules of the New England Journal of Medicine by faxing the study to a friend working for GlaxoSmithKline, in Upper Merion.

The leak was also confirmed by U.S. Sen. Charles E. Grassley (R., Iowa), a noted industry critic who yesterday released a letter he sent to GlaxoSmithKline asking for a detailed accounting of how the firm handled the fax.

Grassley also cited FDA documents showing that Haffner had received at least $75,000 in consulting fees and honoraria from GlaxoSmithKline since 1999.

Haffner was not returning media calls yesterday. "Why I sent it is a mystery," he told Nature. "I don't really understand it. I wasn't feeling well. It was bad judgment."

The article Haffner reviewed was written by Cleveland Clinic cardiologist Stephen Nissen and concluded that Avandia raised the risk of heart attack. After Nissen's study appeared online on May 21, Glaxo's stock sank that day nearly 8 percent and has continued downward. The decline in Avandia, then the firm's second-biggest seller, has helped lead GlaxoSmithKline to lay off thousands of workers worldwide.

Nancy Pekarek, a company spokeswoman, said Haffner faxed the article May 3 to Alexander Cobitz, senior director of metabolism in new medicines development, based in Upper Merion.

"He got the fax because the reviewer had told us he had concerns about the methodology used in the [Nissen] analysis," Pekarek said. "It was a request for technical advice.

"We actually decided it was inappropriate for GSK to respond," Pekarek continued. "We thought it was better to have independent comment."

No senior GSK executive profited from the information by selling shares between May 3 and 21, Pekarek said. "I haven't analyzed the entire company," she added. One of the firm's U.S. headquarters is in Philadelphia.

Confidentiality is key in medical journals. Without it, "somebody else could scoop the conclusion by publishing their own work more quickly," said Christine Laine, senior deputy editor of the Annals of Internal Medicine, the nation's largest specialty journal, published in Philadelphia.

"I worry that people are not always following the confidentiality policy completely," Laine said. "It probably happens a lot more than comes to our attention. We would certainly not use that reviewer anymore." The journal would also notify the reviewer's institution about the "ethical breach," she said.

The New England Journal yesterday issued a statement, saying "any breach of ethics by a reviewer would be taken very seriously by the editors, but would be handled as a private matter."

Last year, the journal slapped another reviewer, Columbia University cardiologist Martin B. Leon, for disclosing an article's findings on heart stents at a medical conference before the article appeared. The journal banned Leon from reviewing articles for five years and said he could not publish commentary during that time.

In San Antonio, medical school dean William L. Henrich also issued a statement, saying: "This issue has just come to light on our campus. We are embarking on a complete investigation of the facts. Once the facts are understood, we will take swift and appropriate action."

Haffner was a co-author of a 2006 New England Journal article that was highly supportive of Avandia. In that study, financed by GlaxoSmithKline, Haffner reported receiving grant support and consulting and lecture fees from the company.

Haffner is a national expert in diabetes. His university's Web site calls him "one of the highest-funded investigators, in terms of [National Institutes of Health] funding, in Health Science Center history."

Contact staff writer Karl Stark at 215-854-5363 or kstark@phillynews.com.