An investigation into whether fraud played a role in an influential report on the health effects of hormonelike chemicals has drawn to a murky close. In a letter on page 1932, the chancellor of Tulane University Medical Center in New Orleans announces that endocrinologist John McLachlan "did not commit, participate in, or have any knowledge of any scientific misconduct" in preparing the paper, which was published in Science 3 years ago and later retracted (25 July 1997, p. 462). The conclusions about the researcher who conducted much of the work are not so clear-cut, however: Tulane found that Steven Arnold's data fail to support "the major conclusions" of the paper.
Questions about the paper, which claimed that mixtures of pesticides could have potent hormonal effects, have reverberated partly because of the prominence of its senior author, a former scientific director of the National Institutes of Health's National Institute of Environmental Health Sciences in North Carolina. "I'm just glad it's starting to clear up for John McLachlan," says Earl Gray, a reproductive toxicologist at the Environmental Protection Agency's health effects lab in Research Triangle Park, North Carolina. Arnold, who resigned from Tulane in 1997, eventually found work at the Roswell Park Cancer Institute in Buffalo, New York. A co-worker said last week that Arnold had just finished his last day and was planning to begin business school in the fall. Arnold could not be reached; McLachlan, through an assistant, declined to comment.
In the paper at issue, the Tulane team used yeast cells with a human estrogen receptor to test the potential estrogenic effects of different compounds. They found that pairs of several pesticides were 1000 times more potent at triggering estrogenic activity than were individual chemicals on their own. The prospect that pesticides could mimic the female sex hormone raised alarm bells among toxicologists and environmentalists and helped convince Congress to include provisions in two 1996 laws requiring manufacturers to screen thousands of chemicals on the market for estrogenic activity. Within a few months after the Science article appeared, however, other labs reported that they could not replicate its results. In July 1997 the authors retracted the paper, and that August Tulane announced an inquiry into what happened.
Although it absolved McLachlan, a Tulane faculty committee "concluded that [Arnold] provided insufficient data to support the major conclusions of the Science paper" and that the "independent review of Arnold's data does not support the major conclusions," writes chancellor John C. LaRosa. This ambiguous denouement--neither exoneration nor a misconduct finding--is not surprising, says Chris Pascal, acting director of the federal Office of Research Integrity. A decade ago, universities were "always trying to find one or the other," he says, but now they "realize there's a lot that falls in a gray area."
Whether McLachlan's lab should have kept better tabs on its raw data is another question. "Most institutions don't enforce data retention or data recording," Pascal says. But Gray says he wouldn't expect a lab chief to check the underlying data "if you have a lot of trust in somebody in your lab." The study's sponsor, the W. Alton Jones Foundation in Charlottesville, Virginia, says it's ready to close the books on the affair.
As for the hypothesis that hormonelike chemicals are dramatically more potent in combination, "it's kind of fallen by the wayside since that paper was retracted," says Gray. But he and others are convinced it's important to test chemical mixtures, because in the real world mixtures abound and the results can be additive. Says John Sumpter, a reproductive toxicologist at Brunel University in the United Kingdom: "I don't think there's any disagreement on that."
Comments on this posting?
Click here to post a public comment on the Trash Talk Bulletin Board.
Click here to send a private comment to the Junkman.