The future of the popular acne-fighting drug Accutane may hinge on the credibility of government estimates that it caused birth defects in as many as 1,300 babies.

Since it was first put on the market in 1982, the government and Accutane's manufacturer, Hoffman-La Roche Inc. of Nutley, N.J., have warned against pregnant women using the drug because of the possibility of birth defects.

Almost from the beginning, some pregnant women and doctors did not observe the stiff warning: Within a year of the first sales of the drug, the company had sent the Food and Drug Administration reports of nine miscarriages and seven babies born with birth defects to women who had used Accutane during their pregnancies. To date, the FDA has received reports of 62 cases of babies with birth defects.

Last week, however, the scientific community was startled by new government figures suggesting that the extent of harm from the drug, which is obtained only through prescription, has been grossly understated.

Researchers with the FDA's division of epidemiology and biostatistics studied records of Michigan Medicaid patients to determine that 928 women of childbearing age were exposed to Accutane between 1982 and 1986, including 52 who took the drug during pregancy. In two cases, the researchers found data suggesting that children had been born with "significant birth defects," as well as one stillbirth and two cases in which researchers suspect "early postnatal death."

Extrapolating from these numbers, the researchers estimated that 16,000 to 23,000 pregnant women nationally were exposed to Accutane between 1982 and 1986, and as many as 900 to 1,300 babies were born with birth defects.

To the FDA researchers, the numbers indicated that past warning efforts -- including changes in the drug's labeling, "Dear Doctor" letters, and patient education campaigns -- have not succeeded. They wrote in a Feb. 11 memo that the data "warrant active consideration of removal of Accutane from the market."

But soon after the memo was disclosed last week, Hoffman-La Roche issued an angry statement calling it "invalid."

"The number of defects is grossly overstated and the statistical methods used to arrive at the numbers would not survive scientific scrutiny," the company said.

And at least one member of an FDA advisory panel of top dermatologists from around the country seems to agree with the company. In an interview Saturday from Boston, Dr. Robert S. Stern, a Harvard professor, said he has "grave reservations about how accurate the {FDA} data is," based on his initial reading of the document.

Stern said he has "real questions" about how the FDA staff verified the birth defects and other adverse pregnancy outcomes culled from the Medicaid data. In addition, he said, the way the FDA staff extrapolated from the data to make national projections is highly questionable.

An FDA spokeswoman said yesterday that the staff study does not "represent {the} agency's position . . . . It hasn't been peer-reviewed or presented to a group of scientists."

On Tuesday, FDA scientists are scheduled to discuss their findings with the dermatological drugs advisory committee at a meeting at FDA headquarters in Rockville. Officials with the company, the Centers for Disease Control, American Academy of Dermatology and Public Citizen Health Research Group are scheduled to make presentations.

Afterwards, the committee could recommend that FDA remove the drug from the market, or any of a range of less stringent recommendations, including asking the company to restrict distribution of Accutane or to step up warning efforts. The drug currently is used by an estimated 250,000 people in the United States annually.

One of the dilemmas facing the committee is that Accutane has proven to have unusual success in fighting severe cystic acne, an often traumatic skin disease that leaves scars across the face, chest and back -- and that disasters seem to have occurred only when doctors or patients ignored explicit guidelines for using the drug.

"If the drug is used correctly, it is a highly effective drug. If it is used incorrectly, it is highly dangerous," said another member of the advisory commitee, who asked not to be identified. "You are allowing a few people to ruin it for the people who are using it correctly. That's not fair."