Routine screen hints at fetal death risk.

Some pregnant women with high blood levels of a commonly measured fetal protein may face up to 11 times the usual risk of losing their babies late in pregnancy, a new study indicates. However, since most women with the high protein levels do carry their babies to term, the study's authors view the test only as an adjunct to other methods for monitoring high-risk pregnancies.

The test measures maternal blood levels of alpha-fetoprotein, a substance of unknown function produced by the developing fetus. U.S. obstetricians already assay this protein early in pregnancy in roughly half of their patients, because expectant mothers with extremely high levels run a greater than 80 percent risk of bearing a baby with neural-tube defects. These birth defects, in which the tissue destined to become the fetal central nervous system fails to develop properly, often lead to open spinal cords.

U.S. screening programs for alpha-fetoprotein as a predictor of neural-tube defects began in the mid-1980s. In the course of such screening, obstetricians noted that the fetuses of some women with high alpha-fetoprotein levels died late in the pregnancy -- even if they did not have neural-tube defects.

Those observations led D. Kim Waller of the University of California, Berkeley, and her colleagues to launch a retrospective study comparing the second-trimester alpha-fetoprotein levels of 612 women whose pregnancies ended in fetal death with those of 2,501 women who gave birth to live infants. The researchers discovered that the fetuses of women with double the average protein level were nearly three times as likely to die before birth as those whose mothers had normal alpha-fetoprotein levels. Women with more than three times the average level of this protein faced 11 times the risk of losing their babies late in pregnancy. The team reports its findings in the July 4 NEW ENGLAND JOURNAL OF MEDICINE.

"It hadn't been clear before our study that alpha-fetoprotein could predict a fetal death occurring at term," says Waller.

Surprisingly, 515 of the 612 fetal deaths occurred in women who had less than two times the average level of the protein, she notes. Until now, this was considered within the normal range of individual variation.

However, "this test would only identify between 8 and 10 percent of women destined to have a fetal death ... so it's not a good screening test [for predicting fetal loss]," she adds. Indeed, one-third of the 78 fetal deaths occurring in women with more than twice the average alpha-fetoprotein levels can be attributed to chance, Waller says.

In an editorial accompanying the report, F. Gary Cunningham and Larry C. Gilstrap of the University of Texas Southwestern Medical Center in Dallas point out that "most fetal deaths in this study were not associated with elevated alpha-fetoprotein levels, and most women with elevated levels did not have fetal death." They suggest that obstetricians might prevent some fetal deaths by closely monitoring third-trimester mothers with high alpha-fetoprotein levels and proceeding with delivery if the baby shows signs of stress.