LOS ANGELES - Researchers believe they have come up with a way to tell whether a newborn infant has a higher-than-normal risk of developing autism - by looking for abnormalities in the placenta shortly after birth.

The abnormalities in question are called trophoblast inclusions, or TIs. They are created when the placenta doesn't develop properly, and they are a marker for various genetic abnormalities. When a placental sample is examined on a slide under a microscope, TIs appear as dark blobs.

Dr. Harvey J. Kliman came up with the hypothesis that TIs might be linked to autism after he was asked to examine two placentas with many abnormalities. It turned out those placentas belonged to children with autism.

To see whether his inkling had any merit, Kliman, a research scientist in obstetrics, gynecology and reproductive sciences at the Yale School of Medicine, found 13 children with an autism spectrum disorder and 154 children who didn't. Then he compared samples from their placentas and found that TIs occurred three times more often when children had an ASD. He and his colleagues reported those results in 2007.

The new study was more ambitious. Kliman teamed up with researchers from the University of California-Davis MIND Institute. They found 117 pregnant women who already had a child with autism and thus were more likely to have another child with the disorder. The placentas from these high-risk pregnancies were compared with placentas from 100 women who had no heightened risk of having a baby with autism.

It turned out there was a marked difference in TIs between the two groups. A full 92 percent of the control placentas had no TIs whatsoever, and none of them had more than two TIs.

On the other hand, placentas from the high-risk pregnancies had as many as 15 TIs. Only 59 percent of placentas had none. That meant that if the researchers found two or more TIs in a patient's sample, the odds that the baby would have autism increased by a factor of 8.

Then the researchers ran the numbers again, this time using only the high-risk women who were most similar to the women in the control group. In this comparison, the presence of two or more TIs meant the risk of autism increased by a factor of 11.5.

The results were published Thursday in the journal Biological Psychiatry.

The test can't say whether a particular infant will grow up to develop autism. At best, it can only serve as a warning that a particular baby has an increased risk of developing autism.

Still, that information could be tremendously useful to parents. Research suggests the sooner parents can begin performing behavioral interventions - eye contact exercises, for instance, and using children's names - the more their autism symptoms will fade.