It's a vexing question: When an infant or very young child develops leukemia, how did it happen? Fewer than 5 percent of pediatric leukemia cases are hereditary. And if cancer is caused by a series of genetic mutations in key regulatory genes that, as much of the data on adult cancers suggests, may take years to develop, how does this happen so rapidly in these children?

For some types of pediatric leukemia, plausible answers have begun to emerge. But it's clear that researchers are just scratching the surface of a complex disease development process that may be influenced by factors such as the contents of a pregnant woman's diet or a child's ear infection history.

One of the more remarkable findings is that many cases of infant and childhood leukemia - diagnosed at 0 to 18 months and 18 months to 14 years, respectively - appear to be initiated prior to birth.

The most convincing evidence of in utero origin, explained Dr. Julie Ross of the University of Minnesota Department of Pediatrics, comes from analyses of archived newborn blood (ANB) samples - the blood samples generated by heel pricks of newborns that are used to screen for disorders like PKU - from infants and children who later developed leukemia.

"Molecular studies of ANB cards have identified mutations that were also found in those children" at the time of diagnosis, Dr. Ross said.

These analyses have turned up what have become the hallmark genetic mutations of pediatric leukemia: specific chromosomal translocations or the presence of too many chromosomes, often referred to as hyperdiploidy.

However, added Dr. Ross, who is leading the largest-ever epidemiologic study of infant leukemia, the available evidence doesn't suggest that all pediatric leukemias are initiated in utero.

That is supported by analyses of ANB blood spots from children with leukemia that failed to detect the genetic mutations associated with their particular cancer.

"These could be postnatally initiated cases," said Dr. Mel Greaves of the Institute of Cancer Research in London. "Or it is perfectly possible that many of these cases are indeed initiated in utero, but the blood spots are negative because there are either too few preleukemic cells in blood at birth to detect…or the molecular marker used is not the initiating event."

The number of demonstrably proven in utero-initiated cases is most likely "underestimated by the available technology," Dr. Greaves argued.

Identifying the Transforming Event

The conclusion that some pediatric leukemias have an in utero origin raises a tough question - one that derives from Dr. Alfred Knudson's widely accepted "two-hit" model of cancer development that (at least) two genetic mutations, or hits, are needed to transform a healthy cell into a cancer cell: In a case in which the first hit occurs in utero, what causes the second, postnatal hit?

For childhood leukemia, the available evidence suggests that chromosomal translocations or hyperdiploidy on their own likely are not enough to cause cancer. In a cross-sectional study published in 2002, for instance, a specific chromosomal translocation (TEL-AML1) often seen in childhood acute lymphoblastic leukemia was also present in cord blood samples from children who had not developed disease. The study authors concluded that for every child who developed TEL-AML1-positive leukemia, 100 children with this translocation would never develop disease.

There is mounting, yet still inconclusive, evidence to support a theory put forth by Dr. Greaves that the transforming event is an abnormal immune response to a common infection. This would especially hold true for children who have had extremely limited exposure to infectious agents such that, when their immune system is finally confronted by a significant infectious threat, it overreacts, fostering genetic instability among the immune system cells in which leukemias take hold.

Several epidemiologic studies, including ongoing large case-control studies led by Dr. Greaves and Dr. Joe Wiemels of the University of California, San Francisco, have put this theory to the test using day care attendance or levels of social activity among other children as a proxy measure of exposure to infectious agents.

The results have been remarkably consistent: The more time spent in day care or engaging in social activity with other children, the lower the risk of childhood leukemia.

"The evidence is indirect but, overall, I believe, persuasive," Dr. Greaves said. "But we still lack unambiguous proof," he continued, such as "identification of a cellular/biochemical mechanism by which an abnormal immune response might trigger and select preleukemic clones."

The situation may be entirely different for infant leukemia, added Dr. Wiemels.

By Carmen Phillips