WHO'S YOUR DADDY

Don't count on DNA testing to tell you

From the science magazine Discover, Vol. 27 No.04, April 2006, p. 68: , By Susan Kruglinski

When celebrity hairstylist Andre Chreky was hit with a paternity suit by
a woman he had not been involved with for years, he was certain he
couldn't lose. Paternity tests are DNA tests, he thought, and DNA tests
never lie. So he unhesitatingly submitted a swab of cells. To his shock,
he was positively identified as the father, with a 99.99 percent certainty.

But last April, after a two-year legal battle that cost Chreky $800,000,
the Fairfax, Virginia, circuit court found that human error in the
testing was probable and that the DNA results were incorrect.

"It hurt my family; my business," Chreky says. "My life will never be
the same."

DNA testing is thought of as definitive. If there is a match between two
samples, then identification is certain. Some DNA experts place the
probability of an error at one in a billion. But recent cases in which
paternity tests were proven to be inaccurate suggest the odds may be
much less certain.

Only a millionth of a person's genetic sequence is examined in a DNA
test. but that tiny portion includes more than a dozen locations with
unique repeating sequences. The sequences are laid out in
black-and-white strips (left) to form patterns more individual than
fingerprints (which, interestingly, have never been proven to be
unique). A so-called paternity index is calculated using the number of
matches between two sets of DNA as well as the likelihood of matches
within a sub' ject's ethnic group. If the resulting index is 300, for
example, the odds that the person is not the father are 1 in 300. That
figure is then recalculated using a simple probability equation to
deliver a percentage that is more understandable in a courtroom. If the
report comes back with an index of 300, the probability that the person
is the father is 99.67 percent. The index does not include the
possibility of error, genetic anomaly, or mitigating circumstances.

"Paternity testing is a human endeavor, and it is complicated," says
Karl Reich, scientific director of Independent Forensics of Illinois, a
DNA research and analysis laboratory. "Nobody really knows the
human-error frequency, and the question is: Can you find these errors
and reduce them to a statistically meaningful level?"

In the case of Chreky, the judge ruled that LabCorp, one of the largest
paternity labs in the country, had performed "shoddy" work. An employee
testified that during his 10-hour shift, he issued an average of one
paternity report every four minutes. Mislabeling, misinterpretation, and
switched samples are not factored into the probabilities.

"Every time there is a transfer point, errors can occur," says William
Thompson, a criminologist at the University of California at Irvine.
"Errors are embarrassing, and labs are not forthcoming about them. But
we know that they do occur."

Human error was not the cause of Lydia Fairchild's unexpected maternity
test results in 2003. A mother of two at the time, Fairchild submitted
her DNA to establish maternity when applying for welfare, says Alan
Tindell, her former lawyer. But the results that came back said she was
not the biological mother of her children, placing her in danger of
losing them. When she gave birth to a third child with DNA that did not
match her own, Tindell says, judges and lawyers involved in the case
were stunned.

Fairchild could not be reached for comment, but Tindell says she seems
to have two sets of DNA. One set matched poorly with her children's DNA,
as if she were their aunt. Fairchild may be a tetragametic chimera. A
chimera is any animal with more than one set of genetically distinct
cells, each set originating from a different fertilized egg. Fairchild's
body could be the result of the fusion of two nonidentical embryos,
which can occur at the earliest stage of development. Different parts of
her body could have come from different cell lines.

The eggs that produced her children may have been spawned from tissues
that came from one of those cell lines; the DNA taken from her mouth for
testing would have been from a different cell line.

Human bodies are imperfect machines in which a range of biology can take
place. Fused embryos may be rare, but cell swapping and genetic mutation
are as natural as crooked teeth or double-jointed fingers. During
pregnancy, mother and fetus may swap blood cells, and twins in utero
often do so. A recent autopsy study suggests that swapped cells can
migrate to organs and transdifferentiate, residing peacefully with
normal cells in the kidneys, liver, heart, or any other organ.

For a paternity test to fail, the gonad producing the egg or sperm must
contain tissues that are genetically different from the blood or tissue
tested. Whether a gonad and a test site will match is pure chance. As
far as anybody knows, the genetically different tissues in a human
chimera materialize as unpredictably as the orange, black, and white
patches on the coat of a calico cat. Many of the known cases of human
tetragametic chimerism have been discovered because the person was born
with patchwork skin, or with different eye colors, or as a
hermaphrodite. Others have no visible signs.

"For all we know, there could be chimeras out there who aren't
recognized because the condition hasn't caused them any problems," says
Lynne Uhl, a pathologist at Beth Israel Deaconess Medical Center at
Harvard University.

Twins researcher Charles Boklage, a biologist at East Carolina
University, says chimerism is underdiagnosed: "The great majority of
people who are spontaneous chimeras will never be detected by any means
whatever. It's a spooky thing. It's very difficult to find when it's
there." He estimates that about 15 percent of people were conceived
alongside a twin who was then lost. As multiple conceptions grow with
the rising popularity of fertility drugs, the percent of the population
with double DNA may increase.

Simple genetic mutations that take place during mitosis can also cause a
discrepancy in DNA test comparisons. "When a fertilized egg divides,
there's a chance that one of the repeats gets lost or increases," says
Reich. "That's a mutation, and it does not show up in the parent. It's
much more common than anyone thought."

Mark Stolorow, executive director of one of the largest and oldest
genetictesting companies, Orchid Cellmark, says that years of
experimentation have perfected testing. "We have selected sites that are
reasonably stable," he says. "The mutation rate for any one location is
less than 1 percent." About 300,000 paternity tests are conducted each
year in the United States.

"Every particular DNA variation that you can think of is in the
population and at a reasonable frequency," says Reich. "So in the case
of paternity testing, you can never be sure that a man is excluded
unless you look at every genetic element. But you can't."

Thompson, who says DNA testing is very reliable, warns that "even a low
rate of error can significantly undermine confidence in the results if
you are doing hundreds of thousands of these tests."

"What particularly disturbs me," Judge David T. Stitt stated in the
summary of the Chreky case, "is that most of the people coming through
our system, particularly criminal defendants, do not have the resources
to mount the kind of challenge to the DNA test results as was done in
this case.... DNA test results are typically accepted at face value by
the court in criminal and civil cases."