The New York Times, February 26, 2016
Marina Stajic worked for nearly three decades as director of the forensic toxicology lab at the medical examiner’s office in New York City. Last week Dr. Stajic, 66, filed a lawsuit against the city, claiming she had been forced into retirement last year in part because of a disagreement with her superiors over the accuracy of certain DNA tests.
There is more at stake here than Dr. Stajic’s retirement. The cutting-edge technique at the center of this legal dispute, called low copy number DNA analysis, has transformed not just police work, but also a range of scientific fields including cancer biology, in vitro fertilization, archaeology and evolutionary biology. Yet some of the technique’s applications have triggered scientific controversy.
The medical examiner’s office has become a strong advocate for the technique. It is the only public lab in the United States that uses low copy number DNA to develop profiles for use in criminal cases. But experts have long warned that investigators must take particular care in interpreting these tests: analyzing so few DNA molecules can lead to errors.
When scientists first began deciphering DNA in the 1970s, they needed large amounts, because the chemical processes that were involved destroyed most of the source material.
In 1983, the NobelPrize-winning biochemist Kary Mullis sped up the process with a kind of photocopying machine for DNA called polymerase chain reaction, or PCR. Dr. Mullis showed the world how to make millions of copies of any particular genetic fragment.
PCR made it possible for scientists to work with DNA in smaller samples, since they could now make more of it. Over the past four decades, researchers have come up with ways to run ever-more-sensitive tests. In the most extreme of these, experts can reconstruct the entire genome using DNA fragments extracted from a single cell.
Similar tests allow scientists to study DNA from fossils dating back thousands of years. Over time, genetic material in fossils degrades until it becomes scarce. Yet scientists only need to rescue trace amounts for analysis.
“We can be down to 10 or 20 molecules,” said Brian Kemp, a geneticist at Washington State University, who regularly extracts DNA from tiny fish vertebrae more than 10,000 years old.
Scientists have found a number of ways to make their tests more sensitive. For instance, Dr. Kemp and his colleagues have discovered that collagen and other compounds found in fossils can slow down the PCR process. By carefully removing those chemicals, they are able to make more copies of ancient DNA.
But the more sensitive DNA tests become, the greater the risk that they will yield the wrong result. Even stray bits of DNA — from a lab worker’s skin cell or an airborne fungal spore — may contaminate the test equipment.
When scientists then run PCR reactions, they may make millions of copies of the contaminating DNA along with the genetic material they want to study. Even a little contamination can skew results.
Scientists have developed many safeguards to prevent contamination. Stephen R. Quake, a biologist at Stanford University, and his colleagues have shrunk down the equipment they use, so that there is less room for contaminating cells to invade.
“Whatever’s floating around in a test tube, you’re going to have a thousand times less of it,” Dr. Quake said, referring to the miniaturized setup.
Like other scientists, forensic researchers also are using less and less DNA. Now investigators can get usable genetic material just by wiping gun grips or other surfaces for a few loose skin cells.
But low copy DNA analysis can detect a mix of DNA from more than one person, and it can be hard to tell which of them is relevant to a crime.
“Maybe there’s not three people bleeding on a steering wheel, but there are three people touching it,” said Kirk E. Lohmueller, a geneticist at the University of California, Los Angeles. “Before, you didn’t have to worry about that.”
People can even leave DNA traces on objects they haven’t touched. In the January issue of The International Journal of Legal Medicine, German researchers vividly illustrated this problem.
They rubbed a cloth on people’s necks, and then gave the cloths to a second group of people. The second group rubbed their hands on the cloth, picking up the DNA from the first group, then handled a plastic bag or a cotton cloth.
When the scientists examined the bags and cloths, they found DNA from the first group of subjects about 40 percent of the time.
Low copy number DNA analysis can also give incorrect results when it pushes PCR chemistry beyond its limits. There can be so few DNA molecules floating around in a test tube that they simply don’t bump into the PCR chemicals needed to replicate them. As a result, some of the original DNA may go unduplicated.
This failure can yield a genetic profile that doesn’t match the source of the DNA.
Each of us carries two copies of every gene. Sometimes the copies are identical, but not always. If a suspect has two different versions of a gene and PCR duplicates just one, then a forensic scientist may conclude both copies are identical.
This error, called allelic drop-out, doesn’t mean that low copy number DNA is useless. Statisticians are developing methods to calculate the reliability of PCR results for each gene. Those methods should help investigators figure out how confident to be in their overall analysis.
Last July, defense lawyers in a Brooklyn murder trial challenged DNA evidence that came from a bicycle handle. They offered testimony from experts who criticized the medical examiner’s methods, such as the way they calculated the odds of allelic drop-out. The judge threw out the DNA results.
Bruce Budowle, the executive director of the Institute of Applied Genetics at the University of North Texas Health Science Center, served as an expert witness for the defense. (His fee was put toward his university’s program for student stipends.) Despite delivering harsh criticisms of the medical examiner’s office on the stand, he sees a lot of promise in low copy number DNA analysis.
He’s using it in his own research, which includes identifying remains of Civil War soldiers and skeletons discovered in Deadwood, S.D. And Dr. Budowle, like other experts, is working on new methods to improve the technique’s sensitivity.
The problem, he said, is that criminal investigations leave far less room for error.
“I’m not pleased with what’s been done with low copy number in forensics to date,” he said. “But if we get better interpretation, I think it could be a better system.”
Copyright 2016 The New York Times Company. Reprinted with permission.