Researchers from Johns Hopkins University have described a new approach for non-invasive prenatal trisomy diagnosis that the group argues is simpler and more efficient than other sequencing-based aneuploidy detection methods.
The team published a paper in PLoS One last month discussing the approach — a sample preparation and sequencing method dubbed Fast Aneuploidy Screening Test-Sequencing System, or FAST-SeqS.
According to the researchers, from the Ludwig Center for Cancer Genetics and Therapeutics at the Johns Hopkins Kimmel Cancer Center, FAST-SeqS — which uses a single primer pair to cull and amplify sections of the genome that occur on every chromosome with small, identifiable differences — offers "significantly increased throughput and decreased cost" compared to current sequencing methods that require more extensive sample and library preparation.
Using the approach in a series of mixing experiments, the group reported it could distinguish samples containing as little as four percent trisomy 21 DNA from normal samples. The researchers are now planning a trial of human subjects to better demonstrate the technique's potential.
Isaac Kinde, the first author of the John Hopkins group's paper, told Clinical Sequencing News that while the paper represents only the earliest tests of the method, the group is excited about its potential.
"This has all been just mixing things together in a lab, so definitely just proof-of-principle experiments. We have a while to go before this can be used in the clinic," he said. "But we consider it certainly something very promising."
Several companies have recently advanced methods for non-invasively detecting fetal aneuploidies by sequencing DNA in maternal plasma. Sequenom, the first to commercialize such a test, licensed its method for detection of cell-free fetal DNA in a mother's bloodstream from the Chinese University of Hong Kong in 2005.
Several other companies have since entered the competitive market, including Verinata, whose method is similar to Sequenom's, and Ariosa Diagnostics, which takes a more targeted approach. Sequenom has taken an aggressive stance in defending its intellectual property, which it has argued precludes other firms from developing non-invasive sequencing-based diagnostics for fetal aneuploidies.
Sequenom filed a lawsuit in January against Ariosa for patent infringement. In turn, Ariosa, Verinata, and another firm, Natera, have all sued Sequenom.
Amid this contentious IP landscape, the Hopkins team is proceeding cautiously with regard to its development plans for FAST-SeqS. "We'd like to prepare for the best case scenario [in all this]. That's our perspective," Kinde said. "If it works out that competition is allowed in the field, we'll be ready and have things put in motion [to commercialize our technology.]"
Kinde said that the FAST-SeqS approach grew out of the team's research on the early detection of cancer and is a modification of a technique called Safe-SeqS that the group developed last year to detect and quantify rare mutations (CSN 6/1/2011).
"We realized this technique to count molecules could be applied to non-invasive trisomy detection. That was how the thought process started," he said.
"Another thing our lab thinks about a lot is how to actually translate a technique into something that would be useful in the clinic. So ease, efficiency — these are things we think about a lot," he said.
Kinde explained that FAST-SeqS is significantly simpler than current approaches to maternal plasma DNA sequencing and differs largely in its sample preparation methods.
In the case of Sequenom and Verinata, "they take DNA fragments in an unbiased way from the whole genome, whether it be the mom or the fetus. Then they align those to the genome and they count," he said. In the Ariosa method, meantime, "they actually hybridize specific oligos to each region that they want to quantitate."
But in either case, he said, "the end result is the same. You have different pieces of DNA from different regions of the genome and you look for differences in terms of chromosome representation … [And] in both of these techniques, there are a lot of steps involved in preparing the samples."
According to Kinde, the Hopkins method's advantage lies in the simplicity of the pre-sequencing steps.
FAST-SeqS involves PCR amplification using a primer pair the group designed to hybridize to many different regions throughout the genome, including "every nuclear chromosome," Kinde explained.
"We wanted to get to the same end result of having DNA fragments from many different chromosomes, but without all the steps," he said.
"The thinking was, if you can find very small regions of a genome that are very, very similar but not identical to other regions on different chromosomes … you can design primers that flank these regions at positions that don't change throughout the genome, but in between those primers there are small differences that will clue us in to whether this is coming from one chromosome or another."
Kinde said the group started off looking for small regions on chromosome 21, knowing that was their target, and in a "brute force manner," looked to see if any 150-base-pair sequence chunks might fit the bill, eventually narrowing down to three, designing primer pairs for them, and then again narrowing down to one best-performing primer pair.
Once the group had its primers in hand, it proceeded to test the method. The researchers first looked at eight plasma samples from seven normal women to establish that the sequencing method did not produce false positives — in this case, so-called z-scores more than three standard deviations from the mean. According to Kinde, the cutoff at a z-score of three is a widely used standard.
In the test, the researchers found that the maximum z-score among the eight samples for chromosomes 21,18, and 13 were 1.3, 1.4, and 1.0, respectively — significantly below the cutoff.
Next, the group recreated this in another eight samples, this time of white blood cells. "We saw the same type of reproducibility," Kinde said. "No individual chromosome from the samples was greater than three or less than negative three."
Kinde said the group used an Illumina HiSeq 2000 for these first two experiments. They then looked at a third set of eight samples using half a lane of an Illumina GAIIx. "We got [fewer] tags per sample," Kinde said — almost three-fold less sequencing, according to the group's paper — "but still the data was just as reproducible. There weren’t any outliers," he said.
Overall, he said, "this was the first real indication that this technique was going to be usable, because in normal samples, it didn't produce a false positive. Of course that could also mean it just doesn’t work at all … But that was our starting point."
The researchers then moved on to establish that the method could identify aneuploidies, first testing four samples with trisomy 21, two with trisomy 18, and one with trisomy 13, then testing the ability of FAST-SeqS to identify the presence of trisomies in lower and lower concentrations.
"This is important," Kinde said, "because if we want to be able to detect a fetus with an aneuploidy, the earlier the better. You want to give the families time to prepare whether they are going to take action or not."
In the full trisomy experiment, the group found that z-scores were "through the roof," Kinde said, ranging from 32 up to 56.
Then the team performed a mixing experiment. "One thing you should expect to see in a technique that can accurately detect low and high levels of trisomy is some sort of dose dependence," Kinde said. "If you mix samples at different proportions, the samples with higher proportions of trisomy DNA should give a higher z-score than samples with lower proportions."
In this test, the group evaluated eight samples mixed from the DNA of one trisomy patient and one normal DNA sample to create two normal controls, two samples with 5 percent trisomy DNA, two with 15 percent, and two with 25 percent.
"We saw pretty much the expected percentages," Kinde said. "You can see the line fits with the data."
Finally, the researchers measured whether FAST-SeqS could pick out trisomies in samples with only 4 percent trisomy DNA. "For each one of the cases, whether it was 0 percent, 4 percent, or 8 percent, we could accurately distinguish them from the normals, so this was a really encouraging experiment for us," Kinde said.
Before the researchers can translate these initial results into the clinic, they'll have to test the method on real pregnancies, Kinde said. Right now, the team is deciding how best to proceed with a large-scale study of pregnant women.
"One of the things we're thinking about is, if we partner with a company or organization that already had samples and we could do it quickly, perhaps we would do a retrospective study," Kinde said. "But it's also possible that we could collaborate with a group that sees a lot of high-risk pregnancies and do a prospective trial. We're open to either one."
Most importantly, he said, the group wants to be able to bring its method to the clinic "as quickly as possible."
He said the team would be "pretty happy" if it could get such a study done in the next few years.
According to Kinde, the method could theoretically work with a variety of sequencing technologies. His lab used Illumina because that is what the researchers were most familiar with, and to maintain consistency, they would likely continue using Illumina machines for a future large study.
Using the MiSeq, he said, could speed things up. "If you're thinking about having the most rapid test, it might be nice not to have to send your samples out, so one approach to implementing this test could be to harness the power and speed of a MiSeq, which would give us plenty of sequencing depth, and could be performed in many labs."
And because their method reduces the number of steps, "this is something you can start thinking about many clinics implementing on their own," he said. "Obviously a kit would have to be prepared and standardized. But we are hoping this is simple enough [so] that is something that is feasible."