By Andrea Anderson
Single-molecule sequencing provides slightly better distinction between fetuses with or without Down syndrome in non-invasive, sequencing-based tests of DNA from maternal blood samples compared to sequencing instruments that rely on PCR-amplified DNA, according to a recent study.
Nevertheless, while the study found that the Helicos Biosciences HeliScope offered higher resolution than the Illumina Genome Analyzer, the investigators said that recent improvements to Illumina's HiSeq 2000 makes it a more cost-effective option for non-invasive prenatal trisomy testing at the moment.
The study, led by researchers at Leiden University Medical Center, included 20 pregnant women at nine weeks gestation or later and was published in Clinical Chemistry earlier this month.
In their head-to-head comparison of the HeliScope and the GA for identifying fetuses with trisomy 21 by whole-genome shotgun sequencing of cell-free fetal DNA in pregnant women's blood samples, the Dutch researchers found that both platforms accurately distinguished between fetuses with two copies of chromosome 21 and those with an extra copy of the chromosome.
Even so, the single-molecule sequencing approach allowed for clearer resolution between trisomy and disomy 21, the team reported, due to lack of GC-bias — the preferential amplification of guanine and cytosine residues, which can lead to skewed read depth for sequences replete with these bases.
"Using both platforms, you can detect the trisomies," senior author Elles Boon, a clinical molecular geneticist with Leiden University Medical Center's Laboratory for Diagnostic Genome Analysis, told Clinical Sequencing News. "But we did see a better distinction between fetuses with or without trisomy 21 when using single-molecule sequencing."
The work represents part of an ongoing effort to validate and optimize a non-invasive, sequencing-based trisomy 21 test that may eventually be offered to pregnant women in the Netherlands. In that country, Boon explained, regulations require such testing to be done through an academic hospital rather than by companies offering diagnostic tests commercially.
"We don't have any commercial interest, to be honest," she said. "We are an academic center and we try to see which method we can use that is the best for diagnostics."
Tackling GC Bias
Fetal DNA was first detected in maternal blood samples in the late 1990s. Since then, intensive efforts have been underway by several groups to find ways of using this DNA for blood-based prenatal testing to replace more invasive methods such as amniocentesis or chorionic villus sampling, which carry a small miscarriage risk.
Indeed, several studies published over the past few years have demonstrated the feasibility of using cell-free fetal DNA sequencing to detect the presence of extra chromosomes and even to map the entire fetal genome (IS 12/14/2010), and a number of firms, including Sequenom, Natera, Aria, LifeCodexx, and Verinata have launched or plan to launch sequencing-based prenatal tests for T21 and other aneuploidies.
To date, though, most of these tests have relied on Illumina sequencing technology, which requires PCR amplification prior to sequencing, Boon noted. While that may not necessarily prevent accurate detection of trisomies, she explained, the GC bias associated with this process could affect the quality of the data used to find the extra chromosomes.
Boon and her colleagues speculated that they might be able to curb this "experimental noise" by using single-molecule sequencing, which does not require DNA amplification prior to sequencing.
To look at this issue in more detail, Boon and her colleagues did a direct comparison between the HeliScope and GA for sequencing-based trisomy 21 testing, using blood samples taken from 20 pregnant women enrolled through Leiden University Medical Center's obstetrics department. The median gestational time for each of the women tested was at least 12 weeks.
Along with DNA from the pregnant women's blood samples, the team also tested cell-free DNA from the blood of four adult male controls.
Across the 23 case and control samples with enough quality DNA for sequencing, the team generated a mean of more than 10 million raw reads on the HeliScope and a mean of 4.65 million reads that could be aligned. For duplexed samples run on the GA, meanwhile, they got a mean of nearly 25 million raw reads and 12.6 million mapped reads.
While sample preparation was shorter, less laborious, and cheaper on the Helicos platform, Boon noted, the actual sequencing step took longer on the HeliScope — four days versus two days per run on the GA.
With the advent of HiSeq 2000 sequencing, though, she said Illumina now beats Helicos on a cost-per-sample basis, since it's possible to multiplex several samples on the HiSeq. Using that approach, she explained, "you can run more samples at the same time, but it will also take three or four days."
Because the types of errors found in Illumina and Helicos reads differ from one another, the researchers used different pipelines for their initial data filtering and analysis steps. They assessed the GA reads with SoftGenetics' NextGene software. Helicos reads were initially analyzed using the company's HeliSphere resequencing pipeline.
Since more labs are using Illumina sequencers to do such testing, Boon noted, there are currently more data analysis tools available for those reads.
"The Illumina platform is used by more labs, so you have more options for analyzing your data, such as commercial software packages or in-house pipelines," she said. "For the HeliScope, you're bound to one program."
Nevertheless, the team ultimately used the same relative sequence tag density, or RSTD, ratio method to detect trisomy 21 fetuses for both platforms. That method uses read depth data across the chromosomes to detect extra chromosome copies.
For the 20 pregnant women tested, the researchers accurately detected all 11 trisomy 21 cases and correctly classified eight more fetuses as having two copies of chromosome 21. The remaining sample, taken from a woman carrying a disomy 21 fetus, did not contain sufficient amounts of quality DNA to generate informative sequence on either platform.
While both platforms could both detect and rule out trisomy 21, the study authors reported that the single-molecule method showed clearer differences between the disomy and trisomy 21 fetuses, due to more uniform read representation across the chromosomes.
"For Helicos, the RSTD ratios for all chromosomes, the normalized total number of reads per chromosome, and the mean number of reads per bin were quite uniform between samples and virtually independent of GC content of the chromosome," they explained, "whereas as reported before, Illumina results showed increased read density in GC-rich areas of the genome."
"I think the single-molecule sequencing really offers a lot of possibilities to reliably detect the trisomies," Boon said.
The potential benefits of using the Helicos platform for trisomy 21 detection came as no surprise to Patrice Milos, former chief scientific officer for Helicos and current senior scientific advisor for Helicos as well as site head at the Pfizer Centers for Therapeutic Innovation.
Milos told CSN she and colleagues at Helicos did a good deal of testing and validation on the use of the HeliScope for non-invasive detection of chromosome 21, 13, and 18 trisomies from maternal plasma, though findings from those studies have not been published.
"This is an area that we followed during my time at Helicos for quite a while," she said. "So I know the area well."
Among the advantages of the platform for this application, she cited its simple methodology, ease of sample preparation, lack of amplification steps, and the ability to sequence very small amounts of DNA.
Milos did not comment on whether Helicos had shared its results with companies in the US that are developing maternal blood-based prenatal tests commercially, but said she is pleased to see the Leiden team publishing its work.
Despite the improved resolution offered on the HeliScope, though, Boon said her group is relying on GC-corrected reads generated on the Illumina HiSeq 2000 platform for non-invasive trisomy 21 detection protocols it is currently testing, primarily due to improvements in the Illumina platform that have increased its throughput and allow for more extensive sample multiplexing.
The Helicos platform, meantime, has not seen any improvements over the past several years and there are ongoing concerns over the company's future given its rocky financial situation (IS 11/15/2011).
Much of the recent research done at Helicos has focused on finding the best applications for the existing instrument rather than platform improvement, Milos explained. This research includes published work describing an isolation-free approach for sequencing the BRCA1 gene associated with breast and ovarian cancer risk (CSN 8/9/2011).
"The research that we have is very much focused, right now, on small quantities of nucleic acids as well as direct RNA sequencing," she said, "and, really, optimization of the applications, with the platform pretty much staying as is based on the type of investment the company could make in it."
While she said reagents for customers already running the HeliScope are still available and new instruments may be available for purchase on a case-by-case basis, Milos noted that Helicos is currently more focused on fee-for-service sequencing.
At the Leiden center, meanwhile, researchers are working with several sequencing platforms besides the Helicos and Illumina instruments tested in the trisomy 21 study, including Illumina's HiSeq 2000 as well as instruments from Pacific Biosciences, Life Tech's Ion Torrent, and Roche 454.
But although the team is interested in exploring the potential applications for each of the instruments, Boon said there are no immediate plans to run the sequencing-based trisomy test on other single-molecule instruments.
"We haven't tried the PacBio for this application, because it's just in," Boon noted. "At this moment, I'm not sure if we will try it for the non-invasive [test], but we are eager to test this new platform as well for different applications."
In addition to the trisomy 21 detection work being done at the Leiden Medical Center, Boon is also coordinating a consortium of academic labs from across the Netherlands that are interested in doing a national validation and implementation study on the use of non-invasive testing for trisomy 21 screening and detection.
"We are comparing protocols and seeing how we can optimize these protocols," Boon said. "So it also involves comparing platforms, data analysis, and a lot of components within the protocol."
Along with work aimed at optimizing protocols and hammering out the technical details of the test, she explained, an arm of that study will also focus on clinical and ethical issues related to the use of non-invasive trisomy 21 testing.
"Eventually we will do a big prospective study within the Netherlands to see if we can really implement this for clinical use," Boon said.
Meanwhile, a number of firms in the US, elsewhere in Europe (CSN 8/17/2011), and Hong Kong (CSN 11/30/2011) have developed, or are in the process of developing commercial tests for detecting trisomy 21 from maternal blood using sequencing.
Sequenom launched the first of these — a lab-developed prenatal sequencing-based test of fetal DNA from maternal blood known as MaterniT21 — in several US cities last fall (CSN 10/19/2011), coinciding with the publication of a clinical validation study on the test involving almost 1,700 pregnant women. The company selected the Illumina HiSeq 2000 for use with MaterniT21 in the fall of 2010 (IS 9/21/2010), prior to the test's commercial launch.
In the clinical validation study of MaterniT21, published in Genetics in Medicine, researchers saw improvements in both false-negative and false-positive rates after adjusting for GC content in sequence data.
Sequenom has said that the cost of goods for the test on the HiSeq run in the neighborhood of $500 to $600 and that the turnaround time is less than eight days. The company expects to reduce the cost of the test by 30 percent to 40 percent over the next year through increased multiplexing; improvements to the instrument, reagents, and flow cells; automation of the sample preparation; and economies of scale via increased testing volume (CSN 1/18/2012).
The test not only reports the results of trisomy 21, but also positive results for trisomies 13 and 18. Earlier this week, the company published a validation study of the T13 and T18 results in Genetics in Medicine (see related story, this issue).
Have topics you'd like to see covered in Clinical Sequencing News? Contact the editor at anderson [at] genomeweb [.] com.