By Monica Heger
This article has been updated from a version published on Jan. 14 to include comments from Good Start Genetics' CEO.
In a step toward moving sequencing-based prenatal testing to the clinic, two groups last week announced results of clinical trials of their respective tests — a noninvasive fetal diagnostic test for Down syndrome and a carrier screening test for more than 400 severe pediatric diseases.
Researchers at the National Center for Genome Resources reported last week in Science Translational Medicine the results of their preconception carrier screening test in 104 unrelated individuals. Meanwhile, Dennis Lo and his group at the Chinese University of Hong Kong reported in the British Medical Journal the results of their fetal trisomy 21 test in more than 700 pregnant women.
Carrier Screening
In the Science Translational Medicine study, researchers from NCGR tested a number of different capture and sequencing methods in order to screen for carrier status of 448 severe pediatric recessive diseases.
The team tested Agilent SureSelect's hybrid enrichment and RainDance's PCR microdroplet technology for the capture, and Illumina's Genome Analyzer and HiSeq 2000 as well as Life Technologies' SOLiD for the sequencing step.
The different technologies yielded similar results in terms of specificity and sensitivity, but moving forward, the team will likely use Agilent SureSelect for the capture and the HiSeq for sequencing, said Stephen Kingsmore, former chief science officer of NCGR, who is now the director of genomics at Children's Mercy Hospital in Kansas City, Mo. However, he added that they were still considering the SOLiD platform, in particular, the 5500.
In addition, the number of diseases covered by the test will be expanded from 448 to 580, Kingsmore said. Several diseases related to severe mental retardation will be added to the list, while a number of other diseases will be dropped because the study indicated that they were "not ready for diagnostics or clinical testing primetime," Kingsmore said. Some of those diseases turned out to have mixed or non-recessive inheritance and in some cases the understanding of the disease turned out to be wrong.
The next step is to begin testing a larger population in a research setting at Children's Mercy Hospital in Kansas City and potentially other hospitals in Kansas City, where testing will be blinded and random, which Kingsmore said he expects to do by summer 2011, and then make it available to other hospitals and institutes, which they will partner with — hopefully by the fall 2011
Kingsmore has previously said that he and his colleagues are targeting a price of around $500 for the test (IS 3/30/2010). He said this week that the test currently costs around $618, including reagents, equipment, and personnel to run it, but not the cost of pathology interpretation or counseling. Reductions in sequencing costs should allow them to bring the price under $500, he said.
The next phase of the validation will be done at Children's Mercy Hospital in Kansas City, and Kingsmore said he will work with pediatricians, geneticists, and obstetricians to get feedback on the test and make sure that the results are understandable to both clinicians and patients.
In addition, Kingsmore said his team aims to partner with other organizations, particularly other hospitals in Kansas City. He said they have also been talking to a children's hospital affiliated with the Max Planck Institute in Germany, Queens University in the UK, as well as physicians in Italy.
Uncovering Misannotations
In the Science Translational Medicine study, the researchers sequenced 7,717 regions from 437 genes to an average of 160-fold target coverage, and were able to detect mutations at 95 percent sensitivity and 99.96 percent specificity for substitutions, indels, splicing, gross deletions, and SNPs. In the 104 unrelated individuals, the carrier burden for severe pediatric recessive mutations was 2.8 and ranged from zero to seven, a finding consistent with previous studies.
Somewhat surprisingly, the researchers found that 27 percent of the mutations cited in the literature as disease mutations were found to be either misannotated or common polymorphisms.
"We were shocked," Kingsmore said. "We had not expected that at all. This certainly speaks to the need for a sequencing-based approach, because genotyping will just perpetuate those mistakes."
He said that the high number of errors could be explained by a number of different factors. For instance, in many cases, the genes were discovered nearly 30 years ago, and the technology was not as accurate and sophisticated as what is available today.
"Part of offering the test will be an ongoing need to improve the database," he added.
[ pagebreak ]
Priming for Commercialization
Moving forward, Kingsmore said that despite the test's initial good performance, there are a number of other issues that pose challenges to bringing it to the clinic. Because it will sequence entire disease genes, the test will yield variants of both known and unknown significance, and deciding which of the novel variants to report to clinicians will be tricky.
Kingsmore said that if the test uncovers an unknown variant that causes a frame-shift or a premature stop codon, they will report it as likely pathogenic. For variants such as amino acid substitutions, which may or may not have functional significance, whether to report those becomes somewhat of an open question. "We've been receiving feedback in both directions," said Kingsmore. In those cases they will likely take direction from the physician, and also report the variants in a database and seek follow-up functional studies, he said.
Also, because the test is screening for so many diseases, ensuring that patients have adequate genetic counseling will be important, said Callum Bell, NCGR's program lead.
When testing for carrier status of one rare disease, you expect the test to be negative, said Bell. But for a test like this, the odds that a parent will be a carrier for at least one of those diseases is much higher, though the odds that both parents will be a carrier for the same disease are still low, Bell said. "It changes your mindset for counseling."
NCGR's test will likely compete with other sequencing-based prenatal tests, such as Ambry's AmbryScreen and Good Start Genetics' test, although each is slightly different (IS 9/14/2010).
While NCGR is screening for more than 500 different diseases in one test, both Ambry and Good Start are taking a more narrow approach. Ambry's test, for example, will screen for approximately 80 diseases, while Good Start CEO Don Hardison said that from his discussions with physicians, "they want a very targeted panel of tests … and the opportunity to customize their ordering patterns based on the individual patient." Therefore, Good Start's test will include only the diseases recommended by the American Congress of Obstetricians and Gynecologists and the American College of Medical Genetics, such as cystic fibrosis, sickle cell anemia, and Tay-Sachs disease. Additionally, the company will offer the option of screening for a panel of diseases, or just one.
Good Start will also not report any novel or unknown variants to physicians, although it will compile those variants in an internal database. "From talking to physicians, they prefer us not reporting mutations that are not very well characterized. The doctor doesn't know what to do with mutations that are not characterized as disease causing," Hardison said.
Fetal Diagnostics
Researchers are also making headway on bringing a noninvasive sequencing-based fetal diagnostic test for Down syndrome to the clinic. Dennis Lo of the Chinese University of Hong Kong and his colleagues recently validated their test on more than 700 pregnant women. Their results were reported in the British Medical Journal last week.
Lo developed the sequencing-based method to noninvasively test maternal plasma for Down syndrome in 2008 and subsequently licensed the technology to Sequenom, which is also developing a Down syndrome test based on the technology that it plans to launch in the fourth quarter on the Illumina HiSeq 2000 (IS 9/21/2010). Currently, the company is clinically validating its test and has said that it will soon publish results of its "locked assay" verification study in the American Journal of Obstetrics and Gynecology (IS 1/4/2011).
Lo said that he retained the IP for the technology in Hong Kong, and plans to launch a sequencing-based Down syndrome test through the Chinese University of Hong Kong this year.
In the study, Lo and his colleagues tested 753 women with high-risk pregnancies, 86 of whom had tested positive for trisomy 21 with karyotyping. They did the sequencing on the Illumina Genome Analyzer and tested both a 2-plex and 8-plex sequencing approach using barcodes.
The 2-plex protocol detected 100 percent of the trisomy 21 cases with a 2.1 percent false-positive rate, while the 8-plex approach detected 79.1 percent of the cases with a false positive rate of 1.1 percent.
Lo attributed the difference in accuracy to the number of reads per sample each approach generated. The 2-plex approach generated 2.3 million aligned reads, while the 8-plex approach generated just 300,000.
For a clinical diagnostic test, Lo said that the 2.3 million reads per sample would be necessary. He said that they have now installed a HiSeq 2000, and are currently validating the test on that technology, which would likely allow them to do more multiplexing since the throughput is so much higher.
Lo added that he plans to launch the test in Hong Kong later this year. "The test is essentially ready to be brought into the clinic, pending the clearance of the necessary regulatory procedures," he said.
He said that his team is also working on nailing down a cost for the test, but could not comment on what he thought it would be.
Additionally, in the BMJ study, the researchers detected the presence of trisomy 18 and trisomy 13 in a number of cases, and Lo said that he plans to develop diagnostic tests for those diseases as well. Eventually, he said, the different trisomies could probably be included in one diagnostic test, but initially he will develop them separately.
Both the work by Lo on fetal diagnostics and that by the NCGR on pre-pregnancy disease screening illustrate the advances that sequencing has made for clinical purposes. "I think next-gen sequencing will become reasonably common as a diagnostic tool," said Lo.
He added that he thought that his test and the carrier screening test could be complementary. The NCGR test "would allow us to identify at-risk couples" who would benefit from the fetal Down syndrome test, he said.
Have topics you'd like to see covered in In Sequence? Contact the editor at mheger [at] genomeweb [.] com.