By Monica Heger
Building on a sequencing-based approach to non-invasively detect fetal trisomy 21 in maternal plasma, Dennis Lo of the Chinese University of Hong Kong and his team have shown that the method, with slight modifications, can be applied to detect trisomy 13 and trisomy 18, as well.
Publishing today in PLoS One, Lo and his team screened 392 women — of which 25 were carrying fetuses with trisomy 13 and 37 were carrying fetuses with trisomy 18 — and showed that by adjusting the bioinformatics analysis of their previous method, they could detect all trisomy 13 cases and 34 of the 37 trisomy 18 cases at specificities of 98.9 percent and 98 percent, respectively.
Lo told Clinical Sequencing News that the three trisomies — 21, 13, and 18 — could be incorporated into one sequencing-based diagnostic test, but that he would likely develop the trisomy 21 test first, and then add trisomy 13 and 18 detection later.
Detecting trisomy 13 and 18 is trickier than detecting trisomy 21, Lo said. Identifying aneuploidies essentially revolves around being able to measure the "relative concentration of different chromosomes," Lo said, "but the precision with which you can measure each chromosome in the plasma differs."
Typically, chromosomes are measured by what is known as the coefficient of variation, which is essentially a ratio similar to signal to noise, which measures how much variance there is when trying to use sequencing to measure the genomic representation of a chromosome. Both chromosome 13 and 18 have coefficients of variation several times higher than chromosome 21, making it more difficult to detect whether they are present at normal or higher levels, Lo said.
In the paper, the researchers demonstrated that by using the exact same method they used in a study of trisomy 21 that they published in the British Medical Journal earlier this year (IS 1/18/2011), they could only detect 36 percent of the trisomy 13 cases and 73 percent of trisomy 18 cases, with specificities of 92.4 percent and 97.2 percent, respectively, which would not be accurate enough for a diagnostic test, said Lo.
The patients in the study were largely from the same set of patients sampled in the BMJ study, with 48 newly recruited women. For 314 women, the same sequencing results were used, but the method for analyzing the results was modified to better detect trisomy 13 and 18.
In order to increase the ability to detect these aneuploidies, Lo said that his team needed to increase the number of reads that could be counted and aligned to chromosomes. One option would have been to simply increase the coverage; however, that would also increase the cost, potentially making it prohibitive for diagnostic use.
Instead, the team pursued a two-step strategy. In the first step, they incorporated repetitive regions of the genome that they had previously masked because they are difficult to align to chromosomes and not needed to detect trisomy 21. Lo incorporated only those repeats that can be mapped back to chromosomes, which "increased the number of aligned reads by about double," he said. While this improved the accuracy, however, it was still not sufficient.
The next step involved GC correction. "The precision in which one can measure chromosomes in the plasma is related to GC content," Lo said. Chromosome 21 has GC content in the middle of the spectrum, around 41 percent. Chromosome 13 has extremely low GC content, while the GC content of chromosome 18 is somewhere between 13 and 21, he said.
The team used a regression algorithm to normalize for GC content, "and when we do that, the result improves dramatically," with sensitivities for trisomy 13 and 18 at 100 percent and 92 percent respectively, and specificities at 98.9 percent and 98 percent, respectively. "We now have a diagnostically useful test," Lo said.
Lo's team used the Illumina Genome Analyzer for the work in the PLoS One study, but plans to launch a diagnostic test on the HiSeq 2000.
While Lo has licensed intellectual property for trisomy 21 detection to Sequenom, which plans to commercialize a test by 2012, he has retained the IP for Hong Kong, and plans to offer a trisomy 21 test out of the Chinese University of Hong Kong, also by 2012. He said he is now just optimizing the protocol for the HiSeq instrument, and determining the level of multiplexing.
Lo declined to comment on whether he plans to license the method for detecting trisomy 13 and 18 to Sequenom or any other company.
In the BMJ T21 study, Lo's team found that a 2-plex protocol was more accurate than an -plex protocol, but ultimately decided that it was really the number of reads that mattered, and that 10 million reads per sample was the ideal number. In the current study, the team used only a 2-plex protocol.
While trisomy 13 and 18 tests could be "easily incorporated" into the trisomy 21 test, Lo said he would likely launch the trisomy 21 test first and incorporate other trisomies once clinicians had begun to adopt the trisomy 21 test.
Other companies such as LifeCodexx and Verinata Health are also planning to commercialize sequencing-based trisomy 21 tests by 2012. LifeCodexx, which uses a method based on Lo's technology, announced last week that it has begun clinical validation of its test, but does not yet have plans to look at other trisomies (CSN 6/29/2011).
Verinata Health, meantime, has said that it will potentially include other aneuploidies in its test (CSN 5/4/2011). In an initial validation study, published in Clinical Chemistry earlier this year, aside from detecting trisomy 21 cases, the company showed that its test could detect eight out of eight trisomy 18 cases, as well as other aneuploidies such as trisomy 13 and trisomy 9.
Verinata uses a slightly different method than Lo's to detect aneuploidies. The approach, called a "normalized chromosome value," calculates the number of reads from the chromosome it is measuring and then compares that to a subset of reads from another chromosome or set of chromosomes. It is based on technology it licensed from Stanford University developed in Stephen Quake's laboratory.
Essentially, the company is comparing the "target chromosome to a chromosome with a similar type of behavior," which acts as a control chromosome, said Lo.
Verinata has not yet determined if it will include other trisomies aside from 21 in its commercial diagnostic test, and has said that decision will depend on the results from its clinical validation study, which is currently underway.
Have topics you'd like to see covered by Clinical Sequencing News? Contact the editor at mheger [at] genomeweb [.] com.