Natera's Prenatal Secret Sauce

By Allison Proffitt

November 14, 2012 | When it comes to prenatal testing, it’s a numbers game. Low risk, hormone-based tests are unreliable. The tests have false positive rates of 5%, which means over 120,000 women each year get positive results—high risk results—leading to high stress situations and unnecessary amniocenteses. The flip side is a false negative rate of 15%. Women get the all clear, when in fact something is wrong.

Jonathan Sheena, Natera’s Chief Technology Officer, minces no words. “Those numbers are not suitable for this century.” 
 

Jonathan Sheena

Natera’s solution is one of a handful of noninvasive prenatal tests (NIPT) based on next-generation sequencing of cell-free DNA in maternal plasma (see, “In Conversation: Tufts Geneticist Diana Bianchi on Noninvasive Prenatal Testing”). And Natera’s numbers are much more appropriate for this century. Data from a clinical trial published in late October in Prenatal Diagnosis showed Natera’s test to be 99.92% accurate.  

Originally called Gene Security Network, the company has been in the prenatal testing space since 2004 offering non-invasive paternity tests, miscarriage testing, pre-implantation screening, and genetic carrier screening out of its CLIA-certified laboratory in San Carlos, California. In April 2011, the company received a $2 million grant from NIH to conduct a clinical trial of its technology for prenatal diagnoses. In January 2012, the company conducted a $20 million financing round led by Lightspeed Venture Partners and changed its name changed to Natera.

Natera’s NIPT takes advantage of high throughput sequencing technology and a computational “secret sauce” to provide consistently accurate diagnoses of aneuploidy across chromosomes 21, 13, 18, X and Y in the first trimester—at 9 weeks. The test is based on an algorithm called NATUS—Next-generation Aneuploidy Testing Using SNPs—which is an advanced form of the version Natera has used in its previous tests.

The five chromosomes are significant, says Sheena. “By covering all five chromosomes, instead of just 21, or just 21, 18, and 13, you’re actually providing a test with two to two and a half times the clinical coverage of a test that’s just looking at 21 and 18.” That’s detecting two to two and a half times as many abnormalities in the population, interjects Gautam Kollu, Natera’s VP of Marketing and Business Development. “Right now people don’t do it because existing blood-based maternal serum screening (MSS) doesn’t allow you to test for sex chromosome abnormalities. However the invasive tests can look for it, but of course the invasive tests carry a risk of miscarriage that nobody wants,” Kollu says.

It starts with sequencing the cell-free DNA found in the mother’s blood. The resulting data is a noisy mix of both maternal and fetal sequences. Statistically separating the two is Natera’s specialty. A directed SNP assay developed in house scans the samples for SNPs of interest. It’s a targeted SNP approach, explains Sheena. “By looking at targeted SNPs—the places where people tend to differ—you can [deduce] the contribution of the fetal DNA that you’re seeing in maternal plasma.” 

“We combine very clean measurements from the mom (and the dad if it’s available), along with data from the Human Genome Project, specifically the HapMap data, which tells you where chromosomes tend to cross over when they’re forming… and using that a priori information, we can clean up the very noisy measurements you’re seeing in that maternal/fetal mixture.”

NATUS predicts the fetal genetic makeup from the cleaned up data. “We’re looking at hypotheses,” Sheena says. “We’re looking at the hypothesis that, for example, the fetus inherited two copies of chromosome 21 from mom and one from dad, or one copy from mom and two copies from dad or one copy from each. And we’re looking at the data and telling you what is the most likely explanation for the data that you’re seeing. By turning the question around like that, you’ve gained a huge amount of statistical power.”

Add likely chromosome crossover points, and there are billions of possibilities. That’s where Natera’s algorithm does the heavy lifting. “You do that in a computer and you come up with the hypothesis that best explains that noisy data, that maternal/fetal mixture that you’re looking at. And what you find is that one of those hypotheses peaks very high, and that’s how you get the statistical power behind our accuracy rates.”

That statistical peak reveals presence or absence of aneuploidy. “You can look at those relative amounts and determine whether there is the presence of a trisomy or more rarely, a monosomy,” Sheena explains. In a maternal blood sample, only 5% to 10% of the cell-free DNA will be fetal sequence. “You’re looking at pretty small variations in that fetal DNA,” he says.

Secret Sauce 

The targeted SNP assay and Natera’s algorithm form the “secret sauce” Sheena referred to, and it’s what differentiates Natera’s approach from others in the NIPT arena.

“Chromosome 21 happens to amplify fairly consistently, so the brute force approach, those qualitative approaches that other companies are using, it may tend to work fairly well. But once you start looking at chromosomes 18, and 13, and especially the sex chromosomes, those methods tend to fall apart. And you can see that in the clinical data.”

He continues: “By using SNPs and by using these parental hypotheses, you are immune to the all the noise and amplification problems that would be a problem if you were looking just at the raw, quantitative signal. By using SNPs and by looking at the fetal signal in the context of mom especially, you can get high confidence of results across all chromosomes.”

Natera’s numbers have been very high across all five chromosomes. Earlier this month, Natera presented data at the annual meeting of the American Society of Human Genetics from 407 maternal blood samples, demonstrating 100% specificity and sensitivity when detecting trisomy 21 (Down syndrome), trisomy 18 (Edwards syndrome), trisomy 13 (Patau syndrome) and sex chromosome abnormalities, such as monosomy X (Turner’s syndrome). Clinical trial data from the NIH-funded trial were published in Prenatal Diagnosis and showed that Natera’s technology correctly identified the number of chromosomes at all five chromosomes examined, from 166 maternal blood samples that passed the quality threshold, for a total of 725/725 correct chromosome calls.

Sheena says the company plans to begin offering the test in December this year.

Forward Thinking 

Currently, Natera is using Illumina sequencers, but the method is “platform agnostic,” Sheena says. “Our vision is that ultimately this will be delivered closer to the point of care, so it’s easier for the patient, it’s better for the provider, and this is where desktop sequencers will come into play,” adds Kollu.  

Natera’s SNP approach also limits some of the data management issues inherent in next generation sequencing. “Because we’re targeting [SNPs], you can distill the data that comes off these sequencing machines… Using our assay, we can distill that to the reads on those SNPs of interest. Just from a pure data perspective, you’re shrinking your files by orders of magnitude. And that allows for all sorts of interesting things. That allows for faster processing; that allows us to process in the cloud because you’re not limited by data transfer. If we wanted to deliver outside of this facility, that’s something that’s possible because of this assay. All those things are made possible because you’re distilling the information into a much smaller package.” 

The technology is also bound to improve. “You can imagine that over time, you’re going to get a finer and finer grain look into the chromosomes themselves,” Sheena says, which will open up a host of new possibilities. “Deletions and duplications—micro-deletions like [those causing] Di George and Prader Willie [syndromes]—those actually account for a much higher percentage of genetic abnormalities at birth than we had all originally understood. It’s going to be vital to be looking at deletions and duplications going forward. And as the technology gets better, that’s the next thing to look at,” Sheena says.

“Once you’re looking at segments, you can imagine going down to the bases and being able to detect conditions like cystic fibrosis or Tay-Sachs and being able to see if the fetus actually inherited that from the parents. That’s actually a test we’ve been offering in [pre-implantation IVF testing] for a few years now. It’s a hard problem, but it’s something that’s definitely on the horizon.”