Over on twitter Manal Mehta pointed out Vanadis, a novel NIPT platform acquired by PerkinElmer in 2016. I’d not heard of Vanadis before, but the technology seems entertaining. This also seems like a excuse to discuss NIPT in general!
History
Noninvasive prenatal testing using Cell-free DNA sequencing introduced commercially in 2011. Initially tests focused on aneuploidies, that is fetuses with abnormal chromosome counts. Most commonly trisomy 21 (Down syndrome). Generally this is detected though karyotyping, essentially imaging chromosomes present in individual cells harvested through amniocentesis.
Not only is sticking a needle in someones uterus painful, invasive and dangerous, early amniocentesis isn’t commonly performed (typically amniocentesis is only performed after ~15 weeks, I assume to a higher risk of miscarriage early than this)1. Amniocentesis is used to confirm positive screening results from NIPT, as potential issues with NIPT mean that it is only used as a screening test. NIPT can be performed a little earlier than amniocentesis, as early as 9 weeks. However national bodies recommend that have positive NIPT result is confirmed using amniocentesis. For more detail on this, I recommend reading my corrections and thoughts here.
Reading karyotyping protocols helps explain the amniocentesis process. Lab technicians need to culture cells, ensure they are well separated on a slide., stain them, and image them:
Chromosomes also need to be relatively well separated, which I imagine is a stochastic process… Next chromosomes need to be classified and counted, either manually or using automated image analysis software.
The whole process seems complex, and seems like it requires a degree of manual oversight. Given that the basic technique is >60 years old these issues are perhaps to be expected.
Cell-free DNA Noninvasive Prenatal Testing
As Next-gen sequencing gained wider adoption it became possible to address more applications. As a helpful reader pointed out, a number of key patents were filed in 2010 on the use of cfDNA sequencing for prenatal testing. These are still valid, and have resulted in numerous lawsuits. Many of these key patents are now owned by Illumina…
NIPT has several key advantages over amniocentesis, the first being that it is far safer. But it can also be used earlier, as early as 9 weeks. Positive results are generally still confirmed by amniocentesis however.
The approach uses a blood draw to detect fetal derived cell-free DNA. A key issue is determining the fetal fraction (fraction of cfDNA derived from the fetus). This tends to be in the 10 to 20% range. Fetal fraction is determined by differences in fragment size (using paired reads), SNPs or even methylation differences.
But for detecting aneuploidies, as long as reads are align-able the error rate isn’t hugely significant. It’s essentially a counting problem. Unlike early stage cancer detection, you don’t even need large numbers of reads, with 400K to 1M reads being sufficient. Commercial offerings use read counts in the 5 to 8M range (costing $10 to $100).
Outside of aneuploidies, providers have also been offering expanded tests covering SNPs, which have proved somewhat contentious, and likely require much deeper sequencing.
Hard numbers on costs are harder to come by, but the basic test covering aneuploidies is likely sold from somewhere in the $500 range.
Interestingly I’ve discussed NIPT as a potential market for very low cost-per-sample ($1) sequencing. Most investors seemed to feel that outside of developing markets the sequencing component of NIPT wasn’t very cost sensitive.
NIPT however still feels like it has yet to become a routine tool. With Natera stating that “about 40% roughly of pregnancies today are getting NIPT, the vast majority of patients are still getting biochemical screening which has a positive predictive value that is significantly less than NIPT”.
Perhaps something closer to the clinic providing a complete sample-to-answer solution could help increase adoption…
Vanadis
With the scene set, let’s talk about Vanadis. Vanadis is an NIPT platform covering aneuploidies but doesn’t use sequencing. Instead they use targeted amplification of cfDNA. It’s kind of like doing the nano ball amplification part of the MGI sequencing process, without the sequencing itself:
During the process they amplify and label only those targets they are interested in, for example sites on chromosome 21. The key advantage you have here is that the instrumentation is vastly simplified. You can throw a large number of samples on a plate and scan them without needing a complex fluidic system or to multiplex samples.
The system provides counts similar to sequencing based NIPT. But there are a couple of key limitations. Firstly, you need a different fluorophore for each target. Practically, this probably limits you to about 6 targets. Vanadis seem to be looking at trisomy 21, 18 and 13. Which with controls feels like it might be near the limit.
Secondly it’s immediately clear how you go about determining the fetal fraction using this approach. One of Vanadis’ patents suggests a potential approach using methylation:
“digesting said sample with an MspJI family restriction endonuclease to produce a population of fragments that are in the range of 20-40 base pairs in length and have a central methylated cytosine”
So they use can methylation differences between mother and child to determine fetal fraction. This seems reasonable, but a little less robust than the plurality of options available using sequencing.
Summing Up
The Vanadis approach doesn’t seem to provide richer or more accurate data than sequencing. So it either needs to be quicker or cheaper. With a 22 hour run time for the core protocol, and batch processing, time to result is unlikely to be very fast. But by removing multiplexing, it’s likely easier for a single lab technician to process a large numbers of samples in batches, and I suspect any cost advantage would come through reduced labor costs.
The Vanadis approach could also help them avoid some of those key patents and associated licensing fees, further reducing costs.
So, if you believe that NIPT is cost sensitive, then Vanadis could have a short term advantage here. In the medium term however I’d be less optimistic, the platform is clearly always going to be limited to a small number of targets, competing against ever cheaper and more versatile sequencing based approaches.
The first version of this article contained statements on the amniocentesis which was not particularly clear, and I’ve revised this text twice. Previously the article contained the following statement “early amniocentesis can also have a false positive rate as high as ~3.6% (however, later stage amniocentesis as commonly performed, is highly accurate as I understand it)”. Discussing this over on twitter, it seems like the study referenced is possibly flawed. And technically it seems hard to see how karotyping could result in a false positive diagnosis.
I’m struggling to find a good reference for amniocentesis false positive and false negative rates. So if you happen to come across one, please get in touch.