NIPT Corrections and Thoughts, 10X Genomics

In this post few observations and corrections to my recent NIPT article, and some quick thoughts on the 10X layoffs.

NIPT - Amniocentesis Accuracy

Over on twitter and via email a couple of people pointed out that I my statements on amniocentesis cast it in an overly negative light. I’ve since revised the article, but reference the previous observations in a footnote. Essentially, the articles I referenced on amniocentesis false positive/negative rates were poor and not particularly relevant.

And one reader pointed out that all regulatory bodies accept Amniocentesis as the bench mark for prenatal fetal genotyping, and all societies / regulatory bodies advise confirmatory invasive testing for all positive NIPT findings. I don’t dispute this, and would not seek to offer medical advice.

I had quite a bit to trouble finding good sources for the accuracy of amniocentesis, but I went hunting again. So far the best source I can find references an early study for the late 70s (please do get in touch if you have something more recent):

“Despite the remarkable safety of the procedure and an accuracy rate of 99.4 percent, there was a considerable concern raised by erroneous diagnoses in the U.S. study. In two cases, a baby with Down syndrome was born despite a report of normal chromosomes from the amniocentesis sample; in one case, a diagnosis of galactosemia was made on amniotic fluid cells, but the pregnancy was carried forward and an unaffected baby was born; and in three cases, the sex was identified incorrectly, although the chromosomal disorder feared was ruled out. Intensive restudy of the six cases failed to establish the causes of error; contamination with maternal cells or interchange of samples in the clinic or laboratory are the likely explanations.”

Prenatal Diagnosis of Genetic Disorders, Gilbert S. Omenn. Science, Vol. 200, 26 May 1978

There were 1040 prospective subjects in this study, so the numbers here suggest a ~0.2% false negative rate in 1978 for Down syndrome using amniocentesis. One might imagine that modern lab protocols have improved this even further. There don’t appear to be any false positives in this study (suggesting to me that the rate is likely significantly less than 0.1%).

The false negative rate for NIPT in some instances appears to be similar (in the 0.01% to 0.1% range). NIPT false positive rates may be higher than amniocentesis (for example this study suggests rates in 0.07% range for trisomies, and Natera quote <0.1%).

Overall to me these results help support the use of NIPT for screening and amniocentesis for diagnostics (but of course, I am not a doctor, and this is not medical advice).

Another report sent in by a reader however highlights other potential issues with NIPT, here the authors sent in samples and “and did not inform any of the five laboratories that the women were in fact not pregnant”. Two labs reported no test result, but three others negative. So… in these cases I guess the fetal fraction calculations previously mentioned failed to provide sufficient quality control.

I’ll always try and post corrections to articles, so you might want to consider a free subscription in order to stay up to date.

NIPT and GRAIL

Manal Mehta over on twitter pointed out the GRAIL/cancer screening NIPT connection. Where FTC documents reference Meredith Miller’s collection of "really really weird" NIPT results. An article by her daughter describes this in more depth and the role her mother played. Where these “weird results” turned out to be cfDNA from cancer. This helped motivate the development of GRAIL and the Galleri test. Of course the history of cfDNA cancer detection is older than this, but an interesting part of the story none the less.

10X Layoffs

So it seems 10X is laying off 100 people (8% of staff). This is after they announced that their Q2 revenues are expected to be down 1% YoY. 10X showed pretty slow growth during 2019 and 2020. Many seemed to put this down to COVID and no doubt this is part of it. And they did indeed have a bit of a revenue boost in 2021, I imagine as users moved out of lockdown. But when I last looked into the market, it also wasn’t super obvious to me that single cell and spacial applications would scale in the same way sequencing has where there are obvious high volume applications you can point to to grow the market.

I’ll continue to watch 10X with interest, and try to better understand where growth in single cell and spacial applications might come from. And of course, I hope the employees effected by the layoffs find success elsewhere.