Hey @GeneSmith, really appreciate you putting this together. I wanted to throw in a few thoughts regarding monogenic disease (the disclaimer/context here is I lead engineering at Orchid, and we’ve put a lot of thought into our monogenic embryo screening).
The 1% risk of monogenic disease you cite above is pretty misaligned with most estimates of monogenic disease. It may be an old reference, or filtering for only catastrophic disease, but common estimates for monogenic screening yield in adults are between 3.5% and 18%:
BabySeq found 18 out of 159 infants (11.3%) with diagnosable and actionable disease variants.
The BabySeq project revealed unanticipated monogenic disease risks in 11% of newborn babies.
DiscovEHR found 3.5% of adults with pathogenic actionable variants
In a survey of 76 clinically actionable disease-associated genes, we estimated that 3.5% of individuals harbor pathogenic or likely pathogenic variants that meet criteria for clinical action. Review of the EHR uncovered findings associated with the monogenic condition in ~65% of pathogenic variant carriers’ medical records.
Vassy et al., 2017 found that of 50 healthy individuals who were sequenced, 18% had a pathogenic or likely pathogenic variant associated with monogenic disease.
(we’ve written a bit more here about how the screening we do translates into a % of embryos with findings).
It’s important to remember that carrier screening is very limited and conservative, generally flagging only recessive variants strongly associated with health conditions (deafness, etc) on a limited list of genes. But in reality there are a large number of low-penetrance, or moderate pathogenicity, pathogenic dominant variants that a WGS screen (like the above studies, and our PGT-WGS) surface.
The GC and reproductive health community is very conservative, and defaults to not scaring patients (as they see it) and not trusting them with probabilistic outcomes, so you end up with pretty conservative carrier screening panels. But this is the same instinct that makes them reflexively opposed to PGT-P! So if we’re talking about monogenic vs polygenic screening it’s reasonable to look at the whole universe of risk alleles (which may have a 3x elevated risk of disease, or a 10% risk of disease) rather than the limited carrier screening lists.
And what we see in practice aligns with these numbers. For example, we have a case whitepaper here (more soon) where we detected a pathogenic variant linked to cardiomyopathy in embryos from an IVF cycle; when we consulted with the patients, it turned out the male partner was on medication for dilated cardiomyopathy! The dominant variant wasn’t on the carrier screening panels, but in practice finding a variant like this is more informative than a 99% PRS for heart disease which may only translate to a 2-3x risk of disease.
I don’t at all mean to cast shade on polygenic screening, because of course we offer PGT-P (and raw data exports), and if you’re in the 90% of couples without a monogenic finding, it’s absolutely the best way to move the needle. But I do really think that:
Monogenic screening, even today, has a very high ROI on future disease prevention. In aggregate rare disease costs something like $1 trillion a year, most of that is genetic, and monogenic screening even today can catch a real % of that (I’m spitballing ~50%, but the ROI even there is ~15x so there’s a lot of room for ballparking).
As large biobanks come online, we are going to see a lot more monogenic pathogenic risk alleles identified. Many of these variants are individually extremely low prevalence and we simply aren’t seeing them linked to disease until we have biobanks with 1mm+ individuals with genotypes linked to health data.
Very happy to go into any of this in more detail, and of course really appreciate the work you spent putting this guide together : ) Most patients have no idea where to start on embryo screening and your guide is an incredible reference for orienting them.
I haven’t read much at all about the Iterated Meiosis proposal but I’d be pretty concerned based on the current state of polygenic models that you end up selecting for a bunch of non-causal sites which were only selected as markers due to linkage disequlibrum (and iterated recombination makes the PRS a weaker and weaker predictor of the phenotype).