ATXN8OS motifs and pathogenicity limits

[dnil]

It came to my attention a while back that our lab is struggling with calls on ATXN8OS, especially when borderline pathogenic around 70 copies.
I have resisted updating since I feel the error is on the part of the clinical literature / limitations of PCR-based testing, but currently feel this is not my battlefield. I will at least pass the torch.

As you know ATXN8OS pathogenicity can be controversial and difficult to ascertain in general, with debated reduced penetrance and test utility, and it does not help that classical testing defines the repeat as "(CTA·TAG)n(CTG·CAG)n" eg in https://www.ncbi.nlm.nih.gov/books/NBK1268/.

They also helpfully note

"While the (CTA·TAG)n portion of the repeat can be polymorphic, it is the (CTG·CAG)n portion of the repeat that expands."

I have interpreted this and literature in general as the CTG being pathogenic, but from a renewed review, I'm no longer certain. And as it is the lab will get different sizing from PCR vs WGS, different results when compared to external commercial labs etc. And no-one the wiser when trying to read up and arbitrate.

I would suggest a 🐩 (sorry for the Swedish; https://en.wiktionary.org/wiki/göra_en_pudel), or at least a temporary retreat, to sizing the the full (CTA·TAG)m(CTG·CAG)n in harmony with GeneTests. Once we find evidence showing more clearly the polymorphism size is irrelevant, we could revert and have the CTG as the pathogenic repeat unit again.

[hdashnow]

Thanks for bringing this to our attention!

This is indeed a tricky issue. In STRchive, we have defined the position based on CTG, but do include the CTA in the TRGT definition (as we do for all compound loci).
You can see how this translates in the reference sequence in your PR here: https://github.com/dashnowlab/STRchive/pull/304/files#diff-719324890b00eb75e48559201deb84377288537c6577dba009b48674a1edbb3a

We're going to have to do a bit more thinking on this! The question in my mind is whether we can disentangle the two motifs in the literature. If we move to interpreting it as a compound locus like this I think we'll gave to provide some better automation of this. Maybe additional metadata?

Related, what are your thoughts on interruptions at this locus? Do you think there is enough evidence to interpret their impact?

[dnil]

If you can find some good data on this, I would agree the global optimum would be to update GeneReviews and manuals for Sanger or PCR based testing accordingly. As it is, old school as it may be, clinical testing is currently done with the locus definition spanning both repeats, likely mostly due to the nearest unique primer sites outside. With a (small) room of local experts, I heard only that we don't really know about the effects of the polymorphic CTA.

We should at the very least modify the normal/premutation/pathogenic limit accordingly. But then the limits will differ from literature.

That above GeneReviews text have this interesting observation on interruptions:

"However, one or more (CCG·CGG) interruptions within the expanded (CTG·CAG)n portion of the repeat have been found more frequently in probands with multiple affected family members compared to probands who represent simplex cases (i.e., a single affected family member). These data suggest that asymptomatic individuals with CCG·CGG interruptions within the (CTG·CAG)n repeat expansion may be at higher risk of developing ataxia [Authors, unpublished data]."

We haven't really checked for it, but perhaps listing a purity/LPS score (https://github.com/PacificBiosciences/trgt-lps) could be a way forward?