Genetic markers for some PANDAS/PANS cases

[FittingLight]

Hi, 

Did anybody see this paper, published in Nature, https://pubmed.ncbi.nlm.nih.gov/35773312/ ? 

I've been on augmentin to treat PANDAS/PANS for three years - its been the best years of my life in terms of mental health stability and no flares. Feel like a changed person. 

I recently did Whole Genome Sequencing (WGS), and found that I have the below gene. 

Gene: NLRC4
Variant: c.2-212C>T
rsID: rs199476292
Ref Allele: G
Alt Allele: A
Freq: 0.3122%rare
CADD: 2.014

 

This gene is hardly mentioned anywhere I can find. 

 

Has anyone else with suspected Pandas/Pans undertaken WGS ? 

 

 

 

 

[bobh]

Yes, I have done WGS for my PANS kid (now adult).  In the study you quote, there were 10 PANS cases with whole genome done, and from discussion groups such as this, I know there are more.

So you listed a variant, presumably because it is from one of the genes listed in the paper.  Hard evidence (by which I mean data, not theory) for that gene being related to PANS is not explicitly spelled out in the paper.  It is not based on a study of thousands of PANS patients that all have NLRC4 variants in common while the general population doesn't (which is the kind of evidence needed to make a strong connection).  Rather, the paper implies or suggests that NLRC4 may be a PANS-related gene because of 1) its function, and 2) they found 4 of the 386 PANS cases with ultra-rare NLRC4 variants.  Note that the 4 cases were not the same variant, they were 4 different ones.

What are the chances of getting an ultra-rare variant on a random gene?  As I say, the paper doesn't spell that out, so there is no statement of strength for the implied connection.  One can try to do some homework to answer such a question; here is one answer: https://www.perplexity.ai/search/in-any-random-gene-in-a-specif-syMcq9GMSoOlKM3bPfnm.w     (that is, approximately 1 in 500 to 1 in 1000). I find I get different answers from my AI, depending on not just how but in what context I ask the question.  For example check out this different answer to the exact same question (scroll down to the 2nd question in this thread): https://www.perplexity.ai/search/in-genetics-what-does-maf-0-00-APKxEoEdQAikjSM7bimZ3g (that is, as much as 5.8% in Europeans for medically actionable genes).  So we have to be careful here, the answer seems quite variable.

Back to the paper's hard data: out of 386 tries, the authors found 4 people with ultra-rare (defined as <0.1%) variants on a particular gene (NLRC4).  That is just over 1% of cases.  Maybe that is something compared to the first link above, but less so if considering these points:
1) If NLRC4 can be considered a "medically actionable gene" (that is what my AI says, if it can be trusted), then this finding of 4 among 386 is nothing special, because it is not out of the ordinary in the general population, per the 2nd link above.
2) NLRC4 is the gene in which they found the most (4) ultra-rare variants.  In 6 of the 11 genes listed, they only found 1 ultra-rare variant, so NLRC4 is not really representative, the real average is 2 for the 11 genes they found ultra-rare variants on (this point is further complicated and somewhat mitigated by the fact that they also found some individuals with ultra-rare variants on more than one gene).
3) The paper authors don't actually report how many genes they tried, only that they found 11 genes with ultra-rare variants.  If they tried with an initial list of say, 20 (I made up this number), and didn't find ultra-rare variants on 9 of them, that would make their finding weaker.  It is really important for researchers to report such background information, or better yet, to pre-register their study (to detail exactly what they will do before they see any of the data), to avoid the possibilities of either "p-hacking", or "HARKing" - two of the problems that contributed to the replication crises in science that is in fact still ongoing.

There was a subsequent paper by some of the same authors, first published as a preprint (https://www.medrxiv.org/content/10.1101/2024.02.20.24302984v2.full), and then as a final, peer-reviewed paper (https://karger.com/dne/article/doi/10.1159/000541908/914745/Ultrarare-Variants-in-DNA-Damage-Repair-Genes-in)

I like to compare the differences between the two (preprint and final) to see what the peer-reviewers insisted that the authors do for the final.  This is not to say that peer reviewers are the be-all and end-all, but it is interesting to note.