Recent progress in the schizophrenia genetics

•

Publication of two schizophrenia papers in @Nature this week by the @PGCgenetics and SCHEMA consortiums marks a new major milestone in the field of psychiatric genetics. A 🧵 on the key findings from these two landmark papers.

These papers were available as preprints since 2020, but Nature publishing them side by side is an important recognition to the whole psychiatric genetics field and deserves celebration.

I take this opportunity to reflect on how far we've come in understanding this devastating disease. Just to be clear, I'm involved in neither of these studies. All credits should go to the amazing people from the PGC and SCHEMA consortiums.

High heterogeneity in the phenotype presentations combined with natural selection actively removing the psychiatric risk variants from the gene pool have made gene discovery in the field of psychiatry near impossible with small to moderate sample sizes.

The pioneers have envisioned this very early in the timeline and passed on an important mantra to subsequent generations: "sample size, sample size, sample size". Which is also the key motivation behind the formation of PGC and SCHEMA consortiums. cambridge.org/core/journals/twin-research-and-human-genetics/article/nick-martin-and-the-genetics-of-depression-sample-size-sample-size-sample-size/B8ECA6926DE980CCCC0052E73ABA3670#

I've mentioned this before few times. This plot represents a simple summary of how discoveries/knowledge-gains progress over time in the GWAS field. twitter.com/doctorveera/status/1450301320066310147?s=20&t=eHwhxpUd7eFK-cPtrt73qQ

This sequence is true for both common variant (GWAS) and rare variant studies. In terms of common variants, the field has made a huge progress is building a sample size large enough to make discoveries all the way up to variant level resolution.

The current GWAS has gathered a sample size of ~76k cases and ~240k controls, which is truly an impressive achievement by the PGC. nature.com/articles/s41586-022-04434-5

As the sample size grows bigger and bigger, the discoveries start saturating from left to write, which we have now seen in a prototype polygenic trait--height. twitter.com/doctorveera/status/1482384613985058816?s=20&t=A9QlmPYGPL9EYan3lnMYgQ

In line with that, it looks like we are already reaching saturation at the genome and gene-set level findings. At the genome level, the common variant heritability is 0.24, of which ~10% (0.024) can be now explained using only genome-wide significant variants.

The large N has resulted in a polygenic score, which might not be predictive enough for clinical application, but will be of immense value for research e.g. to study treatment response, cognitive deficits, age of onset, rare variant penetrance etc.

At the gene-set level, it's abundantly clear that most of the schizophrenia risk genes are expressed throughout the brain, within the excitatory and inhibitory neurons and are involved predominantly in the pre and post synaptic pathways (I love the SynGO sunburst plots)

Note, the discoveries at the gene set level is now limited not by the GWAS N, but by the size and availability of the brain gene expression datasets at better resolution across time and space. PsychENCODE is making great progress in creating this catalog. science.org/collections/psychencode

At the gene and variant level, the increasing N is continuing to yield more genes and variants. These results themselves might not make immediate sense, but in combination with rare variants they will bring extraordinary insights, which we will see in the second paper.

In the second paper, the authors report an exome wide rare variant association analysis in ~24k cases and ~97k controls (~1/3rd of current GWAS sample size) that led to the discovery of 10 schizophrenia risk genes. nature.com/articles/s41586-022-04556-w

Similar to GWAS, the ExWAS discoveries also progress through the same sequence. Most of the past exome studies have been powered enough to bring insights only up to the gene set levels. With increasing N, we are now making discoveries at gene level.

One major finding that was abundantly clear in the earlier exome studies is that loss of function and deleterious missense variants associated with schizophrenia are enriched within the loss of function intolerant genes.

With the now increased same size, this finding has become stronger in terms of P value and gained better resolution in terms of effect size and standard error.

And by aggregating the protein truncating variants and deleterious missense variants within each gene, the authors identify at least 10 risk genes with high statistical confidence including the previously known risk gene SETD1A.

As someone who reads genetics papers across a broad range of diseases, I was always puzzled by the lack of convergence between rare and common variants in psychiatric disorders. This is easily seen in other diseases e.g cardiovascular twitter.com/doctorveera/status/1473176508529135623?s=20&t=qZ47blV0IUOEBub8YUeFpA

One finding that excites me the most in the current papers is at last, the rare and common variant findings are starting to converge for schizophrenia, not just at the gene set level, but also at the individual gene level.

The above plot shows that the rare variants are enriched in the GWAS prioritized genes. And the authors were also able to identify individual GWAS loci where exome data is pinpointing the likely causal gene. twitter.com/doctorveera/status/1408818093636218882?s=20&t=qZ47blV0IUOEBub8YUeFpA

At some of the loci, e.g at GRIN2A locus, this convergence is remarkable with formation of an allelic series where variants representing varying levels of gene perturbations increase the schizophrenia risk at proportional effect sizes. It's amazing to see we've come this far.

Finally, we are also gaining more insights into the genetic overlap between schizophrenia and ASD/neurodevelopmental disorders (NDDs). Previous schizophrenia exome studies have seen clear enrichment of schizophrenia risk variants within NDD genes.

A recent study has also shown that the overlap is seen not at the gene set level, but also at the variant class level. But the sample sizes so far have been not sufficient to evaluate at the level of individual genes. twitter.com/doctorveera/status/1436702938471206916?s=20&t=qZ47blV0IUOEBub8YUeFpA

The current sample size has enabled the authors to zoom in on individual genes to study the schizophrenia and NDD risk variants. It seems at least for some genes there is a separation at the variant types with denovo missense causing NDDs and inherited PTVs causing schizophrenia

It's tempting to conceptualize these findings in terms allelic pleiotropy that we see for other diseases (e.g. INS gene and diabetes) that severe de novo missense variants hitting critical protein domain lead to childhood onset NDDs twitter.com/doctorveera/status/1382766644511043586?s=20&t=qZ47blV0IUOEBub8YUeFpA

and relatively less severe PTVs lead to adolescent/adulthood onset schizophrenia. But it is much more complicated than that. When we talk about these variants we subconsciously assume a 100% penetrance model.

But these variants vary widely in their penetrance, which depends very much on the genetic background. Background PRS or large effect rare variants can change the phenotypic outcomes. twitter.com/doctorveera/status/1378330014207373313?s=20&t=qZ47blV0IUOEBub8YUeFpA twitter.com/doctorveera/status/1363100705469382661?s=20&t=qZ47blV0IUOEBub8YUeFpA

Hence, it's possible that a same variant can lead to different outcomes (NDD or schizophrenia) in different contexts and many do not seem to appreciate these complexities. twitter.com/sapinker/status/1512462414486204416?s=20&t=qZ47blV0IUOEBub8YUeFpA

Nevertheless, at a higher level, there seem to be some level of separation in terms of gene properties (e.g. pre/postnatal exp, strength of the constraints) between NDD and schizophrenia genes. As sample size grow, we'll get more clarity on this in the future.

To conclude, the field is making great progress in understanding the genetics of this devastating disease and these efforts will one day lead to development of newer drugs. Big congrats to all the authors from the PGC and SCHEMA consortiums 👏 👏