Chapter 36: Genetics in Child Psychiatry

Welcome to Last Minute Lecture.

This free chapter overview is designed to help students review and understand key concepts.

These summaries supplement not replaced the original textbook and may not be redistributed or resold.

For complete coverage, always consult the official text.

Welcome to the Deep Dive.

Today we're taking on a really fascinating area, one that's just exploded in the last decade or so.

Genetics in child psychiatry.

It really has.

We're moving past those, you know, straightforward single gene conditions into the much, much tougher world of complex genetic risk.

That's right.

It's important context, actually.

For a long time, human genetics was all about solving those rare single gene puzzles.

The classic Mendelian stuff like neurofibromatosis or Huntington disease.

Critical work, obviously.

Absolute critical today are the common complex conditions like schizophrenia, autism spectrum disorder, ADHD.

Yeah.

And we know the heritability is high for these, right?

Meaning a lot of the variation we see comes down to inherited factors.

Exactly.

But the huge difficulty is we're not just looking for one gene.

It's this incredibly complex, subtle dance between loads of different genes and, of course, the environment.

And that complexity, is that why it's been, well, harder to pin down psychiatry compared to, say, diabetes or cancer genetics.

That's a huge part of it, yes.

The sheer complexity of the phenotypes, the observable traits and symptoms, and just how varied people are within one single DSM -5 diagnosis, like ASD.

Right.

So one label might cover several different biological paths underneath.

It's highly likely.

And that's precisely why alternative frameworks like the NIH's research domain criteria, the RDOC system, have become so important in research.

Ah, okay.

So RDOC moves away from the symptom checklist.

Exactly.

It focuses more on dimensions, things you can measure based on brain circuits and observable behaviors.

It lets researchers look at the underlying biology without being locked into those sometimes messy traditional diagnostic boxes.

Makes sense.

And the really exciting part, the hook, is that this flood of genomic data we have now, thanks to massive efforts like the Human Genome Project and the Psychiatric Genomics Consortium, the PGC, it's starting to connect the dots.

Oh, so?

We're seeing significant genetic overlap between these different traditional diagnoses.

It strongly suggests there are common risk pathways that just cut right across those old boundaries.

Okay, so let's get our terms straight.

When we talk genetic risk, we're talking about these tiny variations in DNA, these polymorphisms, right?

Correct.

And the most basic, most common type is the SMP, the single nucleotide polymorphism.

Okay.

It's just a change in one single DNA letter, like an A swapped for a G.

They're all over the genome, super common, and they're the main thing we look for in large genetic studies of common risk.

But it's not just single letter changes.

There are bigger chunks involved too, the CNVs.

Exactly.

Copy number variants.

These are larger structural changes, deletions, duplications, repeats of DNA segments, usually bigger than a thousand base pairs.

And these have a bigger impact?

Potentially, yes.

They can change the dosage of a gene, how many copies you have, or mess up important regulatory bits of DNA nearby.

They can carry a heavier risk load.

You mentioned an example earlier, the 22Q11 .2 deletion linked to DeGeorge syndrome.

That one's pretty dramatic, isn't it?

It's a classic example.

That single deletion boosts the odds of developing schizophrenia by something like 50 times compared to someone without it.

50 times.

That's huge.

That completely changes the picture for someone.

It absolutely does.

And it highlights this really important concept, the spectrum of risk.

There's this sort of inverse relationship going on.

Okay, explain that.

Inverse relationship between what?

Between how common a variant is in the population and how big an effect it has on disease risk.

Think of it like figure 36 to 3 in the text, showing this trade -off.

Right, right.

So walk us through the two ends of that spectrum.

Okay, so on one end, you've got the common variants.

These are mostly those SMPs we talked about.

They're found in more than 1 % of people.

But individually, they only carry a tiny bit of risk.

Maybe they increase your odds by, say, 1 .2 times, 1 .5 times, small effects.

So one SMP doesn't really do much on its own.

Not much at all.

They only really contribute to illness when you inherit thousands of these little risk factors all adding up together.

That's what we mean by polygenicity.

Polygenicity.

So risk spread out across loads and loads of genes.

Precisely.

Then at the other extreme, you have the rare variants.

These are often those big CNVs or mutations that really break a gene.

They're found in less than 1 % of the population.

But they pack a punch.

A big punch.

These are the heavy hitters.

They can increase disease risk dramatically,

maybe 20 -fold, 50 -fold, even 100 -fold.

Wow.

So modern genetics has to grapple with both.

You need to account for the combined effect of maybe some rare high -impact variants and that background polygenic score from all the common SMPs to get the full picture of someone's risk.

Okay, so that's the landscape.

How did we actually find all these risk factors, especially the common ones, with tiny effects?

That sounds incredibly difficult.

It was incredibly difficult.

Older methods like linkage analysis just weren't powerful enough for common complex traits.

The real revolution came with new technologies.

And the first massive breakthrough was GYs.

Genome -wide association studies.

Exactly.

GWs was the game changer.

It uses these microarray gene chips, basically, a way to cheaply and quickly scan millions of common SMPs across the whole genome and thousands of people at once.

Okay, scanning millions of spots.

But because each SMP has such a tiny effect, you need enormous statistical power.

We're talking studies with tens, even hundreds of thousands of participants to reliably detect those subtle signals.

And when they finally reached that scale, the discovery started pouring in, especially for schizophrenia.

That was one of the first big psychiatric successes.

Huge international efforts like the PGC identified, I think the latest count is over 270 locations in the genome associated with SCZ risk.

The earlier count mentioned in the chapter was 128 associations at 108 loci.

Wow.

And what did those findings tell us?

Just a list of genes.

Oh, much more than that.

It immediately pointed towards key biological pathways.

Genes involved in how brain cells communicate and change synaptic plasticity and crucial neurotransmitter systems like dopamine, glutamate, and calcium signaling.

It gave us a biological starting point.

A roadmap source.

A biological roadmap, exactly.

And now we're seeing GWA successes for childhood disorders too.

Right.

You mentioned ADHD and autism.

Yes.

Recent large GWs have found risk loci for ADHD.

I think 12 were highlighted and several for ASD, pointing towards genes involved in things like synaptic development and how DNA is packaged, chromatin regulation.

And anorexia nervosa too, with that interesting metabolic link.

That was striking.

Finding genetic links not just to psychiatric factors, but also to metabolic traits really makes us rethink the boundaries of that disorder.

So finding these shared genes across different diagnoses, SCZ sharing with bipolar or ASD,

does that challenge how we think about these as totally separate conditions?

It absolutely does.

That's the concept of pleiotropy 1 gene influencing multiple traits.

These cross disorder studies keep finding genetic hot spots that don't respect our traditional diagnostic lines.

Which suggests shared biology underneath.

Shared underlying mechanisms, yes.

And that brings us back to thinking dimensionally, using models like the liability threshold model.

Okay.

Explain that model.

It sounds key.

Like figure 36 -2 tries to illustrate.

Exactly.

Imagine genetic risk, like a continuous scale, maybe like neurotic existing across the whole population.

That's the liability.

It's dimensional.

A formal diagnosis, like an anxiety disorder, only happens when an individual's total liability, that's their genetic risk, plus environmental hits, crosses a certain point, a clinical threshold.

So the underlying risk is a spectrum, but the label is yes, no, once you cross the line.

Precisely.

The genes contribute dimensionally, but the diagnosis becomes categorical when impairment hits that threshold.

And all this GTOA's data lets us actually estimate that total underlying genetic risk for someone using a polygenic risk score, or PRS.

That's the idea.

PRS tries to sum up the effects of thousands, even millions, of those common SNPs identified in GTOA to give an individual score reflecting their overall genetic liability.

Oh, that has a big but here, isn't there?

You mentioned the Eurocentric bias.

A huge but.

It's a critical limitation right now.

The overwhelming majority of participants in these massive GWAS studies have been of European ancestry.

So the scores work best for them.

They work much better for individuals of European descent.

The predictive accuracy drops off significantly for people from other ancestral backgrounds because the genetic markers and their effects might differ.

Wow.

That's a massive health equity issue if we start trying to use these clinically.

It's a fundamental barrier to equitable clinical implementation right now.

We absolutely have to diversify these study populations urgently.

So that's common variance.

What about finding those rarer, high impact CNVs?

Different tools?

Different tools.

We typically use microarray methods again, but designed differently.

Techniques like comparative genomic hybridization or CGH arrays, specifically to detect those larger deletions or duplications.

And what have those CNV studies found, Bradley?

They consistently show that the overall burden of rare, large CNVs is significantly higher in individuals with neurodevelopmental disorders like ASD and schizophrenia compared to controls.

And the affected genes point to similar pathways.

Often, yes.

Genes involved in building synapses, calcium signaling, core neurodevelopmental processes keep popping up.

And then there's the ultimate tool, actually reading the entire genetic code, genome sequencing.

Right.

You can do whole exome sequencing, WES, which focuses just on the protein coding bits, about 3 % of the genome, or whole genome sequencing, WGS, which aims to capture pretty much everything.

And what have WES and WGS revealed, say, in autism?

They've been particularly powerful for studying de novo mutations, new mutations that appear in the child but aren't inherited from either parent.

Ah, spontaneous changes.

Exactly.

And these studies found that severe disruptive de novo mutations in certain critical brain genes, like SCN1A or CHD8, carry very large risks for ASD.

And there was a link to paternal age.

Yes.

Often these studies show a correlation that the rate of these de novo mutations tends to increase with the father's age at conception, potentially linked to sperm cell division over time.

So, okay, we find hundreds of these risk genes through GWS, CNVs, sequencing.

Then what?

How do we know what they do?

That's functional genomics.

That's the next massive challenge.

Finding the genes is just step one.

We need to know when and where in the brain they're active and how they influence biology.

How do you figure that out?

We use bioinformatics tools, pathway analysis, databases like gene ontology, to see if the identified genes cluster in particular biological processes more than you'd expect by chance.

Looking for patterns.

Looking for patterns.

And then there are lab techniques, really cutting -edge stuff like Hi -C or 8 -Taxec, that help us understand how genetic variation affects the regulation of genes, not just the gene itself, but how it's switched on or off, like a dimmer switch.

Okay, let's bring this back to the clinic.

With all this data, what does it actually mean for a child psychiatrist seeing a patient today?

Where is genetics genuinely actionable?

Right now, it's really actionable in two main areas.

First is diagnostic genetic testing in specific situations.

Professional groups, like the American Academy of Pediatrics, actually recommend chromosomal microarray analysis, that's CGH for CNVs, and often fragile X testing as part of the standard evaluation for any child with suspected autism spectrum disorder or any child with intellectual disability.

Rett syndrome testing, MECP2, is also key for girls with specific features.

So that's become routine for ASD and ID.

It's recommended as routine, and the yield is significant.

Doing this testing can lead to a specific molecular diagnosis in maybe up to 20 % of those children.

20%, that's substantial.

Finding a specific cause must be huge for families.

It can be transformative, it can guide management, give prognostic information, inform family planning, it provides answers.

Okay, so diagnosis is one area, what's the other?

Pharmacogenomics, using genetics to predict drug response, especially for safety.

Safety first, like avoiding bad reactions.

Exactly.

The absolute clearest, most compelling example is the FDA recommendation about the HLA -B158 -O2 allele.

Right, you mentioned this for carbamazepine.

Yes, carbamazepine to Gretel, and also oxcarbazepine, treleptil.

Before prescribing these drugs to someone of Asian ancestry, testing for this HLA allele is strongly recommended.

Because?

Because carrying that specific allele massively increases the risk of developing Stevens -Johnson syndrome, or SJS, a potentially life -threatening skin reaction, knowing the genotype prevents the catastrophe.

That's clear cut.

What about predicting how well a drug works, or the right dose?

That's mostly about drug metabolism, primarily the cytochrome P450 enzymes, especially CYP2D6 and CYP2C19.

People vary genetically in how active these enzymes are.

Some are poor metabolizers, meaning the drug builds up to potentially toxic levels.

Others are ultra -rapid metabolizers, chewing through the drug so fast it might not work effectively at standard doses.

And we can test for that.

We can, and there are clinical guidelines, like from the CPIC group, that advise based on genotype.

For example, avoiding certain triceclic antidepressants if someone is a CYP2D6 ultra -metabolizer, or using much lower doses if they're a poor metabolizer, it helps predict those extreme responses.

Okay, but now for the flip side.

There are companies selling these broad genetic panels,

these decision support tools, or DSTs, claiming to guide all sorts of psychiatric prescribing.

What's the real story there?

Yeah, this is where we need major caution.

These commercial DSTs are widely marketed, often directly to consumers or clinicians.

But be evidence.

The supporting evidence is generally extremely limited.

Many of these tools use proprietary algorithms.

They haven't been independently validated.

The studies often lack replication.

It's not solid science yet for most applications.

So proceed with skepticism.

Absolutely.

There was actually a concerning finding from a trial in adolescent depression.

Doctors who got results from one of these DSTs were sometimes less likely to choose a standard, evidence -based, first -line treatment.

It seems they were swayed by the weak, often ambiguous genetic information, prioritizing it over established clinical practice guidelines, which is worrying.

Yeah, trusting the tech over the established evidence shows the danger.

It really does.

Providers need to be able to critically evaluate these things and not just assume that because it's genetic, it's definitive.

So even with all this incredible progress, there are still major hurdles.

Huge hurdles.

We're still wrestling with the sheer complexity of polygenicity, thousands of genes and pleiotropy genes affecting multiple things.

And fundamentally, we usually can't get direct access to brain tissue during those critical early developmental periods.

There's still a big gap between finding a gene association and truly understanding how it alters brain development or function.

Why does one mutation lead to this outcome and not another?

Think about Huntington's.

The mutant protein is everywhere in the body, but the striatum in the brain is hit hardest.

Why?

We need functional genomics to bridge that gap.

And it sounds like there's a gap in clinical practice, too.

There really is.

We have this strange paradox right now.

The well -validated diagnostic genetic testing for ASD and ID, where guidelines are clear, often underutilized.

Meanwhile, those commercial pharmacogenomic panels with much weaker evidence, frequently overutilized, sometimes driven by patient demand or marketing.

So a real need for better education for clinicians.

Definitely.

Mental health providers need more training to critically appraise genetic evidence and know when and how to use this information appropriately.

Looking ahead, then, where's the field moving?

What's next?

Well, beyond just the DNA sequence, there's huge interest in epigenetics.

That's studying modifications to DNA like methylation that change how genes are expressed without changing the code itself.

And environment plays a role there.

A huge role.

Things like early life stress or adversity can leave epigenetic marks, potentially altering stress response systems long term.

It's about how experience shapes biology.

Fascinating.

And the ultimate goal is heading towards precision psychiatry.

The idea is to integrate everything, genetics, epigenetics, brain imaging, cognitive data, behavioral measures, maybe even data from wearables or smartphones, what's called digital phenotyping.

Wow, all of it together.

All together, potentially using sophisticated tools like machine learning to build predictive models that can guide truly personalized prevention and treatment strategies for each individual.

That's the future hope.

So wrapping up this deep dive into child psychiatric genetics, it's clear how far we've come.

From single genes to tackling complex risk with GWAS and CNV analysis, we see real clinical utility now in diagnosing conditions like ASD and ID and critically in improving drug safety through pharmacogenetics like the HLA testing.

Definitely key takeaways.

But that promise of precision psychiatry, especially using things like polygenic risk scores,

it comes with a major caveat we need to end on.

It does.

And it's a crucial one for everyone to think about.

While PRS could be amazing for personalized medicine down the road, using them widely in the clinic today, given their current massive bias towards European ancestry populations due to the underlying research data,

doing that now would actively exacerbate existing health disparities.

Meaning the benefits wouldn't be shared equally.

Exactly.

It highlights an urgent ethical and scientific imperative.

Researchers must prioritize diversifying genetic studies.

We need data that reflects everyone if we want the benefits of genomic medicine to be equitable.

That has to be a central goal moving forward.

A powerful and necessary point to end on.

Thank you for joining us on this deep dive.

We'll talk to you next time.