Episode Transcript
[00:00:00] Speaker A: Foreign.
[00:00:14] Speaker B: Welcome to Base by Base, the papercast that brings genomics to you wherever you are. Okay, Imagine a newborn, just tiny and vulnerable, needing life saving heart surgery. It's a really tough reality for well over 40,000 babies in the US every single year and globally. These congenital heart defects, CHDs, they affect roughly one in a hundred live births. It's quite common. Now, surgery offers incredible hope, obviously, but predicting how well a child will recover or if they'll face, you know, serious complications like cardiac arrest or needing a ventilator for a long time, that's always been an incredibly challenging puzzle for doctors. But what if, what if we could look into that child's unique genetic blueprint before the surgery and get a much clearer picture of their individual risk? How could that kind of foresight actually transform their care, maybe even save lives?
[00:01:02] Speaker A: That's exactly the powerful question a new deep dive into genomics is starting to answer. And this isn't just about diagnosis anymore. It's really about a new level of prediction. Absolutely. And this incredible leap forward, it's largely thanks to the groundbreaking work of the Pediatric Cardiac Genomics Consortium, the pcgc, and also the dedicated researchers from places like the University of Utah. Their collaborative efforts, meticulously bringing together just vast amounts of genetic and clinical data, are genuinely pushing the boundaries of personalized medicine. It's offering this new level of critical foresight in pediatric cardiac care.
[00:01:39] Speaker B: Okay, pushing boundaries, that's a huge claim for someone like me, maybe just vaguely remembering high school biology. What does that actually mean, you know, in practical terms for these tiny patients and their obviously very worried families? Because for clinicians, predicting outcomes and CHDS has always been tough, right?
[00:01:56] Speaker A: Oh, absolutely.
[00:01:57] Speaker B: These are complex, often life threatening disorders. And despite all the amazing surgical advances we've seen, just the sheer variation in how severe the defect is and all the intricate medical stuff required, it makes forecasting recovery a real challenge. We kind of understood the what that genetics matters in CHDs, but the how much and for whom that remained elusive. It's like having puzzle pieces but no picture on the box. Right. So how did these researchers finally start connecting those elusive pieces?
[00:02:26] Speaker A: You've really hit on the core problem there. Previous large scale genetic studies, including earlier work by the PCGC itself, show that the genetic landscape of CHT is incredibly diverse.
So to give you an example, we've known for a while that dominantly inherited forms are often linked to damaging variants in genes that modify chromatin.
[00:02:46] Speaker B: Chromatin, that's the packaging for DNA, right?
[00:02:48] Speaker A: Precisely. Think of it as the system that packages Organizes our DNA within our cells. Very important stuff. And then on the flip side, you have recessive forms. These are frequently enriched for variants in genes related to cilia.
[00:03:00] Speaker B: Cilia? Those tiny hairs on cells?
[00:03:02] Speaker A: Yeah, exactly. Those tiny hair like structures on cells all over our bodies. They're crucial for everything from clearing airways to sensing signals. But, you know, despite knowing these kinds of links, a broader practical understanding of genetic testing's predictive power across the full spectrum of CHD types and clinical situations, well, that remained out of reach. That's the challenge this new research really tackles head on.
[00:03:26] Speaker B: Okay, so we're talking about taking all that raw genetic and clinical information and making it truly actionable, making it useful for predicting a child's unique path after surgery. How did they even begin to manage something so complex? It sounds like just an enormous amount of data.
[00:03:41] Speaker A: It was. And what's truly fascinating here is their sophisticated approach to handling this complexity. They used a really massive prospective observational cohort study from the PCGC. We're talking 2,253 CHD patients, and each patient came with a wealth of information, pre and post operative clinical variables, plus their exome sequencing data.
[00:04:01] Speaker B: The exome, that's the protein coding parts of the genome?
[00:04:04] Speaker A: That's right. Basically reading all the protein coding parts. So this isn't just a large data set. It's a really deep dive into individual patient journeys that makes the potential findings incredibly robust. Now, to manage the sheer diversity of the heart defects, they first did something innovative for what they call phenotype classification.
So instead of looking at every single type of heart defect individually, which frankly, would be overwhelming, they condensed these different CHD phenotypes into five clinically relevant categories.
[00:04:33] Speaker B: Okay, like what?
[00:04:34] Speaker A: Things like left ventricular outflow tract obstructions or LVOs, heterotaxylaterality defects, atrioventricular canal defects, conotruchal defects, and then a category for other defects. And they didn't just, you know, eyeball it. They used an explainable AI approach, specifically an XG boost model.
[00:04:52] Speaker B: Like a really smart sorting algorithm?
[00:04:54] Speaker A: Kind of, yeah. This AI was trained on how expert physicians classify these heart defects and then probabilistically assign each patient to one of those categories. This made the data much more manageable for analysis. It gave them a structured, reliable way to group different types of heart defects.
[00:05:09] Speaker B: Okay, so they streamlined the clinical picture with AI. That makes sense. What about the genetic side? That must be even more complex with all those variations we carry.
[00:05:17] Speaker A: Oh, absolutely. So to overcome that high genetic variation, they used another AI based tool. This one was for genome interpretation called gm. Think of GM as like an advanced genetic scanner. It's designed to flag particularly damaging genotypes.
[00:05:33] Speaker B: Damaging genotypes meaning ones likely to cause problems?
[00:05:36] Speaker A: Exactly. Those genetic variations that have a strong statistical and biological likelihood of causing disease or impacting function.
Specifically, they look for variations with a GM score of 1.0 or higher.
That indicates really strong evidence for pathogenicity. And once they identified these, these damaging genotypes were then categorized into specific molecular pathways or gene types. It's essentially making sense of the vast genetic information by grouping it into biologically meaningful categories. Yeah. Okay. And finally, and this is where the predictive power truly comes into play. They connected all these refined phenotypic and genotypic classifications using powerful statistical tools, probabilistic graphical models, specifically Bayesian networks.
[00:06:18] Speaker B: Bayesian networks. Okay, that sounds complex. Like a detectives board.
[00:06:21] Speaker A: That's actually a great analogy. Imagine that board where every clue a patient's age, their specific heart defect category, and now these genetic red flags is linked together. It builds a probability map of potential outcomes.
These networks are incredibly powerful because they can model really complex dependencies, even nonlinear relationships.
And they're particularly well suited for making predictions when data points for certain rare outcomes are limited.
Which, you know, is often the case in rare diseases like complex ehd.
[00:06:51] Speaker B: Right, right. Makes sense.
[00:06:53] Speaker A: So after all that incredible work categorizing heart defects, decoding genetic variations with AI, building these powerful predictive models, what insights did they finally uncover? What did that crystal ball reveal for these children and their families?
[00:07:07] Speaker B: Well, the first major revelation was just how prevalent these genetic links actually are. They found that damaging genotypes were identified in 238 participants. That's about 10.6% of the total group. Yeah, that's roughly 1 in 10 children with CHD carrying a hidden genetic risk factor. They found new de novo variantso genetic changes not inherited, but arising spontaneously in the child in important genes like KMT2D and CHT7. And they also saw biallic genotypes. That's where a child inherited a damaging variant from both parents for the same gene, like dync2h1 and dnah5, both types.
[00:07:44] Speaker A: Pointing to different genetic pathways that can influence outcomes.
[00:07:46] Speaker B: Interesting.
[00:07:47] Speaker A: And digging deeper into that, they saw significant genetic enrichment in specific CHD phenotypes. For instance, that left ventricular outflow tract obstructions class, the LVO group, especially hypoplastic left heart syndrome. Well, it was nearly twice as likely to have those damaging de novo genotypes in the chromogen modifying genes we talked about earlier.
[00:08:06] Speaker B: Ah, okay, so it highlights how certain heart defects are strongly driven by specific genetic changes.
[00:08:11] Speaker A: Exactly. And what's truly striking for patients with heterotaxia, that complex condition where internal organs are abnormally arranged, they found a 2.6 fold enrichment for damaging recessive or biallelic variants in those cilia related genes.
[00:08:23] Speaker B: 2.6 times higher.
[00:08:24] Speaker A: Yes. And what's even more remarkable is that for those specifically linked to something called the FoxJ1 pathway, the enrichment was a staggering 6.9 fold.
[00:08:34] Speaker B: Wow. Nearly seven times.
[00:08:36] Speaker A: It really reinforces prior knowledge. But with new incredibly specific numbers, it shows a direct genetic vulnerability there.
[00:08:43] Speaker B: So these damaging genotypes weren't just present, they had a clear, measurable impact on, well, bad outcomes. This is where it really gets tangible. For families, I think.
[00:08:53] Speaker A: Absolutely.
[00:08:54] Speaker B: Damaging genotypes in both those chromatin modifying and the cilia related genes significantly increased the risk of significant severe post operative complications. For the chromatin variants, we're talking about nearly doubling the risk of mortality.
[00:09:06] Speaker A: Doubling.
[00:09:06] Speaker B: Yeah, and increasing cardiac arrest risk by 1.7 fold and prolonged mechanical ventilation by 1.6 fold. And for the cilia variants, the risks were also significantly elevated across the board. But here's something crucial, maybe often overlooked. What isn't there can be just as powerful as what is.
[00:09:24] Speaker A: Hmm. Tell me more.
[00:09:25] Speaker B: Well, if a child lacks a specific damaging genotype, say they don't have a damaging de novochromatin genotype, their risk for mortality after surgery plummets by nearly half.
[00:09:36] Speaker A: So a relative risk of about 0.55, that's significant.
[00:09:39] Speaker B: Exactly. Think of it like getting a powerful all clear sign. It offers incredible reassurance. Right. And maybe it can guide less aggressive interventions when that's appropriate. It's not just about finding risk, but also identifying lower risk.
[00:09:53] Speaker A: That's a really key insight. And these impacts were often amplified by other factors, making the predictions even more precise. For example, for the highest risk surgical procedures, the ones categorized as stat 4 or stat 5, patients who unfortunately died were nearly twice as likely to have a damaging chromatin variant and 1.7 fold more likely to have a damaging cilia variant.
[00:10:14] Speaker B: So it really highlights how genetics plays a critical role in the most challenging, most vulnerable cases.
[00:10:19] Speaker A: Precisely. And if we connect this back to the specific CHG phenotypes, the numbers get even more stark. LVO patients who died were 2.3 fold more likely to have a damaging chromatin genotype. And those HTX patients, the heterotaxia ones, who died were 2.8 fold more likely to have a damaging cilia genotype.
[00:10:38] Speaker B: That level of granularity, being able to predict risk that specifically is truly remarkable, allows for really tailored assessments.
[00:10:45] Speaker A: Absolutely. Furthermore, the presence of extra Cardiac anomalies, or ECAs, those are other structural problems not related to the heart, significantly amplified the risk.
Patients with ECAS already showed a 2.8 fold increased mortality risk and a 1.7 fold increased risk of prolonged ventilation just baseline.
[00:11:02] Speaker B: Okay.
[00:11:03] Speaker A: But what's even more telling is having a damaging chromatin genotype made it 2.5 fold more likely to have an ECA and die.
[00:11:10] Speaker B: Wow. That combination.
[00:11:11] Speaker A: Yeah. And for HTX patients who also had an ECA, having a damaging cilia genotype increased the probability of prolonged ventilation by a substantial 4.0 fold.
[00:11:21] Speaker B: Four times higher.
[00:11:21] Speaker A: It's this layered approach to risk assessment that's really a potential game changer.
[00:11:25] Speaker B: Okay. Wow. That's a lot of incredible detail. It paints such a vivid picture of risk. But for me, I'm thinking about that moment in the hospital.
How does this information shift, shift from just a nice to know genetic insight to something a surgeon or a pediatric cardiologist can actually use? Maybe even today? What does personalized risk stratification look like on the ground?
[00:11:48] Speaker A: That really is the million dollar question, isn't it? And this deep dive clearly defines a critical role for genome sequencing and forecasting outcomes after congenital cardiac surgery. Especially, like we said, in the context of those higher risk procedures, specific CHD types, and the presence of those extra cardiac anomalies. It's about giving clinicians a clearer, more precise roadmap for each child. The power of combining that explainable AI framework with the Bayesian networks. Well, it allowed these researchers to unravel incredibly complex dependencies among all these diverse clinical and genetic factors. Things that were hidden before exactly hidden connections. This provided detailed risk estimates that were frankly, previously unattainable, given the inherent data limitations you always face in rare disease studies. And Bayesian methods are especially useful for these low data scenarios because they can incorporate prior knowledge and. And also model uncertainty. It's a significant leap forward in understanding individual patient risk.
[00:12:45] Speaker B: And it also seems to refine our biological understanding too. Right. Reinforcing and maybe adding new layers to what we already knew.
[00:12:51] Speaker A: That's right. For example, it really showed that hypoplastic left heart syndrome is a principal driver of those chromatin related genetic signals, making that connection even stronger.
And that these cilia related gene variants are strongly tied to heterotaxy, particularly impacting postoperative respiratory complications.
So it's not just predicting what might happen, but also helping understand why. And that understanding opens doors for very targeted interventions.
[00:13:17] Speaker B: This sounds incredibly promising, but let's get practical again. If a clinician gets this genetic information before surgery, what's the immediate actionable step? Can you maybe walk us through a hypothetical scenario? Let's say a child comes in with a heterotaxy defect. How does this new genetic insight actually change the conversation a doctor has with the family and change the care plan for that child?
[00:13:41] Speaker A: That's precisely where the rubber meets the road. So take your example. Preoperative knowledge of a damaging cilia genotype in a child with heterotaxy. Knowing that child is predisposed to respiratory issues because their cilia don't work. Quite right.
Well, that could allow for incredibly early proactive interventions.
[00:13:59] Speaker B: Like what kinds of things?
[00:14:01] Speaker A: We're talking about maybe starting aggressive airway clearance therapy before surgery. Or using eucalyptic therapies, medicines to help clear airways. Or even employing specific ventilation strategies after the operation. Strategies designed specifically to promote mucociliary function. It can even inform choices in anesthesia, maybe avoiding certain agents known to impair cilia function.
[00:14:21] Speaker B: I never would have thought of that connection.
[00:14:23] Speaker A: Right. It translates directly into actionable preventative care right at the bedside, optimizing every single step for that child from the moment they enter the hospital.
It's really about moving from just reactive treatment to proactive, personalized prevention based on their unique genetics.
[00:14:41] Speaker B: That's truly remarkable foresight, but, you know, every study has its boundaries. What were the limitations here? And where do we go from here? What's the next frontier for this kind of research?
[00:14:50] Speaker A: Yeah, that raises important questions for future research.
And the study authors were quite upfront about limitations. For instance, the cohort wasn't technically an inception cohort. That means some very early deaths, perhaps before enrollment, might have been missed. That could potentially lead to an underestimation of morbidity.
[00:15:08] Speaker B: Okay, that makes sense.
[00:15:09] Speaker A: It was also underpowered for some of the less severe CHD phenotypes.
So its predictive power is currently strongest for the more complex cases.
And while large clinical databases are invaluable, they can sometimes have data quality issues, which the researchers had to navigate carefully. Replication in an independent external cohort is always the gold standard. Right?
[00:15:31] Speaker B: Right.
[00:15:32] Speaker A: But because of the unique depth and scale of this PCGC dataset, it's actually challenging to find an exact match elsewhere. However, the findings align exceptionally well with existing biological and clinical literature, which gives them strong plausibility it fits with what we know. Looking ahead, the future really calls for even larger cohorts that would help capture those rarer associations and also gain more power for less severe phenotypes. We also need to move beyond just exome sequencing to hold genome sequencing. Eventually look at all of a person's DNA, not just the protein coding bits.
[00:16:04] Speaker B: Get the bigger picture.
[00:16:06] Speaker A: Exactly. And implementing new technologies like long read sequencing will provide an even more comprehensive and precise view of complex genetic variations. As genomics continues to get faster and cheaper and it is rapid, genome sequencing is truly poised to become the standard of care for all critically ill newborns. This will empower clinicians with unprecedented personalized risk stratification. It allows us to proactively tailor care for every single child, moving towards a truly individualized approach to pediatric medicine.
[00:16:35] Speaker B: So what does this all mean for you, the listener?
Well, the central insight from this deep dive seems clear.
Genomic sequencing isn't just for diagnosing conditions anymore. It's emerging as a powerful, critical tool for forecasting clinical outcomes after congenital heart surgery. It offers truly personalized risk assessments for patients.
Knowing a child's genetic risk, and crucially, even knowing about the absence of certain damaging genotypes empowers clinicians. It lets them deploy targeted therapeutic strategies and optimize care, potentially leading to significantly better outcomes and ultimately saving lives. It makes you wonder, what does this mean for the future of precision medicine and pediatric surgery? And how quickly can these powerful insights become standard practice for every child who needs it? This episode was based on an Open Access article under the CCBY 4.0 license. You can find a direct link to the paper and the license in our episode description if you enjoyed this, follow or subscribe in your podcast app and leave a five star rating. If you'd like to support our work, use the donation link in the description. Thanks for listening and join us next time as we explore more science Base by base.
