TP53 germline variant analysis using functional assays finds reduced-penetrance variants have intermediate activity, higher frequency, later onset, and 106 predicted candidates.
0:00Welcome to Base by Base, the paper cast that brings genomics to you wherever you are. Thanks for listening, and don't forget to follow and rate us in your podcast app. You know, usually when we talk about a medical diagnosis, there is this expectation of absolute precision.
0:15Right, like a clear, binary answer. Exactly. You break your arm, the x-ray shows a jagged white line, and the doctor points to the film and says, there it is. It's broken or it's not broken. The lines are clean, the categories are visible, and as patients, we find that incredibly comforting.
0:31Oh, absolutely. But when you step into the world of genetics, that x-ray machine is, well, it's basically useless. Yeah, you're suddenly looking at a diagnostic landscape that is entirely murky. Take the TP 53 gene, for example.
0:44It's widely considered the most important gene in the human body, serving as the guardian of the genome. When it breaks, your lifetime risk of developing cancer shoots to nearly 100%. It's mask. Right.
0:58You would think a genetic mechanism that dangerous would give you a simple yes or no test result. But today's deep dive reveals a massive, frankly terrifying middle ground. We are looking at what happens when a gene doesn't break completely, but just, you know, bends.
1:12How could this change the way we treat patients stuck in genetic limbo? What really happens when our body's ultimate defense system only sort of works? Today we celebrate the work of an international team of researchers.
1:24including scientists from QYMR Burkofer and Ambri genetics, who have advanced our understanding of the TP 53 gene. They tackled this exact gray area, analyzing genetic mutations that just flat out refuse to fit into our standard diagnostic boxes.
1:39Yeah, and to understand why a spectrum of risk creates such a massive headache for modern medicine, you 1st have to understand the extreme standard version of the disease. Right. So what does a definitively broken TP 53 gene actually look like in a human being?
1:52Well, when someone inherits a standard known pathogenic variant of TP 53, it causes a condition called life from many syndrome, or LFS. Because the primary cellular defense mechanism against rogue cells is just completely gone, the patient's body becomes a highly favorable environment for aggressive tumors.
2:12And these aren't just your standard cancers later in life, right? Not at all. Classic LFS is associated with very early onset, breast cancer, brain tumors, rare cancers of the adrenal glands called adrenal cortical carcinomas, and various sarcomas in the bone and soft tissue.
2:28Wow. So because that cancer risk is astronomically high, the medical response has to be incredibly intense. It is. Patients with standard pathogenic TP 53 variants undergo a really grueling surveillance protocol.
2:40I mean, they're subjected to annual whole body MRIs, annual brain MRIs, regular blood work, and this protocol begins when they are very young children. That sounds like a heavy, lifelong psychological and physical burden.
2:52It really is, which puts an enormous amount of pressure on the diagnostic guidelines. The current genetic guidelines, like those established by the American College of Medical Genetics and genomics, are inherently dichotomous.
3:06Meaning they force everything into 2 categories. Exactly. They demand that every genetic variant be put into one of two main boxes. Benign, meaning completely harmless, or pathogenic, meaning definitively disease causing.
3:20But, you know, biology rarely respects the neat binary categories humans invent. So true. If the laboratory evidence is conflicting or if a genetic mutation doesn't look quite as severe as the classic pathogenic ones, it just gets dumped into a frustrating 3rd box called a variant of uncertain significance, a VUS.
3:39Right. It's essentially the medical system's giant we don't know pile. I think about the TP 53 gene, kind of like the brakes on a car. If you have a standard pathogenic variant, it means your brake lines are totally cut, you are going to crash, it's just a matter of when.
3:50That's great analogy. And if you have a benign variant, your brakes are perfectly fine. Right. But with these variants dumped into the middle pile. It's like asking what happens when the brakes are just um, squishy.
4:02They sort of work most of the time, but maybe they take a little longer to stop the car. How does a doctor treat a squishy break? Yeah, you can't just tell the patient they're completely fine, but you also don't want to overhaul their entire car and put them through annual MRIs if you don't actually have to.
4:17Exactly. The central problem the researchers faced is that our current diagnostic toolkits are built exclusively to detect cut brake lines. They aren't calibrated for squishy ones. So they had to figure out a methodology to track down these elusive, intermediate risk mutations, which they call reduced penetrance variants in the laboratory.
4:36Right, and they started by digging into Clinvar, which is this massive public database of genetic variance. From that mountain of data, they managed to isolate 11 suspected reduced penetrance variants.
4:48Oh, wow, only 11. Yeah, 11. This included a well-known genetic mutation called PR 337 H, which is a famous Brazilian founder variant that has actually puzzle geneticists for years. But I imagine they couldn't just look at those 11 isolated cases in a vacuum, right?
5:03No, exactly. they had to benchmark them So they set up a comparative study, putting those 11 suspected squishy variants up against 62 variants widely accepted as totally benign, and 113 variants known to be standard pathogenic. And to see how these mutations actually functioned, they ran them through 4 massive distinct functional assays.
5:25For anyone listening who isn't in a lab every day, an assay is essentially a controlled test to see how the genetic protein behaves in a biological environment. Yeah, you're taking the genetic code off the computer screen and forcing it to perform a physical task.
5:39The researchers looked at data from 4 independent studies, referred to as Cato, Giacomelli, Cottler and Funk. The Cato essay is particularly fascinating to me because its mechanics highlight exactly what a swishy brake looks like on a cellular level.
5:55It's actually a yeast based test. Wait, yeast? Like for baking? Yeah, well, living yeast cells. The researchers took them and engineered them to rely on the human TP 53 protein. Normally a functional TP 53 protein acts as a transcription factor, meaning it binds to DNA to turn other protective genes on.
6:14Right, it's the trigger. Exactly. So in the Cato assay, they introduced the different human genetic variants into the yeast and measured the exact level of gene activation. And what makes the Cato study unique among the 4 is that it actually features a specific, predefined mathematical category for partial function.
6:32And the resulting data clustered beautifully, didn't it? It really did. When analyzing the Cato classes, the benign variants mostly landed in Class D, meaning they function perfectly, activating the target genes just like normal protein.
6:45Right. And the standard pathogenic variants. They almost entirely grouped into class A, showing complete loss of function. The yeast cells couldn't activate the protective genes at all, but our 11 reduced penetrance variants landed uniquely in the middle.
6:58in class C. They possessed partial transcriptional activity. And it wasn't an anomaly either. Across all 4 independent functional studies, which were measuring different aspects of the protein's physical behavior, the median distribution of scores for these reduced penetrance variants consistently fell right in between the benign and the pathogenic variants.
7:18So they were mathematically and functionally in the middle. Exactly. Okay, so yeast cells and isolated protein assays tell us these variants are acting weird in a vacuum. But biology doesn't happen in a vacuum.
7:30It happens in incredibly complex human bodies. Right. The researchers needed to know if our modern predictive algorithms could actually spot this intermediate risk when looking at the genetic code itself.
7:41They used advanced computational bioinformatics tools, specifically ones called Bayesdell and Alphemous M. And those are algorithms designed to look at the sequence and physical structure of a genetic mutation.
7:53Yeah, they predict whether the mutation will be deleterious or harmful to the patient. But if the algorithm is just looking at the deformed physical structure of the mutated gene. You might imagine it just flags, these squishy variants is completely broken.
8:05That's the logical assumption. Yeah, but it didn't. The algorithms successfully flag them as harmful. But critically, the mathematical risk scores came back notably lower than the scores for standard pathogenic variants.
8:19Oh, that's really interesting. Yeah, the computer models could tell the protein structure was deformed, but, you know, not deformed enough to completely destroy its function. Which perfectly matches the yeast data.
8:28But the metric that truly brings this out of the Petri dish and into human biology is immune fitness. Oh, this part is wild. It is. This measures how visible a mutated cell is to your body's immune system.
8:42Because your immune system's job isn't just fighting off external viruses like a cold. It's constantly scanning your tissues, looking for, and destroying rogue pre-cancerous cells. The mechanism here is so elegant.
8:54When a cell turns cancerous, It usually pushes mutated proteins to its outer surface. Think of it like a cell waving a giant red flag that alerts passing T cells to attack and destroy it. So if you have a benign variant, Your cells are highly visible.
9:08The red flag goes up, and the immune system does its job perfectly. Right. But the classic pathogenic TP 53 mutations, the ones that cause live Romney syndrome. They physically alter the cell so much that it stops presenting those flags.
9:24It's like they have low immune susceptibility. The immune system is basically blind to them. Exactly. The mutated cell sneaks past the defenses and the cancer is allowed to grow unchecked. So if a benign variant waves a red flag, and a classic pathogenic variant is wearing an invisibility cloak, where do these 11 reduce penetrants variants land?
9:45They sit perfectly in the middle? They are less visible than benign variants, allowing them to sometimes evade the immune system, but they aren't nearly as good at cloaking as the standard pathogenic ones.
9:55It like a partial cloaking device. The immune system catches them sometimes, but misses them other times. Yeah, every single angle we look at. The yeast, the algorithms, the immune visibility, it all points to this exact same middle ground.
10:07But then the researchers looked at population data. They did. They tracked allegal frequency using global databases, like gnome ID, to see how often these genetic variants show up in the general public.
10:17And they found that these reduced penitence variants show up slightly more frequently in the general population than the standard pathogenic variants. Which, you know, forces us to ask a difficult question, if these variants are showing up more often in the general public, doesn't that just imply they are harmless?
10:34Like if lots of people are walking around with them, why not just classify them as benign? Well, we have to define what more often means in the context of global genetics. They are slightly more common than the most devastating variants, but they are still incredibly rare overall.
10:50Okay, so why haven't they been wiped out? They haven't been completely wiped out of the human gene pool through natural selection, precisely because they're squishy brakes. The math of evolution is brutal.
11:02Standard pathogenic variants often cause fatal cancers before a person reaches reproductive age, which keeps their frequency in the overall population, extremely low. But because these reduced penetrance variants take much longer to cause disease, The people who carry them live long enough to have children and pass the gene on.
11:20Exactly. Natural selection is literally leading these intermediate variants in the population because they operate on a time delay. Wow. Which transitions us to the most critical part of the study, the clinical phenotypes.
11:33How do these functional assays and evolutionary statistics translate into actual health outcomes for living patients? This is where it gets very real. The researchers analyzed a massive data set provided by Ambri genetics.
11:46We are talking about over 256,000 control patients who did not have a pathogenic variant. That's a massive data set. It is. They compared them to 505 people carrying standard pathogenic variants and 203 people carrying our reduced penetrance variants.
12:02256,000 control is definitely large enough to draw some very hard conclusions. So does this middle ground actually show up in real patients' medical records? It absolutely does. And the average age of a patient's 1st cancer diagnosis perfectly illustrates the spectrum.
12:18For the control group without the mutation. The average age of 1st cancer diagnosis was 50.3 years old. Okay, 50.3. And for the standard pathogenic group, the classic LFS patients. Their average age was 36.6 years old.
12:34And their reduced penetrance group. Their average age of first cancer diagnosis was 47.7 years old. Wow. 50, 36, 47. The clinical reality perfectly mirrors the laboratory spectrum. It really does. The researchers also looked at odds ratios to measure how heavily the disease is enriched within each group.
12:53Okay. Instead of using statistical jargon, let's make this concrete for the listener. If you take a large group of average women under the age of 31, a suit and small number will inevitably develop breast cancer.
13:05But if you look at women who carry the standard pathogenic TP 53 mutation, they are 16 times more likely to get breast cancer before their 31st birthday, than the average person. That's an oz ratio of roughly 16.
13:18But for the reduced penetrance variants, that massive risk drops significantly. You are only 3 times more likely to get breast cancer under 31 compared to the general population. Still higher than normal, obviously.
13:30Oh, yeah. Meaning these variants are definitely dangerous. But it is vastly different from the 16 times multiplier of the standard risk. Which creates an absolute clinical nightmare for the patient sitting in the doctor's office.
13:43The current system forces doctors to make a binary choice. If a patient's squishy brake variant is misclassified by a diagnostic lab as a variant of uncertain significance or is benign. They get 0 preventative care.
13:57Right. You are sent home thinking your genetics are totally fine, and you might get blindsided by an aggressive tumor in your 40s because nobody told you to look for it. And the reverse scenario is just as damaging.
14:06If a diagnostic lab sees that the variant is somewhat deleterious and decides to dump it into the standard pathogenic bucket. That patient is suddenly given a lie from many syndrome diagnosis. Which means they are subjected to aggressive, terrifying annual MRIs, immense psychological stress, and potentially even preventative surgeries that they might not actually need.
14:28Because their risk isn't near 100%. And it likely won't strike in their 20s. The classification system is actively failing this middle group of patients. It is. And the researchers knew that if identifying just these 11 variants took this much manual digging, cross-referencing and laboratory testing.
14:46There was no way human researchers could manually sort through the 1000s of unknown variants currently sitting in genetic databases. So if you want to find the rest of the squishy brakes, you need a detective that processes data at an unimaginable scale.
15:01So they turn to machine learning. Exactly. The researchers took all those disparate data points, the functional yeast assays, the bioinformatics algorithms, the immune fitness metrics, the illegal frequency, and they fed them into a machine learning algorithm called a random force model.
15:17A random force model. How does that work? Well, it essentially builds 100s of decision trees, each looking at different features of the data, and then they all, like, vote on the outcome. They taught the artificial intelligence what a completely benign variant looks like, what a completely broken pathogenic one looks like, and crucially, what a reduced penetrance variant looks like across all those different biological dimensions.
15:39And when they analyzed what features the AI relied on most heavily to make its categorizations, the functional assays, specifically the Cato yeast assay and the funk assay, proved to be the most influential.
15:51So the AI weighted the actual physical behavior of the mutated proteins far higher than just the population statistics or the basic structural algorithm. That makes a lot of sense. So they trained this AI detective and then unleashed it on 590 TP 53 variants currently sitting in Clinvars uncertain or conflicting pile, that massive frustrating VOS box.
16:15And what did it find? The model flagged 106 new suspected reduced penetrance variants, 106 genetic mutations currently in diagnostic limbo, which the AI believes actually belong in this critical middle category.
16:28That's incredible. But, um, we just establish how messy human biology and immune systems are. How can we possibly trust a machine learning algorithm, which was only trained on 11 confirmed reduced penetrance variants to make a life or death classification?
16:40Are we really ready to use this random forest model in a doctor's office tomorrow? Like running a patient's genetic test through it to give them a customized risk percentage? We absolutely cannot. And the researchers themselves are exceptionally clear about the limitations of this study.
16:55This is a preliminary model. As you pointed out, it was trained on a very small sample size. Right. And the error rate for categorizing the reduced penetrance variants was around 30%. It is not a final diagnostic tool ready for clinical use.
17:10Okay, so if it cannot be used to diagnose a patient, What is the practical application of the model? Think of it as a treasure map. It is pointing researchers directly toward the 106 variants they urgently need to study next.
17:23Ah I see. Instead of randomly testing the 1000s of uncertain variants out there. Diagnostic labs now have a prioritized hit list. You know exactly which genetic sequences to run through formal penetrant studies, comprehensive family history evaluations, and clinical modeling to confirm their actual risks.
17:40It's the starting line for the next decade of genomic research rather than the finish line. It proves that we do not necessarily need entirely new diagnostic technologies. We just need to look at our existing tools with much more granularity.
17:55Yeah, we need to build systems that recognize intermediate function instead of forcing every complex biological process into a simple loss of function or normal function binary. The central insight here is that our genetic destinies are rarely black and white.
18:09From partial protein function in a yeast cell to the immune system's blind spots to delayed cancer onset in human patients, disease exists on a vast, complicated spectrum. And as genetic panel testing becomes cheaper and a more routine part of standard healthcare, more and more people without traditional family histories of disease are going to get their genomes sequenced.
18:31Right. And when they do, they will increasingly find them themselves sitting in this exact middle ground. What does this mean for something as fundamental as family planning when our exact shades of gray leave us waiting for a shoe to drop later in life?
18:43It's a profound question If genetic sequencing tells you that you carry a classic 100% fatal variant that strikes in childhood that radically alters your decision to have biological children. But what if you find out you have a squishy brake, a variant that might not cause cancer until you are 48, and even then, only at a moderately increased rate?
19:03How does knowing that precise intermediate percentage change who you date when you marry, or whether you decide to pass those genes onto the next generation? Science is getting incredibly good at giving us the raw data, but is leaving us to navigate the deeply personal consequences of that information entirely on our own.
19:22This 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 5 star rating.
19:36If you'd like to support our work, use the donation link in the description. Now stay with us for an original track created especially for this episode and inspired by the article you've just heard about.
19:46Thanks for listening and join us next time as we explore more science base by base.
