A longitudinal study of recipients of medically actionable secondary genomic findings develops a Bayesian approach that integrates variant, family genotypic, and phenotypic data to estimate the probability that a secondary finding represents a true clinicomolecular diagnosis, with a detailed analysis of BRCA1/BRCA2 families and implications for screening policy and clinical management.
0:00Welcome to Base by Base, the papercast that brings genomics to you wherever you are. Thanks for listening, and don't forget to follow and rate us in your podcast app. So, um, I want you to put yourself in a very specific scenario.
0:13Imagine you decide to take a DNA test. Like one of those ancestry kits or something? Exactly. Maybe you bought a kit online to figure out, you know, where your ancestors came from or maybe you volunteered to give a saliva sample to a biobank for a massive research study.
0:27So you spit in the tube, you drop in the mail, and honestly, you completely forget about it. Right. out of sight, out of mind. Exactly. But a few months later, your phone rings, and it's the testing facility, and they are not calling to talk about your family tree.
0:41The voice on the other end informed you that they found a medically actionable genetic variant. Oh, wow. That's a heavy call. Right, like a mutation in a gene like BRCA one or BRCA two, which, you know, are notorious for their link to hereditary breast and ovarian cancer.
0:58I mean, receiving a call like that completely blindsides you. It flips your entire understanding of your own health upside down in a matter of seconds. And my immediate reaction, and I think the reaction of almost anyone listening, would be sheer panic.
1:13You weren't looking for a life altering health diagnosis, but suddenly it feels like you have one. Or do you? Right. that's the big question Yeah, is that surprise result a 100% guarantee of the disease syndrome?
1:27What really happens when we uncover severe genetic risks in people who weren't even looking for them? We really have to answer that question and quickly, because this scenario is playing out constantly right now, genetic testing has just become so cheap and accessible.
1:40Well, a new study reveals that the actual odds attached to those surprise phone calls are far more complicated and honestly way more fascinating than we thought. Yeah, our fundamental understanding of how to interpret a surprise genetic test is going through this massive evolution right now.
1:54Today we celebrate the work of Julie C. Sapp, Leslie G.B. Sekker, and their expansive research team at the Center for Precision Health Research at the National Human Genome Research Institute, as well as Geiseigel, who have advanced our understanding of disease likelihood in genomic ascertainment.
2:12The sheer scale of what they pulled off and the rigor is just remarkable. And, uh, this open access research was actually published in the American Journal of Human Genetics in May 2026. So to really grasp why this paper changes the game, we need to, I guess, look at the clinical problem it addresses, right?
2:31Like when a lab finds a scary gene variant in your DNA. Why is that considered a secondary finding in the 1st place? Well, it comes down to the guidelines set by the American college of medical genetics and genomics, the ACMG.
2:43When doctors or researchers sequence a genome, they generate just a massive amount of data. And if they're looking at your DNA for one specific reason, say, trying to figure out why you have some rare metabolic issue, they might accidentally stumble upon a mutation in an entirely unrelated gene.
2:58Right. So that's the secondary finding. Correct. And the ACMG has this specific list of genes that they highly recommend reporting back to the patient if they are found incidentally. Because they're actionable.
3:10Right. Exactly. These are genes linked to severe but preventable conditions. So, cancer predispositions, sudden cardiovascular risks, things like that. The logic is preventative. You know, if we spot a massive sinkhole on the road ahead, we have an obligation to warn you so you can hit the brakes.
3:28I mean, that logic makes total sense on paper. But the context of how we find that sinkhole matters immensely. Let's talk about selection bias. Because finding a pathogenic variant in a high risk breast cancer clinic is a very different beast than finding it in a random, healthy person.
3:44completely different. Okay, let's unpack this. It's like fishing. If you drop a line in a heavily stocked commercial trout pond, which is our diagnostic clinic. A tug on the line is almost certainly a trout.
3:54Right. You know what's in there? Exactly. But if you drop a line in the middle of the open ocean, the general population, a tug could be anything, including, you know, a false positive. That is a brilliant analogy.
4:06The location completely changes the prior probability. Prior probability. Yeah, so in statistics, prior probability is just your baseline chance of an event happening before you introduce new evidence.
4:18So in the trout pond, the prior probability of catching a trout is huge. In the open ocean, it's incredibly low. Translating that to medical genetics, you know? If a patient walks into a clinic with, say, 3 ants who had early onset breast cancer, The prior probability that a BRCA variant in her blood is truly causing a family cancer syndrome is massive.
4:40Because the family history already set the stage. Exactly. But when you test the general public through, like, biobanks, the prior probability drops to the general population's incidence rate of the disorder, which, you know, might be one in 400.
4:52So a piece of paper from a lab that says pathogenic variant does not mean the person actually has the clinical syndrome, the tug on the line in the ocean might just be, I don't know, a piece of seaweed.
5:02Right. The lab is accurately reporting the chemical makeup of the DNA, sure, but they cannot tell you the biological reality of how that DNA is behaving in your specific body or, you know, your family.
5:13And that disconnect is what this research team wanted to quantify. Yes. They wanted to know when a secondary finding pops up in the general population. What is the actual mathematical likelihood that the person truly has the disease syndrome?
5:28Measuring that sounds like an absolute logistical nightmare. I mean, you can't just send out a simple survey for something this complex. Oh, not at all. They had to cast an incredibly wide net and then filter it down ruthlessly.
5:40They actually recruited participants who received secondary findings from 41 different sources. Wow. 41. Yeah, including biobanks, direct to consumer genetic testing companies, telehealth counseling services, and all sorts of research studies.
5:55Getting data from 41 different pipelines. I mean, they must have had 1000s of people to soar through. They started with roughly 1500 inquiries. But, you know, rigorous science requires strict boundaries.
6:08They independently verified the lab reports, making sure every single variant was officially classified as pathogenic or likely pathogenic. Oh, so no ambiguous results. Right. And they also had to confirm these were true secondary findings, meaning the person wasn't like secretly getting tested because they already felt sick or had a long...
6:28They had to be truly incidental. After putting all 1500 through that gauntlet, how many actually made the final cut? 163 individuals. Okay, quite a drop. Yeah. And in genetic research, these starting individuals are called pro bands.
6:41Out of those, they completed deep data collection and genotyping for all 163 of them. And they did cascade testing too, right? Yes, exactly. So just to clarify, cascade testing means mapping out the family tree.
6:54Like testing the parents, the siblings, the grandparents, to see who else carries the variant and who actually develops symptoms. You got it. They needed to move beyond just looking at the DNA in isolation.
7:06And the critical innovation in this paper is how they synthesize all that family data. The team developed a Besian mathematical model to calculate the exact percentage likelihood of something called a clinical molecular diagnosis or CMD.
7:20Okay, wait, before we plug in the numbers, let's define Besian for a second. It sounds like this intimidating math jargon, but it's really just a formalized way of updating our beliefs as new evidence comes in, right?
7:31a great way to put it. Like a detective solving a mystery. If you know a crime happened in a city of a 1000000 people, your baseline probability for any random person being the culprit is tiny. Right. But if you find out the suspect has red hair, your probability updates and narrows.
7:47If you find out they drive a blue truck, it updates again. The math just keeps multiplying the new clues against the old baseline. That is the perfect analogy. The Besian model takes a starting probability and continuously refines it as new data points are introduced.
8:00In this study, they wanted to calculate the probability of a CMD. A clinical molecular diagnosis. Exactly. Meaning you don't just have the molecular mutation, the typo and the DNA, but your family also exhibits the physical clinical symptoms that the mutation is supposed to cause.
8:17And to test this, they focus deeply on 59 families who had a secondary finding of a BRCA one or BRCA 2 variant. Okay, I want to see this math in action. Because the paper details a specific case. Family 83334 that totally blew my mind.
8:33Walk me through what the Besian model did with this family. Yeah, Family 8334 is the ultimate showcase of why this model matters. The pro band our starting patient was a healthy 38 year old woman, 0 symptoms of cancer.
8:45She took a genetic test, and boom, a likely pathogenic variant in her BRCA 2 gene is flagged. I mean, if I'm 38 and healthy and I get that result. I am immediately assuming I have hereditary cancer syndrome.
8:57Understandably. But the Bayesian model doesn't care about my panic. What was her actual baseline probability before they looked at her relatives? Factoring in general population data and the specific variant, her baseline probability of having the true hereditary cancer syndrome was calculated at 58.2%?
9:13Wait, 58.2%? It's basically a coin flip. Pretty much. The lab report feels like a death sentence, but the math says it's only slightly more likely than getting heads on a quarter. And this is where the cascade testing changes everything.
9:26The researchers went up the family tree and tested her mother, and the mother carried the exact same BRCA to variant. Okay, let me guess. When they looked at the mother's medical history. The completely healthy picture started to fall apart.
9:42It did. The mother had actually been diagnosed with breast cancer at age 47. Oh, wow. Okay. That is a massive clue. So how does the model handle that? The Besian model takes that new clue, a positive genotype and a cancer diagnosis at a relatively young age and updates the 58.2% baseline.
10:00But they went even further. They tested the maternal grandmother. She also had the variant. And she had been diagnosed with breast cancer at age 55. Jeez. So we now have 3 generations with the variant, and 2 of them have early onset cancer.
10:13The detective has a mountain of evidence now. What did the math say? When the model plugged this family history into the formula, it pushed the family's probability of actually having the hereditary cancer syndrome to 99.2%.
10:2799.2%. The tug on the line was definitively a trout, or in this case, a true hereditary syndrome. The 38-year-old might be healthy now, but her personal risk is verified as incredibly high. Exactly. The math worked perfectly to validate her risk.
10:43But, you know, the real shock came when they applied the same formula to everyone in the 59 families in the BRCA group. Right. Did they all land in the 90s? Not even close. The probability that these families actually had the hereditary cancer disorder varied wildly.
10:57The results ranged from a mere 26.2%, all the way up to 100%. Wait, 26.2%. You're telling me there are people walking around holding a medical document that says they have a pathogenic cancer mutation, and there's barely a one in 4 chance their family actually suffers from the syndrome.
11:12That is exactly what the data shows. The average across all the families was 86.9%. But, I mean, an average is completely useless to the individual patient sitting in a clinic who happens to be at 26%.
11:25That variance is staggering. It completely shatters the illusion that a genetic test gives you a simple yes or no answer. It really does. And looking closely at the findings, there was another data point that jumped out at me, regarding the specific genes involved.
11:39Usually, if you walk into a clinical testing setting, doctors find a lot more BRCA1 mutations than BRCA2. Right. Usually it's about 66% BRCA1. But in this study, the ratio was entirely flipped. The SKU was incredibly pronounced.
11:53A whopping 83% of the variants were BRCA2. and only 17% were BRCA1, which is highly skewed compared to standard clinical testing. Why such a drastic reversal? Are BRCA 2 variants just more common in the general population?
12:09It all comes down to penetrance. Think of penetrance as the aggressiveness of the gene. The RCA1 is highly penetrant. It acts like a blaring siren. Right, it's very obvious. Yeah, if you have a BRCA one variant.
12:20The cancer risk is very high, and the onset is usually very young. Because it burns through a family tree so aggressively. those families know they are sick. They end up in diagnostic clinics. And BRCA too?
12:32BRCU 2 has slightly lower penetrance. It's more of a quiet, intermittent beep. People might carry the BRCA 2 variant without realizing it because the cancer might strike later in life or skip a generation, making the family history look, you know, less devastating. So when you cast a massive net into the open ocean.
12:50You scoop up all these hidden lower penetrants BRCA 2 variants that never triggered a clinical alarm bell. Exactly. But here is the piece of data that should give every healthcare professional pause. We keep calling these secondary or surprise findings, assuming these patients had no idea they were at risk.
13:06But when the researchers analyzed the family histories, they discovered that 501% of these families actually already met the NCCN diagnostic criteria for hereditary cancer testing. Here's where it gets really interesting.
13:19Over half. Wait, let's define the NCCN criteria for the listener. The national comprehensive cancer network. Essentially provides the checklist that doctors use, right? Yes. Like if you have a certain number of relatives with cancer, at certain ages.
13:34You check the boxes, and your doctor is supposed to order a primary genetic test. That is the established protocol, yes. But if over half, 51% of these families actually already met the criteria based on their family history alone, why didn't their doctors catch this earlier?
13:51Does this mean secondary screening is actually just catching miss primary cases? Yes, exactly. You are entirely justified in calling it a massive healthcare system failure regarding genetic testing utilization.
14:03That's crazy. The researchers point to several compounding issues. First, think about modern primary care. A patient gets a 15 minute appointment. The doctor is checking blood pressure, renewing statins and rushing to the next room.
14:15They simply do not have the time to sit down and draw a three generation family tree. And even if they did have the time, the patient has to actually know the history. I barely know what my cousins do for a living, let alone the exact age my great aunt was diagnosed with a tumor.
14:29Exactly. Patient recall is a huge barrier. Add to that, the fact that genetic literacy among non-specialist physicians is still catching up. A doctor might hear a patient mention a mother with breast cancer, but fail to connect it to the grandmother's ovarian cancer, missing the pattern entirely.
14:46So these people were falling through massive cracks, and opportunistic screening just happened to catch them. which is great for them, but it brings us back to the clinical dilemma. If the odds of disease range from 26% to 100%.
15:00How should this change what doctors actually do with patients? It demands a complete paradigm shift in clinical practice. Historically, seeing the word pathogenic on a lab report might trigger an automatic, highly aggressive medical response.
15:13But this study proves that a pathogenic variant doesn't equal a 100% guarantee of disease. A patient with a 26.2% probability should be managed very differently than someone with a 99.9% probability. We want to avoid unnecessary prophylactic surgeries based on false or low probability assumptions.
15:33Prophylactic surgery. Meaning preventative removal of the breasts or ovaries to stop the cancer before it can ever start. Yes. For the 99% patient like our 38 year old and family 8334, who might prudently consider prophylactic surgery.
15:49It makes sense. It's drastic, but it saves lives when the risk is definitively proven. Exactly. But if a doctor applies that same logic to the patient sitting at 26%. That borders on catastrophic, prophylactic surgery is irreversible.
16:04So what do you do for the 26% patient? Their management should pivot to careful monitoring, you know, regular MRIs, frequent blood work, just keeping a very close eye on things. Surveillance gives you the flexibility to adapt.
16:15As mathematically elegant as all of this is, I have to imagine this data isn't flawless. Oh, for sure. The authors are highly transparent about the study's limitations. Recall bias is a major hurdle. Because much of this relied on participant recall.
16:29Right. When doing deep phenotyping, asking someone to remember if their grandmother was 55 or 65 when she got sick, introduces a margin of error. There was also the question of who actually participated in this study.
16:41Ascertainment bias is definitely a factor here. Because so many came from biobanks about 57% or direct to consumer testing, which was 18%. Right. That might not perfectly mirror clinical diagnostic settings.
16:55Exactly. We are looking at a subset of the population that is generally proactive, interested in science, or financially capable of buying DNA kits. We have to keep that in mind before applying these exact percentages to the entire globe.
17:09Even with those limitations, bringing it all together. What is the ultimate lesson for the listener? If you had to distill this entire deep dive down? What is the central insight? The context in which a genetic variant is found completely changes its predictive power, a pathogenic genetic finding from a general screening, is not a definitive diagnosis, but rather a starting point that requires careful clinical and familial investigation to determine true risk.
17:36It is the difference between feeling a tug on your line in the open ocean and instantly assuming it's a trout versus taking the time to reel it in and seeing what you actually caught. You have to do the work to figure out what the finding means.
17:47Exactly. And as these tests become more accessible, That ocean is only gonna get more crowded. What does this mean for the future of population wide genetic screening as it becomes more and more common?
17:58That's the $10000 question, isn't 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 5 star rating.
18:15If 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.
18:25Thanks for listening, and join us next time as we explore more science based by base.
