A systematic review of 24 full economic evaluations assessing polygenic risk score (PRS)–based clinical strategies across cancer, cardiovascular disease, and other conditions, summarizing methods, cost components, and evidence on cost-effectiveness.
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. Imagine for a 2nd that you're walking into your doctor's office for a routine checkup.
0:12Right, just a normal annual visit. Exactly. They draw a single vial of blood, or maybe just swab your cheek. And from that one simple DNA test, they hand you a highly specific timeline of your future health.
0:26Like a crystal ball, but biological. Yeah. They look at the results and say, you know, you need to start screening for prostate cancer a full decade earlier than the standard guidelines recommend, or they tell you to start taking a preventative heart medication right now today, even though your cholesterol looks completely normal.
0:44I mean, we've been chasing that specific vision of precision medicine for over 2 decades now. The whole idea that we can frontload our healthcare system by catching risk before it ever actually manifests into disease.
0:56And the wild part is the biological tools to do this already exist. We have the technology right now. It's called a polygenic risk store. Right. The PRS. Yeah. Think about how a bank calculates a financial credit score.
1:08They don't just look at one massive financial event, right? They look at 100s of tiny financial behaviors. A slightly late payment 5 years ago, your current credit utilization, the number of open accounts, and they aggregate all of that to calculate your overall financial risk.
1:24It's a cumulative picture. Exactly. A polygetic risk score works on that exact same principle, but for human health. It takes 1000s of tiny genetic variants and combines them to calculate your lifetime risk for specific complex diseases.
1:39But the bottleneck holding this back is no longer the genetics. I mean the science is largely solved. The bottleneck is the economics. Can our heavily burdened resource strapped healthcare systems actually afford to deploy it?
1:51That is the multibillion dollar question. And to explore whether we can actually afford to bring this into the clinic. Today we celebrate the work of Leonardo Maria Siena and the incredible team at Sabienza University of Rome, who have advanced our understanding of the economics behind precision medicine.
2:06Yes, we are doing a deep dive into their 2025 systematic review, which was published in the American Journal of Human Genetics. They decided to look past all the technological hype, and, well, rigorously follow the money.
2:20Because following the money is the only way this gets out of the lab. I mean, over the last decade, the cost of gene sequencing has absolutely plummeted. We went from the human genome project costing 1000000000s to sequencing a whole genome for the price of a decent pair of noise canceling headphones.
2:35It's incredible. Genomic data is suddenly everywhere, but taking that raw data and seamlessly integrating it into routine clinical practice, that's hitting a massive wall of reality. What kind of wall?
2:47Like IT infrastructure. Oh, absolutely. Healthcare systems globally lack the IT infrastructure to securely store and dynamically analyze that volume of data. Plus we are facing a severe personnel shortage.
2:58Ah, like genetic counselors. Exactly. We need professionals to translate these complex, probableistic scores to everyday patients. And above all of that, there is a desperate need to prove cost effectiveness.
3:09Right, because public health systems and insurance models don't just fund cool science. No, they don't. You cannot introduce a new widespread diagnostic tool unless you can mathematically prove that it saves money or significantly improves health outcomes over the long term.
3:25That is the major hurdle for polygenic risk scores. So okay, let's unpack this. By aggregating 1000s of independent single nucleotide polymorphisms. S&Ps, a PRS, quantifies a person's complex disease predisposition.
3:40But are we using these scores to, like, treat sick people or to screen healthy people before they get sick? Well, it's evaluated across the whole spectrum. There's a massive economic difference between using a PRS to screen a completely healthy population of 1000000s just to find a few high-risk individuals versus using it on a small group of people who are already sick to figure out which specific chemotherapy will work best.
4:02Right. The clinical utility changes the math completely. It does. And the Sepienza University team captured that entire spectrum. To figure out where the math actually works, they utilize the prisma methodology to filter the noise.
4:14So they cast a massive net. A huge net. They pulled 2183 records from major scientific databases searching for any study that married polygenic risk with formal economic evaluation. Over 2000 papers. That's a lot of reading.
4:31But they filtered it down to just 24 full economic evaluations. Wait, really? Out of over 2000 Polish papers, hyping up the clinical potential of this technology. Only 24 actually evaluated the economics properly.
4:45Yeah, just 24. That highlights a massive blind spot in translational science. I mean, we pour 1000000000s of dollars of grant money into engineering these incredible genomic tools, but we completely forget to calculate the downstream bill for the hospital trying to use them.
5:00But disparity is striking, you know? However, those 24 papers that made the final cut were exceptionally rigorous. Most of them were cost utility analyses or seaways. Okay, what makes a cost utility analysis so rigorous?
5:13They're sophisticated models that don't just look at raw dollars spent. They weigh the financial cost against a metric called QALY's quality adjusted life year. Ah, QALYs. Let's break down the QAA math for the listener because it's kind of the engine of these economic models.
5:29A QAY assigns a numerical value to a year of life based on the quality of that health. Exactly. So one year in perfect health equals one. 0.0. But a year spent going through severe chemotherapy or managing chronic heart failure might only equal, say, 0.5 or 0.6.
5:47Right. The models are trying to calculate how much money it costs the healthcare system to buy the patient one full perfectly healthy year of life. And were these 24 studies actually good at calculating that?
5:58Very good. The quality check on these studies was highly stringent. The researchers applied the quality of health economic studies checklist, which scores each paper out of 100. 11 of the 24 studies scored in 90 or above.
6:11Wow, so the pool of data is narrow, but the depth and reliability of that data is formidable. Exactly. To make sense of it all, the researchers categorize the studies based on clinical utility into 3 buckets.
6:22Cancer, cardiovascular disease, and other diseases. Let's look at the cancer data 1st because it makes up the bulk of the review, right? 16 of the 24 studies landed here. Yep, covering prostate, colorectal and breast cancer.
6:36In the cancer arena, PRS is fundamentally used defensively. Defensively, like as a tool to optimize existing screening programs. Precisely. Deciding exactly who gets a mammogram or a colonoscopy and when they get it.
6:50Take prostate cancer, for example, which had 6 dedicated studies. And what did they find? Four of those 6 concluded that PRS-based screening is highly cost effective. The standard of care right now relies heavily on the PSA blood test.
7:03But the economic flaw with the PSA test is its notorious rate of false positive. Exactly. A high PSA leads to an expensive invasive prostate biopsy. Often, that biopsy reveals a slow growing cancer that would have never harmed the patient in their lifetime.
7:18Which leads to overdiagnosis and costly unnecessary treatments. But by introducing a PRS as an initial filter, doctors can stratify who actually needs that PSA test and subsequent biopsy. You spare the healthcare system, the cost of unnecessary procedures, and you spare the patient, the physical toll.
7:37That makes total sense. But the economics get complicated fast when you look at breast cancer. And they really do. One study in the review looked at using a PRS to estimate the risk of recurrence in women with early stage breast cancer to guide whether they needed adjuvant chemotherapy.
7:52The math there was brilliant. Because chemotherapy is astronomically expensive and highly toxic. Right. So using a genetic score to confidently telepation, you can safely skip chemo, generates massive cost savings and preserves a huge number of QALYs.
8:06But another study completely flipped the script. It compared PRS-based screening against screening guided by artificial intelligence. Oh, wow. And who won? The economic model found that the PRS screening was actually inferior to the AI guided approach.
8:20That is fascinating from a cost perspective. And the mechanism behind why the AI one is pretty simple when you think about it. An AI model analyzing breast tissue density usually runs on mammograms that the hospital has already taken.
8:34Right, the image exists. The sunk cost is paid. Exactly. The marginal cost to run an algorithm over that image is basically just the electricity for the server compute. A polygenic risk score, on the other hand, requires a brand new physical workflow.
8:49You have to draw blood, ship it to a lab, run it through a sequencer, pay a genetic counselor to interpret it. The AI had a massive economic head start because it leveraged existing infrastructure. That perfectly illustrates why biological elegance does not automatically equate to public health efficiency.
9:06A tool must fit the logistical reality of the clinic. That's a great point. Now, if we shift from defensive screening and cancer to proactive treatment in the 2nd category cardiovascular disease, the economic narrative changes dramatically.
9:21This was the most consistent success story in the entire systematic review, wasn't it? Yes. There are 5 studies focusing on CVD. Here's where it gets really interesting. In cardiovascular disease, the math shifts, because we aren't just screening for the sake of monitoring, we are proactively altering the treatment plan.
9:40Exactly. The PRS is used specifically to refine a patient's eligibility for preventive therapies, most notably, statens. Standard cholesterol meds. So currently, cardiologists rely on standard risk equations like the American heart association guidelines.
9:55Right, which factor in your cholesterol levels, blood pressure, age, smoking status, you know, to decide if you need a statin. Those equations are good, but they are blunt instruments. Very blunt, but by incorporating a coronary artery disease PRS into those existing guidelines, the studies consistently demonstrated high cost effectiveness.
10:14I mean, a myocardial infarcion, a heart attack is catastrophic for the patient and astronomically expensive for the healthcare system. An ER visit, surgery, cardiac rehab, lifelong care. It costs 100s of 1000s of dollars.
10:26While a daily statin is a cheap generic pill. Using a PRS to identify the hidden high risk patients and putting them on a cheap pill to prevent a massive hospital bill. That is the definition of compounding economic return.
10:42The review also looked at a 3rd category. Other diseases, which included type 2 diabetes and primary open angle glaucoma. The diabetes studies are critical, I think, because they highlight the difference between a computer simulation and real life.
10:56Right, like that Finnish study. They didn't just build a spreadsheet. They tracked a massive real world cohort of 313,000 individuals. A sample size of 313,000 moves the data out of the realm of theory and into hard epidemiological evidence.
11:10It definitely does. And they found that PRS-based prevention programs for type 2 diabetes were still highly cost effective, even with the friction of the real world. Which brings us to the obvious tension here.
11:20If the cardiovascular data is consistently positive. And if massive real-world cohorts show it works for diabetes, why is this not standard of care? Why isn't my doctor giving me a PRS test for my heart right now?
11:32To understand the gap between these successful models and current clinical practice, We have to examine the structural limitations of the economic models themselves. Okay, so the finished study was an exception?
11:42A very rare exception. The vast majority of these 24 economic evaluations rely entirely on hypothetical cohorts. Using Markov models or micro simulations, right? Exactly. A Markov model is essentially a mathematical framework.
11:57It simulates a population moving through distinct health states over time. So it models a hypothetical patient transitioning from healthy to diseased, to treated, to deceased, calculating the accrued costs in QEL-wise at every transition.
12:13Yes, but the entire model rests on the transition probabilities, the assumptions, the researchers program into the map. And what's the biggest assumption they make? The most glaring assumption built into almost all of these models is 100% adherence.
12:27The math assumes that every single person offered the genetic test will gladly take it, and that every physician will flawlessly execute the resulting clinical guidelines. But human behavior ruins the perfect math every time.
12:40In reality, behavioral economics plays a massive role. There is significant resistance to genetic testing. Oh, massive resistance. People have deep privacy concerns about handing their DNA over to a massive health system.
12:53They worry about life insurance discrimination. Right. If your Markov model assumes 100% uptake, but the real world acceptance rate is closer to 30%. Your projected cost savings evaporate instantly. Because the fixed costs of setting up the screening program remain, but the volume of prevented disease drops off a cliff.
13:12Exactly. But beyond behavioral resistance, there is a structural data issue that threatens to completely invalidate these economic models for diverse population. Wow, the ancestry blind spot. Yes, across almost all the studies reviewed.
13:25Ethnicity and ancestry were rarely factored into the cost effectiveness equations. Because polygenic risk scores are heavily calibrated on genome wide association studies that overwhelmingly sampled people of European descent.
13:37Exactly. The genetic architecture of complex diseases varies. Allele frequencies differ across populations. A specific cluster of S&Ps that accurately predicts heart disease risk in someone of European descent might not carry the same predictive weight in someone of African, Asian, or indigenous descent.
13:56It might be completely absent, which creates an immediate logistical and economic nightmare. A total nightmare. If a national healthcare system rolls out a PRS screening tool based on European data across a diverse population, the model misfires.
14:12It generates false positives for some groups and misses high risk individuals in others. And not only do you risk widening global health disparities by providing inaccurate medicine to minority populations, but the economic rationale collapses.
14:25You are paying for a diagnostic tool that is statistically broken for a large percentage of your patients. That structural bias is a silent cost. But the research has also pointed out that these models completely ignore the loud, obvious costs of actually putting this into practice.
14:39The implementation costs. They measure the price of the genetic test panel and the price of the statin, but they ignore everything else. Right. Expanding laboratory sequencing capacity to handle 1000000s of routine DNA samples requires immense capital.
14:53Integrating genomic data into legacy electronic health record systems requires custom IT development. You have to train 1000s of general practitioners. You cannot simply hand a family doctor a specialized report detailing 10,000 SMPs and expect them to confidently translate that into a care plan without extensive paid clinical training.
15:15You really can't. But applying these models universally hits an even harder wall when you look at how disease actually spreads across a population. This is Jeffrey Rose's prevention paradox, right? Yes, it's a classic epidemiological concept.
15:27It basically states that the bulk of sick people don't actually come from the high risk pool. So how does a highly targeted genetic test survive that paradox economically? Well, Rose observed that a very large number of people exposed to a small risk will inevitably generate many more cases of a disease than a very small number of people exposed to a high risk.
15:46Let's apply that math to polygenic risk scores. A PRS is phenomenally good at identifying the extreme tail end of the bill curve, right? The top 5% of the population at the highest genetic risk for a disease.
15:58And finding and treating those individuals is highly beneficial for them personally. But the paradox is that hospitals aren't filled exclusively with the top 5%. The sheer volume of heart attacks and cancers comes from the massive middle of the bell curve population, simply because that group is so incredibly large.
16:17Exactly. 1000000s of people at a baseline 10% risk will mathematically produce far more total heart attacks than 10,000 people at a 50% risk. Which forces a difficult conversation about return on investment.
16:29It does. If a public health system spends 100s of 1000000s of dollars implementing a nationwide PRS screening program, it will successfully identify the genetic outliers, but it will inevitably miss the vast majority of disease cases.
16:43Because those cases occur in people with normal genetic risk profiles who develop illnesses due to poor diet, environmental exposures, or sheer biological bad luck. The total disease burden on the healthcare system remains largely unchanged.
16:58Right. And when we calculate that total burden. I think it's important to push back on how these studies define a cost. Oh, absolutely. These cost utility analyses meticulously track lab fees, hospital bed days, and drug prices, but they don't have a spreadsheet column for the psychological toll.
17:14The human element. Exactly. If someone takes a PRS test at age 25 and discovers they have a sky high genetic risk for early onset Alzheimer's, they have to carry that invisible heavyweight around for decades.
17:26Does that severe anxiety not lower their real world QAAY score, even if the model ignores it? The indirect societal and psychological costs are a massive blind spot in the literature. And the behavioral impact swings both ways.
17:39Consider the danger of false reassurance. Oh, that's a huge one. A patient receives a PRS report indicating a fantastically low genetic risk for cardiovascular disease. They misinterpret that as being genetically bulletproof.
17:52Right. They see a low score, skip the gems, start smoking, and eat fast food every day. A low genetic risk does not make you immune to a terrible lifestyle. No, it doesn't. And if the false reassurance of a good score triggers behavioral backsliding that ultimately causes a heart attack, that is a severe negative health outcome directly caused by the test itself, something the micro simulation models are completely blind to.
18:15We are attempting to map highly linear economic math onto highly irrational human behavior. a tough fit. So if we synthesize everything Leonardo Maria Siena and the Sapienza team uncovered in the systematic review.
18:29The central insight is clear. The underlying math of polygenic risk scores works. It does. They show highly promising trends for making healthcare more proactive and cost effective, particularly when used to target statin therapy in cardiovascular disease and refined cancer screening protocols.
18:45But until researchers and policymakers account for the friction of human behavior, The staggering implementation costs, and the absolute necessity of diverse ancestral calibration, their true value remains a brilliant theoretical calculation.
19:00rather than a functional clinical reality. It is like we've engineered a high performance Ferrari engine. Genomic science, but we haven't paved any roads or trained any mechanics, which is the healthcare infrastructure to support it.
19:14What does this mean for the future of your own personalized medical care? 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.
19:27If you enjoyed this, follow or subscribe in your podcast app and leave a 5 star rating. If 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.
19:41Thanks for listening and join us next time as we explore more science, base by base.
