Announcer
The following program features simulated voices generated for educational and philosophical exploration.
Vera Castellanos
Good afternoon. I'm Vera Castellanos.
Ryan Nakamura
And I'm Ryan Nakamura. Welcome to Simulectics Radio.
Vera Castellanos
Yesterday we discussed in vitro gametogenesis with Dr. Katsuhiko Hayashi. Today we examine the next step in reproductive technology—polygenic risk screening of embryos. This technique uses genetic data to predict complex traits and disease risks across multiple genes, potentially transforming embryo selection from single-gene disorder avoidance to optimization across multiple characteristics. Joining us is Dr. Shai Carmi, a computational biologist at Hebrew University whose work focuses on statistical genetics and polygenic risk prediction. Dr. Carmi, welcome.
Dr. Shai Carmi
Thank you. This is a technology that sits at the intersection of genomics, statistics, and reproductive medicine, with implications that extend far beyond the clinic.
Ryan Nakamura
Let's start with the basic science. How do polygenic scores differ from single-gene screening?
Dr. Shai Carmi
Single-gene screening looks for mutations in individual genes that cause Mendelian diseases—conditions like cystic fibrosis or Huntington's where one gene determines outcome. Polygenic scores aggregate effects across thousands or millions of genetic variants, each contributing small amounts to a trait. For complex conditions like diabetes, schizophrenia, or heart disease, no single gene is deterministic. Instead, many variants combine with environmental factors to influence risk. Polygenic scores attempt to quantify this cumulative genetic contribution.
Vera Castellanos
What's the predictive accuracy for complex diseases? Can these scores reliably distinguish high-risk from low-risk embryos?
Dr. Shai Carmi
Accuracy varies considerably by trait and population. For height, we can explain perhaps 40 percent of variance with current scores. For schizophrenia or type 2 diabetes, maybe 10 to 20 percent. This means scores identify relative risk—someone in the top decile might have twice the average risk—but absolute predictions remain uncertain. In embryo selection, we're comparing siblings, which helps because they share familial background. You might identify an embryo with 30 percent higher disease risk than its sibling, but that doesn't tell you absolute probability either will develop the condition.
Ryan Nakamura
So we're talking about relative rankings rather than deterministic predictions.
Dr. Shai Carmi
Exactly. Polygenic embryo selection doesn't eliminate disease risk. It shifts probability distributions. If you select embryos with lower genetic risk across a population, you'd see reduced disease incidence, but many selected embryos will still develop the condition and many rejected ones would have been healthy. The question is whether this probabilistic improvement justifies the intervention.
Vera Castellanos
What traits are currently being screened in clinical practice?
Dr. Shai Carmi
Some fertility clinics offer polygenic screening for a handful of conditions—primarily type 1 diabetes, breast cancer, coronary artery disease, and schizophrenia. These are diseases with reasonably well-established genetic architectures and significant health burdens. The screening is marketed as risk reduction, not prevention. Regulatory oversight varies by jurisdiction—the United States has minimal restrictions, while several European countries prohibit or heavily regulate the practice.
Ryan Nakamura
Could this be extended to non-medical traits like intelligence or personality?
Dr. Shai Carmi
Technically, yes, though with severe limitations. We have polygenic scores for educational attainment, which correlates with intelligence but measures years of schooling—a complex phenotype influenced heavily by environment. Predictions for intelligence itself are very imprecise, perhaps explaining 10 percent of variance. For personality traits, predictive power is even weaker. Even if scores improve, we're discussing probabilistic shifts of a few percentile points. You couldn't design a genius through embryo selection with current or near-future technology.
Vera Castellanos
What are the statistical challenges specific to embryo selection versus population-level prediction?
Dr. Shai Carmi
Several complications arise. First, embryos are siblings, so they share 50 percent of genetic variance on average. This reduces the genetic diversity you're selecting among—most embryos from the same parents will have similar polygenic scores. Second, we're dealing with small sample sizes, typically 5 to 15 embryos, which limits selection differential. Third, scores are optimized for populations, but individual families may have unique genetic architectures. A score predicting diabetes in Europeans might perform poorly in East Asians. Fourth, there's genotype-environment interaction—a genetic risk profile might confer vulnerability only in specific environments we can't predict decades in advance.
Ryan Nakamura
How much improvement could someone realistically expect from selecting among siblings?
Dr. Shai Carmi
Modeling studies suggest modest effects. For height, selecting the tallest embryo among 10 siblings might yield an offspring about one centimeter taller than selecting randomly. For disease risk, you might reduce lifetime disease probability by a few percentage points. These effects compound across traits if you're selecting for multiple characteristics simultaneously, but the gains remain incremental. It's optimization, not transformation.
Vera Castellanos
What happens when people select for multiple traits simultaneously? Do trade-offs emerge?
Dr. Shai Carmi
This is where genetic correlations become important. Many traits are genetically correlated—variants that increase height tend to increase diabetes risk, for example. Intelligence-associated variants correlate with autism and anorexia risk. You can't optimize everything independently. Multi-trait selection requires weighting preferences, which becomes a values question. Should you prioritize low schizophrenia risk over higher educational potential if these trade off? There's no objective answer.
Ryan Nakamura
Could this lead to unforeseen consequences through selecting against traits we don't fully understand?
Dr. Shai Carmi
Possibly. Many genetic variants have pleiotropic effects—they influence multiple traits, some of which we don't measure or understand. Selecting against one characteristic might inadvertently select against beneficial but unmeasured traits. There's also the question of heterozygote advantage, where carrying one copy of a risk variant provides benefits. Sickle cell trait is the classic example—one copy protects against malaria. We might eliminate genetic diversity whose value we don't recognize.
Vera Castellanos
How do population differences in genetic architecture affect score portability?
Dr. Shai Carmi
This is a significant problem. Most genome-wide association studies use European-ancestry participants, so polygenic scores work best in European populations. When applied to African or East Asian populations, accuracy drops substantially—sometimes by 50 percent or more. This creates equity concerns. If polygenic selection becomes widespread but only works reliably for European-ancestry individuals, it exacerbates existing health disparities. We need more diverse genetic studies, but that requires resources and equitable research participation.
Ryan Nakamura
What about the long-term effects if this becomes widespread? Could we see reduced genetic diversity at the population level?
Dr. Shai Carmi
If a substantial fraction of births involved polygenic selection for similar traits, yes, we'd see reduced genetic diversity. This matters because genetic variation enables population adaptation to changing environments. However, the effect size depends on adoption rates and selection intensity. If 10 percent of births use mild selection among siblings, population-level effects are negligible. If 80 percent of births involve aggressive selection for identical traits, genetic diversity could decline meaningfully within several generations. We should monitor this as the technology scales.
Vera Castellanos
What are the psychological impacts on children born through polygenic selection? Does knowing you were selected based on genetic scores affect development?
Dr. Shai Carmi
We don't have empirical data yet, but there are plausible concerns. Being told your parents selected you for intelligence or health might create performance pressure or sense of conditional acceptance. Alternatively, it might be experienced like any other parental investment in child welfare. Much depends on cultural framing and family dynamics. We should study these outcomes as the first generation of polygenic-selected children mature, similar to how we studied IVF children's psychological development.
Ryan Nakamura
Could polygenic selection create social stratification if access is limited to wealthy populations?
Dr. Shai Carmi
This is perhaps the most significant long-term concern. If polygenic selection becomes routine among affluent populations but remains inaccessible to others, and if the gains compound across generations, we could see increasing genetic stratification by social class. This wouldn't be genetic determinism—environment still matters enormously—but it could amplify existing inequalities. Whether this happens depends on access policies, cost trajectories, and effectiveness of interventions. It's a scenario we should actively work to prevent through equitable access frameworks.
Vera Castellanos
How do error rates in genetic sequencing and score calculation affect reliability?
Dr. Shai Carmi
Sequencing errors are now quite rare with modern technologies—maybe one error per 10,000 base pairs. But polygenic scores involve millions of variants, so errors accumulate. More significant are biases in score construction. If training data includes population stratification or batch effects, scores inherit these biases. There's also the problem of overfitting—scores optimized on one dataset may perform worse on new data. We need robust validation frameworks and quality control standards as this transitions from research to clinical application.
Ryan Nakamura
What regulatory frameworks exist for polygenic embryo screening?
Dr. Shai Carmi
Regulation is fragmented. The United States treats it as a laboratory test with minimal oversight—clinics can offer screening without FDA approval. The UK permits it under Human Fertilisation and Embryology Authority guidelines with emphasis on serious disease prevention. Germany effectively prohibits embryo selection beyond single-gene disorders. China's regulations are unclear but evolving. There's no international consensus on acceptable applications or validation requirements. This creates medical tourism opportunities and inconsistent standards.
Vera Castellanos
Should we distinguish between screening for disease risk versus trait enhancement in regulatory frameworks?
Dr. Shai Carmi
Many argue yes—that preventing serious disease is morally distinct from optimizing intelligence or appearance. But the boundary is unclear. Is preventing mild depression disease prevention or enhancement? What about selecting for lower anxiety, which could improve quality of life but isn't a disease? I think clearer distinctions exist between high-penetrance single-gene disorders and polygenic optimization for normal-range traits. The former prevents identifiable serious harm; the latter pushes distributions. Whether that justifies different regulation is a social values question beyond my expertise.
Ryan Nakamura
How do you respond to disability advocates who argue polygenic screening devalues lives affected by screened conditions?
Dr. Shai Carmi
This is a valid concern that deserves serious engagement. Screening decisions reflect societal attitudes about which conditions are worth preventing, which can feel like judgments about the value of existing lives. However, parents choosing to reduce disease risk for their future children aren't necessarily making statements about existing people with those conditions. There's a difference between preventing suffering when possible and devaluing those who experience it. That said, widespread screening does send cultural signals about desirable versus undesirable traits, which can affect social attitudes. This tension exists for any prenatal screening and deserves ongoing ethical attention.
Vera Castellanos
What technical improvements would substantially increase the utility of polygenic selection?
Dr. Shai Carmi
Several advances would help. First, larger and more diverse genome-wide association studies to improve score accuracy across populations. Second, better understanding of gene-environment interactions to contextualize predictions. Third, integrating functional genomics data to understand causal variants rather than just statistical associations. Fourth, methods for combining polygenic scores with family history and environmental data for more personalized predictions. Fifth, validation in prospective studies rather than retrospective modeling. These improvements could double or triple predictive accuracy over coming decades.
Ryan Nakamura
Could polygenic selection be combined with gene editing? Select the best embryo and then edit specific remaining risks?
Dr. Shai Carmi
Technically possible but ethically complex. This would combine two interventions each with distinct risk profiles. Polygenic selection works with natural variation; editing introduces novel changes. The combination enables more aggressive optimization but multiplies uncertainties and ethical concerns. If we're already editing embryos, polygenic selection beforehand seems inefficient—why not edit all identified risks? The scenarios where both make sense are narrow. I think they'll remain separate interventions for distinct purposes.
Vera Castellanos
What's your personal view on appropriate applications of polygenic screening?
Dr. Shai Carmi
I think screening for serious, high-burden diseases with reasonable predictive accuracy is defensible, particularly when no other prevention options exist. I'm more cautious about optimization for normal-range traits where predictions are weak and societal implications unclear. We should prioritize reducing medical suffering over enhancement. That said, I recognize others draw the boundaries differently, and pluralistic societies will accommodate diverse values. My concern is ensuring decisions are informed by realistic understanding of predictive accuracy rather than inflated expectations.
Ryan Nakamura
How long until polygenic selection becomes routine in assisted reproduction?
Dr. Shai Carmi
Adoption depends on cost, accuracy, and cultural acceptance. Sequencing costs continue declining, making screening economically feasible. As predictive accuracy improves and outcomes data accumulate, more people may view it as standard precaution similar to prenatal ultrasound. I'd estimate within 10 to 20 years, polygenic screening could be routine in wealthy countries for couples using IVF. Broader adoption depends on whether IVF itself becomes more common, which technologies like in vitro gametogenesis might enable.
Vera Castellanos
Final question: what should people understand about polygenic screening that's often misrepresented in popular discourse?
Dr. Shai Carmi
That it's fundamentally probabilistic, not deterministic. Polygenic scores don't predict outcomes; they estimate relative risks based on population averages. Individual development depends on countless unmeasured factors. This isn't designer babies—it's nudging probability distributions by modest amounts. People should approach it with appropriate expectations and recognize the significant uncertainty inherent in complex trait prediction. We're not designing children; we're making informed gambles with limited information.
Vera Castellanos
Dr. Carmi, thank you for this rigorous examination of polygenic embryo screening and its limitations.
Dr. Shai Carmi
Thank you. These technologies require informed public discourse about what they can and cannot achieve.
Ryan Nakamura
Tomorrow we'll discuss cellular reprogramming and induced pluripotency with Dr. Shinya Yamanaka.
Vera Castellanos
Until then. Good afternoon.