Genetic Test Scores Predicting Intelligence Are Not the New Eugenics
"A world where people are slotted according to their inborn ability – well, that is Gattaca. That is eugenics."
This was the assessment of Dr. Catherine Bliss, a sociologist who wrote a new book on social science genetics, when asked by MIT Technology Review about polygenic scores that can predict a person's intelligence or performance in school. Like a credit score, a polygenic score is statistical tool that combines a lot of information about a person's genome into a single number. Fears about using polygenic scores for genetic discrimination are understandable, given this country's ugly history of using the science of heredity to justify atrocities like forcible sterilization. But polygenic scores are not the new eugenics. And, rushing to discuss polygenic scores in dystopian terms only contributes to widespread public misunderstanding about genetics.
Can we start genotyping toddlers to identify the budding geniuses among them? The short answer is no.
Let's begin with some background on how polygenic scores are developed. In a genome wide-association study, researchers conduct millions of statistical tests to identify small differences in people's DNA sequence that are correlated with differences in a target outcome (beyond what can attributed to chance or ancestry differences). Successful studies of this sort require enormous sample sizes, but companies like 23andMe are now contributing genetic data from their consumers to research studies, and national biorepositories like U.K. Biobank have put genetic information from hundreds of thousands of people online. When applied to studying blood lipids or myopia, this kind of study strikes people as a straightforward and uncontroversial scientific tool. But it can also be conducted for cognitive and behavioral outcomes, like how many years of school a person has completed. When researchers have finished a genome-wide association study, they are left with a dataset with millions of rows (one for each genetic variant analyzed) and one column with the correlations between each variant and the outcome being studied.
The trick to polygenic scoring is to use these results and apply them to people who weren't participants in the original study. Measure the genes of a new person, weight each one of her millions of genetic variants by its correlation with educational attainment from a genome-wide association study, and then simply add everything up into a single number. Voila! -- you've created a polygenic score for educational attainment. On its face, the idea of "scoring" a person's genotype does immediately suggest Gattaca-type applications. Can we now start screening embryos for their "inborn ability," as Bliss called it? Can we start genotyping toddlers to identify the budding geniuses among them?
The short answer is no. Here are four reasons why dystopian projections about polygenic scores are out of touch with the current science:
The phrase "DNA tests for IQ" makes for an attention-grabbing headline, but it's scientifically meaningless.
First, a polygenic score currently predicts the life outcomes of an individual child with a great deal of uncertainty. The amount of uncertainty around polygenic predictions will decrease in the future, as genetic discovery samples get bigger and genetic studies include more of the variation in the genome, including rare variants that are particular to a few families. But for now, knowing a child's polygenic score predicts his ultimate educational attainment about as well as knowing his family's income, and slightly worse than knowing how far his mother went in school. These pieces of information are also readily available about children before they are born, but no one is writing breathless think-pieces about the dystopian outcomes that will result from knowing whether a pregnant woman graduated from college.
Second, using polygenic scoring for embryo selection requires parents to create embryos using reproductive technology, rather than conceiving them by having sex. The prediction that many women will endure medically-unnecessary IVF, in order to select the embryo with the highest polygenic score, glosses over the invasiveness, indignity, pain, and heartbreak that these hormonal and surgical procedures can entail.
Third, and counterintuitively, a polygenic score might be using DNA to measure aspects of the child's environment. Remember, a child inherits her DNA from her parents, who typically also shape the environment she grows up in. And, children's environments respond to their unique personalities and temperaments. One Icelandic study found that parents' polygenic scores predicted their children's educational attainment, even if the score was constructed using only the half of the parental genome that the child didn't inherit. For example, imagine mom has genetic variant X that makes her more likely to smoke during her pregnancy. Prenatal exposure to nicotine, in turn, affects the child's neurodevelopment, leading to behavior problems in school. The school responds to his behavioral problems with suspension, causing him to miss out on instructional content. A genome-wide association study will collapse this long and winding causal path into a simple correlation -- "genetic variant X is correlated with academic achievement." But, a child's polygenic score, which includes variant X, will partly reflect his likelihood of being exposed to adverse prenatal and school environments.
Finally, the phrase "DNA tests for IQ" makes for an attention-grabbing headline, but it's scientifically meaningless. As I've written previously, it makes sense to talk about a bacterial test for strep throat, because strep throat is a medical condition defined as having streptococcal bacteria growing in the back of your throat. If your strep test is positive, you have strep throat, no matter how serious your symptoms are. But a polygenic score is not a test "for" IQ, because intelligence is not defined at the level of someone's DNA. It doesn't matter how high your polygenic score is, if you can't reason abstractly or learn from experience. Equating your intelligence, a cognitive capacity that is tested behaviorally, with your polygenic score, a number that is a weighted sum of genetic variants discovered to be statistically associated with educational attainment in a hypothesis-free data mining exercise, is misleading about what intelligence is and is not.
The task for many scientists like me, who are interested in understanding why some children do better in school than other children, is to disentangle correlations from causation.
So, if we're not going to build a Gattaca-style genetic hierarchy, what are polygenic scores good for? They are not useless. In fact, they give scientists a valuable new tool for studying how to improve children's lives. The task for many scientists like me, who are interested in understanding why some children do better in school than other children, is to disentangle correlations from causation. The best way to do that is to run an experiment where children are randomized to environments, but often a true experiment is unethical or impractical. You can't randomize children to be born to a teenage mother or to go to school with inexperienced teachers. By statistically controlling for some of the relevant genetic differences between people using a polygenic score, scientists are better able to identify potential environmental causes of differences in children's life outcomes. As we have seen with other methods from genetics, like twin studies, understanding genes illuminates the environment.
Research that examines genetics in relation to social inequality, such as differences in higher education outcomes, will obviously remind people of the horrors of the eugenics movement. Wariness regarding how genetic science will be applied is certainly warranted. But, polygenic scores are not pure measures of "inborn ability," and genome-wide association studies of human intelligence and educational attainment are not inevitably ushering in a new eugenics age.
A new type of cancer therapy is shrinking deadly brain tumors with just one treatment
Few cancers are deadlier than glioblastomas—aggressive and lethal tumors that originate in the brain or spinal cord. Five years after diagnosis, less than five percent of glioblastoma patients are still alive—and more often, glioblastoma patients live just 14 months on average after receiving a diagnosis.
But an ongoing clinical trial at Mass General Cancer Center is giving new hope to glioblastoma patients and their families. The trial, called INCIPIENT, is meant to evaluate the effects of a special type of immune cell, called CAR-T cells, on patients with recurrent glioblastoma.
How CAR-T cell therapy works
CAR-T cell therapy is a type of cancer treatment called immunotherapy, where doctors modify a patient’s own immune system specifically to find and destroy cancer cells. In CAR-T cell therapy, doctors extract the patient’s T-cells, which are immune system cells that help fight off disease—particularly cancer. These T-cells are harvested from the patient and then genetically modified in a lab to produce proteins on their surface called chimeric antigen receptors (thus becoming CAR-T cells), which makes them able to bind to a specific protein on the patient’s cancer cells. Once modified, these CAR-T cells are grown in the lab for several weeks so that they can multiply into an army of millions. When enough cells have been grown, these super-charged T-cells are infused back into the patient where they can then seek out cancer cells, bind to them, and destroy them. CAR-T cell therapies have been approved by the US Food and Drug Administration (FDA) to treat certain types of lymphomas and leukemias, as well as multiple myeloma, but haven’t been approved to treat glioblastomas—yet.
CAR-T cell therapies don’t always work against solid tumors, such as glioblastomas. Because solid tumors contain different kinds of cancer cells, some cells can evade the immune system’s detection even after CAR-T cell therapy, according to a press release from Massachusetts General Hospital. For the INCIPIENT trial, researchers modified the CAR-T cells even further in hopes of making them more effective against solid tumors. These second-generation CAR-T cells (called CARv3-TEAM-E T cells) contain special antibodies that attack EFGR, a protein expressed in the majority of glioblastoma tumors. Unlike other CAR-T cell therapies, these particular CAR-T cells were designed to be directly injected into the patient’s brain.
The INCIPIENT trial results
The INCIPIENT trial involved three patients who were enrolled in the study between March and July 2023. All three patients—a 72-year-old man, a 74-year-old man, and a 57-year-old woman—were treated with chemo and radiation and enrolled in the trial with CAR-T cells after their glioblastoma tumors came back.
The results, which were published earlier this year in the New England Journal of Medicine (NEJM), were called “rapid” and “dramatic” by doctors involved in the trial. After just a single infusion of the CAR-T cells, each patient experienced a significant reduction in their tumor sizes. Just two days after receiving the infusion, the glioblastoma tumor of the 72-year-old man decreased by nearly twenty percent. Just two months later the tumor had shrunk by an astonishing 60 percent, and the change was maintained for more than six months. The most dramatic result was in the 57-year-old female patient, whose tumor shrank nearly completely after just one infusion of the CAR-T cells.
The results of the INCIPIENT trial were unexpected and astonishing—but unfortunately, they were also temporary. For all three patients, the tumors eventually began to grow back regardless of the CAR-T cell infusions. According to the press release from MGH, the medical team is now considering treating each patient with multiple infusions or prefacing each treatment with chemotherapy to prolong the response.
While there is still “more to do,” says co-author of the study neuro-oncologist Dr. Elizabeth Gerstner, the results are still promising. If nothing else, these second-generation CAR-T cell infusions may someday be able to give patients more time than traditional treatments would allow.
“These results are exciting but they are also just the beginning,” says Dr. Marcela Maus, a doctor and professor of medicine at Mass General who was involved in the clinical trial. “They tell us that we are on the right track in pursuing a therapy that has the potential to change the outlook for this intractable disease.”
Since the early 2000s, AI systems have eliminated more than 1.7 million jobs, and that number will only increase as AI improves. Some research estimates that by 2025, AI will eliminate more than 85 million jobs.
But for all the talk about job security, AI is also proving to be a powerful tool in healthcare—specifically, cancer detection. One recently published study has shown that, remarkably, artificial intelligence was able to detect 20 percent more cancers in imaging scans than radiologists alone.
Published in The Lancet Oncology, the study analyzed the scans of 80,000 Swedish women with a moderate hereditary risk of breast cancer who had undergone a mammogram between April 2021 and July 2022. Half of these scans were read by AI and then a radiologist to double-check the findings. The second group of scans was read by two researchers without the help of AI. (Currently, the standard of care across Europe is to have two radiologists analyze a scan before diagnosing a patient with breast cancer.)
The study showed that the AI group detected cancer in 6 out of every 1,000 scans, while the radiologists detected cancer in 5 per 1,000 scans. In other words, AI found 20 percent more cancers than the highly-trained radiologists.
Scientists have been using MRI images (like the ones pictured here) to train artificial intelligence to detect cancers earlier and with more accuracy. Here, MIT's AI system, MIRAI, looks for patterns in a patient's mammograms to detect breast cancer earlier than ever before. news.mit.edu
But even though the AI was better able to pinpoint cancer on an image, it doesn’t mean radiologists will soon be out of a job. Dr. Laura Heacock, a breast radiologist at NYU, said in an interview with CNN that radiologists do much more than simply screening mammograms, and that even well-trained technology can make errors. “These tools work best when paired with highly-trained radiologists who make the final call on your mammogram. Think of it as a tool like a stethoscope for a cardiologist.”
AI is still an emerging technology, but more and more doctors are using them to detect different cancers. For example, researchers at MIT have developed a program called MIRAI, which looks at patterns in patient mammograms across a series of scans and uses an algorithm to model a patient's risk of developing breast cancer over time. The program was "trained" with more than 200,000 breast imaging scans from Massachusetts General Hospital and has been tested on over 100,000 women in different hospitals across the world. According to MIT, MIRAI "has been shown to be more accurate in predicting the risk for developing breast cancer in the short term (over a 3-year period) compared to traditional tools." It has also been able to detect breast cancer up to five years before a patient receives a diagnosis.
The challenges for cancer-detecting AI tools now is not just accuracy. AI tools are also being challenged to perform consistently well across different ages, races, and breast density profiles, particularly given the increased risks that different women face. For example, Black women are 42 percent more likely than white women to die from breast cancer, despite having nearly the same rates of breast cancer as white women. Recently, an FDA-approved AI device for screening breast cancer has come under fire for wrongly detecting cancer in Black patients significantly more often than white patients.
As AI technology improves, radiologists will be able to accurately scan a more diverse set of patients at a larger volume than ever before, potentially saving more lives than ever.