How Your Polygenic Risk Score Can Predict Disease

One of the most famous principles in classical philosophy (one of our MBAs double-majored in case the whole business degree didn’t work out) is called Occam’s razor. It states that one should not make more assumptions than are needed to explain a certain phenomenon. In other words, start with the simplest answer first. This is the exact opposite of conspiracy theories, which require a vast amount of crazy assumptions to explain reality. Easier to believe a shadowy group like the Illuminati controls world events rather than go with the obvious conclusion that most of us can’t plan a picnic, let alone planetary domination. Conversely, understanding what causes certain diseases such as cancer or even mental illness is incredibly difficult and complex. But an emerging science within genetics called polygenics promises to simplify things by predicting risks for many different conditions, even all the way back to the womb.

What is the Polygenic Risk Score?

The idea behind polygenics is that scientists believe that much of our future – height, health, happiness – is writ in our DNA. It’s just a matter of finding the right combination to unlock the safe. However, where we once thought that only a few genetic markers predicted whether one would get breast cancer or make a career as a safecracker, researchers now think such traits are the result of hundreds if not thousands of genes acting together. Thanks to the dirt-cheap genetics tests on the market, not to mention huge government studies, scientists now have access to the genetic profiles of potentially millions of people. That enables them to determine with greater accuracy who is more at risk for developing type 2 diabetes or riding on the short bus to school.

Who knew life sciences could be so easy to understand?
Who knew life sciences could be so easy to understand? Credit: Journal of Clinical Lipidology

That’s what we’re calling the polygenic risk score. MIT Technology Review even listed this new sort of genetic fortune-telling as one of the top breakthrough technologies in 2018. All the experts out there that we’ve read say the game changer has been those big datasets, which is driving a technological revolution in everything from mental health to how Coca Cola now operates its business. For example, as the number of diabetics enrolled in genetic studies increased exponentially, scientists went from identifying 12 genes that influence type 2 diabetes to at least 400, though each with a tiny effect that was masked in smaller samples sizes. No doubt artificial intelligence or some sophisticated software is helping automate the whole process in some cases.

Potential of the Polygenic Risk Score

There are obvious benefits to this predictive power. For instance, women with higher risk for breast cancer would get more mammogram check-ups over those with a lower polygenic risk score. We already know that the best cure for breast cancer is early detection. Another advantage, as MIT Technology Review notes, is that clinical trials for new drugs that target heart disease, for example, would benefit by selecting participants who are more likely to experience problems with the old ticker as they age. Genetics is also part of the bigger trend of what is called precision medicine or personalized healthcare, with therapies custom-tailored for each person based on their polygenic risk score and other factors.

Pitfalls of the Polygenic Risk Score

It’s important to remember that the Polygenic Risk Score only offers probabilities, not conclusions or diagnoses. Just because you are predisposed to diabetes or even Alzheimer’s disease doesn’t mean you’ll succumb to those conditions. There are obviously other factors involved in health, chiefly environment and lifestyle. If you swim in the river next to a nuclear power plant and eat a Big Mac for lunch every day as a matter of course, even the genes of an Olympian god might not save you from chronic illness.

Ethically, it’s a slippery slope from making DNA predictions about disease to using that information for selecting babies at birth that are more likely to excel academically or athletically, leaving the rest of us to work blue-collar jobs like forklift driver that will disappear when the machines take over. Will insurance companies be able to deny coverage for pre-existing genetic conditions based on a high polygenic risk score?

And, of course, we have to be wary about all of the slimy ways companies may try to commercialize these tests. For instance, we’ve seen a company called Helix release a line of DNA apps that target everything from the best diet options to your genetic wine preferences. There’s even an at-home DNA test from DNA Diagnostic Center that identifies your skin’s genetic potential, as well as a host of companies promising better love connections through DNA matching. More on some of these scammy-type outfits below.

The First Commercial Polygenic Risk Score

The first polygenic test actually came on the market back in 2017 from our old friends at Myriad Genetics (MYGN), a “personalized medicine” company that specializes in genetic diagnostics, particularly in cancer. Dubbed riskScore, the test combines 86 DNA variants with a person’s family and medical history, to determine a woman’s five-year and lifetime risk of breast cancer. The results of a clinical trial last year found that the new test can accurately make these predictions even when the women test negative for a hereditary disease using the company’s myRisk Hereditary Cancer test. One assumes that would also be the case for Myriad’s competitors such as startup Color Genomics, which offers hereditary tests for cancer and high cholesterol.

Polygenic risk score for breast cancer.
Credit: Myriad Genetics

Until recently, Myriad’s tests were only good for those of European descent, because that’s the population where most of the big data are derived. However, the company just announced last month that it has validated polygenic risk score results for Hispanic women based on 14,000 women.

Polygenic Risk Score for Other Diseases

Click for company websiteCancer has been the primary target for polygenic risk scores, but researchers and companies are branching out – though we’re rightfully cautious given the early stages of this technology. For example, a startup out of Rome called Allelica took in an undisclosed amount of funding in a Seed round a year ago. For $50, it will send you a polygenic risk score for heart disease, type 2 diabetes, hypertension, atrial fibrillation, and even Alzheimer’s disease after you upload your 23andMe results. Allelica relies on technology from Illumina (ILMN) to read the DNA test, which it says allows the genotyping of more than 750,000 genetic variants with an accuracy greater than 99.9%. We’d tell you more about how it works, but apparently the Seed round last year didn’t cover costs to update the company’s webpage:

Hopefully, the science is more solid than the website.
Hopefully, the science is more solid than the website. Credit: Allelica

Click for company websiteMeanwhile, New York-based Phosphorous, which raised $10 million in 2016, is expanding into polygenics with its first polygenic risk score for high cholesterol. The three-year-old startup already offers a single predictive test called Phosphorous One that it says covers a range of conditions, including heart disease, cancer, infertility, vision loss, neurodegenerative diseases like Alzheimer’s, and even adverse drug interactions. Its new polygenic test catches those patients who do not carry the gene mutations normally associated with high cholesterol.

Polygenic Risk Scores of In Vitro Fertilization

Click for company websiteYes, at least one company has already gone there: Genomic Prediction, founded in 2017 and with an office in New Jersey, offers parents using in vitro fertilization a way to evaluate genetic risks to their embryos due to high polygenic risk scores, as well as chromosomal abnormalities and single-gene mutations. The company says it uses machine learning to help process the large datasets that it uses to validate polygenic risk scores. In addition to DNA predictions for disease, Genomic Prediction claims its technology can project a person’s height, for example, within a few centimeters. Eenie meenie miney mo, I pick this super tall embryo. The company does boast a list of peer-reviewed research to back up its claims.

Polygenics and Mental Health

Click for company websiteA Swiss startup called Karmagenes has raised about $260,000 to help people improve their lives through a combination of polygenics and psychology. The idea is that Karmagenes will build a personalized profile of its clients using a proprietary algorithm that employs a standard psychological assessment and 14 different behavioral characteristics extracted from a DNA sample. It says it can link multiple genes to each of the behavioral characteristics (though we only count 13 in this diagram):

The wheel of karma.
The wheel of karma. Credit: Karmagenes.

The road to DNA-based self-actualization begins at $95, but you’ll have to shell out at least $199 for the genetic test to reveal your behavioral characteristics.


The science behind polygenic risk scores has the potential to rewrite the playbook on how we approach healthcare in the future. It also has the potential for serious abuse and dubious applications, like the app that purported to tell parents what their baby would look like using machine learning and neural networks to create polygenic risk scores. The $3 million startup behind that one was gobbled up by Helix last year. Still, when respected companies like Veritas Genetics (one of the many startups co-founded by George Church, a leading figure in genetics research today) invest in technologies like AI to boost its own efforts to produce valid polygenic risk scores, investors should take note.

Worried about DNA privacy? You should be. Now, Nebula Genomics allows you to learn about your DNA without giving it away. They even offer anonymous DNA testing.

Leave a Reply

Your email address will not be published. Required fields are marked *

  • This field is for validation purposes and should be left unchanged.