December 2, 2025 | Increasing use of genetic testing to identify people at risk for various diseases has led to a skyrocketing number of never-before-seen variants with uncertain connections to those health conditions, making them impractical to investigate one by one as they once were. Clever bench scientists therefore began making all possible variants in genes with a suspected link to a disease or a patient's symptoms about 15 years ago, to determine which ones are most likely pathogenic, according to Dan Roden, M.D., senior vice president for personalized medicine at Vanderbilt University Medical Center.
The multi-institutional CardioVar Consortium, which Roden co-leads, is heavily focused on Mendelian heart diseases like familial hypercholesterolemia (FH) and hypertrophic cardiomyopathy (HCM) that can be caused by a single change in the 3 billion base pairs in the genome. One key effort recently was mapping the variant effects for the low-density lipoprotein receptor (LDLR) gene commonly mutated in FH, strongly associated with the development of premature atherosclerotic heart disease. That functional landscape was detailed in research that was published recently in Science (DOI: 10.1126/science.ady7186).
Their mission was to determine the disease impact of the roughly 17,000 variants of the 860-amino-acids-long LDLR gene, he says, a painstaking exercise requiring high-throughput cellular assays and advanced computational techniques. The paper presents a color-coded variant effect map for each of the 860 amino acid positions, indicating the likelihood that any amino change at each of those positions is pathogenic.
Further, functional scores were linked with hyperlipidemia phenotypes sourced from human cohorts in the UK Biobank and All of Research program and augmented with polygenic risk scores to highlight the diagnostic value of the tool, says Roden. Patients having a variant that is normal have a low chance of having high levels of LDL, but if they carry a variant that is pathogenic, their chances of having a high LDL are three to four times higher, he adds. If those with a pathogenic variant also have a high polygenic risk score, the probability is eightfold higher.
LDLR is but one of approximately 20,000 genes in the genome, and many others can also influence cholesterol, either a lot or just a bit, he continues. “We know from the last decade and a half of work in genome science that we can identify literally dozens, often hundreds, of genes where variants produce a tiny effect on things like LDL cholesterol.”
Individuals may be unconcerned with a variant that increases their LDL by 1 mg/dL, but tiny effects add up across dozens or hundreds of places in the genome, says Roden in explaining the rationale behind polygenic risk scores. People having that sort of polygenic makeup are at “just as high risk as someone carrying a single, protein-disrupting variant in the LDLR.”
There are literally thousands of Mendelian diseases, many of them rare, as well as high-penetrance gene mutations significantly increasing a person's likelihood of developing specific diseases, says Roden. Mutations in the BRCA1 and BRCA2 genes, for example, famously elevate the risk of breast and ovarian cancers. Likewise, mutations associated with HCM are commonly found in the MYBPC3 and MYH7 genes.
But multiple other genes can significantly raise the risk of developing cancer or a heart attack, especially when combined with lifestyle and environmental factors, he says. “There are people at risk for heart attack because they don’t exercise enough, eat too much fatty meat, have diabetes ... [or] high blood pressure.”
In the 2010s, people who had a heart attack at 35 and had highly elevated LDL were often sent to a specialty clinic where a genetic test would be ordered costing thousands of dollars to sequence for variants in the three major culprit genes, Roden continues. With the advent of much-cheaper technologies, the test-takers now also include people interested in learning if they’re at risk of an early heart attack. The number of variants of unknown significance (VUS) has thus ballooned beyond scientific capacity to decipher which ones increase the risk of disease, and to what degree.
In a review article published a few months ago in Nature Reviews Cardiology (DOI: 10.1038/s41569-025-01201-7), Roden and his colleagues describe how multiplexed assays of variant effects can enable the functional assessment of nearly all variants in a target sequence to identify their functional significance ahead of their discovery in patients. It highlights the need for studies, like the latest one on the LDLR gene, to determine the pathogenicity of the now-massive number of VUS.
As revealed in the review article, for the 50 genes associated with Mendelian cardiovascular disease, the number of “missense” variants—the type where a single change in the DNA sequence causes one amino acid to be substituted for a different one in the resulting protein—79% were VUS and 12% had
conflicting interpretations. In the UK Biobank and All of Us cohorts, more than 75% of all detected missense variants in three key genes linked to cardiovascular disease (KCNQ1, LDLR, and MYH7) are either VUS, have conflicting annotations in the ClinVar database, or are absent from the archive.
Generating gene variants, for the purpose of studying their disease potential, is the easy part, says Roden. “That’s sort of the blood and guts of basic bench laboratory work.”
The complicated work was coming up with the high-throughput methods to screen a large library of variants simultaneously. In a nutshell, the technique employed involves putting the LDLR gene into a circular piece of DNA called a plasmid and controlling where it will be expressed using molecular biology tools, he says. It’s quite simple to change one of the four nucleotide bases—adenine (A), thymine (T), cytosine (C), and guanine (G)—at a particular place in the targeted sequence to make every conceivable mutant.
Researchers set up a system where each of the many millions of these circular molecules goes into a cell to express that one version of LDLR. That cell can then be studied in a big pool with all the other possible gene variants, says Roden. The resulting variant effect map indicates the likelihood that any variant is likely to cause disease or be normal “at any given position with any given amino acid.”
Variant effect maps have been created for genes implicated in numerous diseases, and, as the latest LDLR experiment suggests, these are important starting points for understanding pathogenicity, Roden says. “If a variant really, really looks like it disrupts protein function ... then the likelihood that a patient who carries that variant has the disease is drastically increased.” That individual is at high risk of having a heart attack before age 40—and the risk is “multiplied drastically” if the person also smokes and has a bad lifestyle or hypertension or a high polygenic risk score.
But Roden is careful not to say that a patient who carries a pathogenic LDLR variant will absolutely have a heart attack or tell someone at low risk that they will not, since many other factors go into determining that eventuality, he adds. “That’s going beyond the data.”
Even so, a variant that was wildly misbehaving in a test tube in cells correlated with very high LDL levels in individuals carrying those variants in the UK Biobank and All of Us cohorts but “don’t yet carry the label of familial hypercholesterolemia,” he notes. “So, there are undiagnosed cases.”
Roden is particularly intrigued with the clinical implications for individuals who have both a pathogenic LDLR variant and a high polygenic risk score, who face the greatest risk of an early heart attack. Current polygenic risk scores are imperfect because a lot of it depends on the ancestral background of the person being studied, and part of it depends on how many genetic variants you assay, and ... statistical curves” used in interpreting an individual's genetic risk relative to the general population, he says. But those statistical calculations, on their own, are generally considered a reliable way of estimating a person's likelihood of developing heart disease based on their overall genetic makeup.
There is significant crossover of disease risk between high cholesterol, type 2 diabetes, and hypertension based on shared genetic factors. “The worst combinations will be people who are at high risk for [all three] ... largely independent of each other.”
There remain many schools of thought on the role of genetics in the preventive treatment of heart attack, says Roden. At one end of the spectrum are those who dismiss genetics altogether and advocate for just measuring LDL cholesterol and, if it’s high, giving them statins and, if necessary, increasing the starting dose or switching to another drug option.
Another school of thought, he says, holds that the longer that high levels of LDL cholesterol persist, the more likely a heart attack is going to happen, so they focus on early detection and treatment of people at high risk with statins. They also encourage everyone with FH to have their family members screened and affected family members started on cholesterol-lowering treatment. “Instead of having a heart attack when they’re in their 40s, they’ll have their heart attack much, much later, if at all.”
The other frequently heard argument is that the people with an “incredibly high” LDL cholesterol of 200 need to be treated and their genetics, and that of family members, should be tested to know whether there is or isn’t a genetic basis for the condition, says Roden. People with a very low LDL cholesterol reading of 50, on the other hand, are at such low risk that genetic testing can be skipped.
Then there are the many individuals “in the middle,” with an LDL cholesterol level of, say, 105. “We now have medicines that are very well tolerated [statins] and can drive that LDL cholesterol down and we think reduce risk,” Roden says. “People argue in the gray zone if it’s worth pushing one way or the other ... genetics may or may not help there, because [some believe] ... if you’re genetically preprogrammed to have high cholesterol that’s a more pressing indication than if you just happen to have your cholesterol measured after a fatty meal.”
The tiebreaker for those folks regarding whether to be more aggressive with the statins is variably genetics or their coronary calcium score, says Roden, noting that he works primarily with people having arrhythmias. That said, he is a “big believer in an early start to statins and long-term statin therapy” to help curb the heart attack epidemic that can be attributed to modern lifestyles.
Roden says he foresees an era where people, at the point of transitioning out of the care of a pediatrician, get genetic testing done as part of their first adult wellness check. “There are of course cases of 8-year-olds who need to start statins because they have familial hypercholesterolemia, and the way to find those people is by testing their parents [and] using polygenic risk scores” that are easily and inexpensively obtained and cover dozens of different diseases.
Whether people pay attention to their elevated disease risk is another matter. There is now a modest amount of evidence showing that giving people their polygenic risk score changes behavior for the better. On the other hand, he adds, there is often also an inexplicable reluctance to take statin drugs despite their well-established safety and effectiveness.