Choosing the wrong therapeutic target can cost years of research and investment. Protective human gene variants, multi-omics and population-scale data are helping researchers identify stronger targets earlier and with greater confidence.
Developing a new medicine requires years of research, substantial investment and decisions made long before anyone knows whether a therapeutic hypothesis will ultimately succeed.
For researchers working at the start of that process, choosing the right target is critical. A compelling biological hypothesis does not guarantee that modulating a particular target will benefit patients, while unexpected safety effects may not become apparent until much later in development.
Human genetics offers a way to investigate some of these questions before committing to a therapeutic programme.
At Regeneron Genetics Center® (RGC®), researchers use de-identified population-scale genomic and health data to discover and validate potential drug targets. Luca A. Lotta, MD, PhD, Vice President and Chief Translational Genetics Officer at RGC, leads a translational genetics team with expertise in biology, epidemiology and medicine.
Their aim is to use genetic evidence found in human populations to understand which genes contribute to disease, what happens when their function changes and whether those effects could be reproduced therapeutically.
But what makes genetic evidence strong enough to pursue a drug target? Lotta explores how protective variants can reveal new therapeutic opportunities, why population diversity matters and how combining genetics with proteomics and metabolomics could provide a deeper understanding of disease biology.
Starting with evidence from humans
Human genetics can provide direct evidence linking genetic variation to disease, helping researchers identify and prioritise potential drug targets.
For Lotta, that evidence matters in an industry where failure remains common.
“If you’re starting from human genetics, you’re much more likely to pick the right target and understand efficacy and safety better up front,” he explained.
This evidence can help researchers focus their efforts on targets with a higher likelihood of success.
If you’re starting from human genetics, you’re much more likely to pick the right target and understand efficacy and safety better up front.
Genetic variants can be thought of as naturally occurring changes in human biology. By comparing people who carry particular variants with those who do not, researchers can investigate how altering the function of a gene affects disease risk and other aspects of health.
The approach becomes particularly informative when a variant disrupts the function of a gene and is associated with protection from disease.
Learning from protective mutations
RGC focuses extensively on exome and genome sequencing at scale. One aim is to find rare variants that alter gene function, including predicted loss-of-function variants that effectively “break” one copy of a gene.
When people carrying such variants have a lower risk of disease, the finding can reveal both a potential disease mechanism and a possible therapeutic strategy.
“If you can inactivate that gene with, for example, an inhibitory drug, this may be beneficial in the vast number of people that have both functional copies. Therefore, you can try to block that pathway and try to mimic these protective genetic variants in some way,” said Lotta.
This can provide a relatively direct therapeutic hypothesis. If reduced activity of a gene protects people from disease without producing unacceptable consequences, researchers can investigate whether a drug could reproduce that effect.
RGC teams have used this approach to uncover protective genetic associations across cardiometabolic disease, including GPR75 in obesity and CIDEB in liver disease.
More recently, Lotta highlighted rare mutations in FNIP1 associated with favourable metabolic effects. According to Lotta, these mutations occur in approximately one in every 7,000 people and appear to alter a mechanism involved in limiting calorie burning and metabolism.
Individuals carrying one inactivated copy of the gene had less liver fat, lower glucose levels and a more favourable distribution of body fat. Lotta said they ultimately had “60 percent lower odds of cardiometabolic disease”.
Protective variants are not the only useful genetic evidence. Variants associated with increased disease risk can also reveal important biology, but translating those observations into a drug strategy may be less straightforward. Researchers may need to determine how a pathway should be activated rather than inhibited and whether the effect applies broadly or primarily to people carrying the mutation.
Adding more biology to the genetic signal
Identifying an association between a gene and disease is an important starting point, but researchers also want to understand the biological events connecting the two.
Advances in proteomics and metabolomics are making it possible to add this information at population scale. Technologies that once measured these molecular features in relatively small studies can increasingly be applied across thousands of individuals.
By combining these measurements with genomic and health data, researchers can investigate how a genetic variant affects proteins or metabolites and whether those changes are associated with disease.
“It provides us with a step-by-step understanding of how this gene and then, ultimately, this biological pathway change, and how these changes relate to the risk of disease,” Lotta explained.
This can deepen understanding of a potential therapeutic target and identify biomarkers that could later be used to monitor its pharmacology.
The same datasets can also capture aspects of the environment. Diet, environmental exposures and other factors can influence circulating metabolites, protein levels and gene expression. Studying these alongside genetics can help researchers examine how inherited biology and environmental exposures interact.
“There is a continuous interplay between genes and environment,” said Lotta.
Importantly, genetics is not being used to define an individual’s inevitable disease trajectory. As Lotta explained, “We use genes mostly as a tool to really understand the underlying biology, rather than defining whether somebody has a destiny of developing a disease or not.”
Why population diversity matters
The value of population-scale genetics also depends on who is represented in the data.
Rare variants do not occur at the same frequencies in every population. Historical population differences mean a variant that is extremely uncommon in one part of the world may be more prevalent elsewhere.
Limiting genomic research to a narrow population can therefore mean missing informative genetic variants and, with them, potentially valuable biological discoveries.
“If you limit yourself to only one region of the world and one population, then you’re going to exhaust the experiments of nature that have been happening in that specific population,” Lotta explained. “And you may miss out on the ones that have happened everywhere else in the world.”
If you limit yourself to only one region of the world and one population, then you’re going to exhaust the experiments of nature that have been happening in that specific population.
This is particularly important when studying rare loss-of-function variants. Hundreds of thousands or even millions of participants may be required to generate sufficient statistical evidence across large numbers of genes.
Yet large genomic datasets have historically been concentrated in Europe and the US, creating what Lotta described as a “Eurocentric bias” in genomics.
Expanding representation is therefore important both scientifically and for the ultimate benefit of all human populations. Doing so will require infrastructure, expertise and access to sufficiently detailed health information across different regions.
When is the evidence strong enough?
Even compelling genetic evidence does not automatically make a target suitable for drug development.
Researchers still need to consider the unmet medical need, predicted efficacy and safety, existing biological knowledge and where the target is expressed in the body. They also need to determine whether it can be reached using an available therapeutic modality.
“If you have individuals that have mutations in that gene that you’re trying to target, is this associated with a favourable phenotype?” Lotta said. “Does it have any associations with diseases that are undesirable, that could predict potentially a safety concern?”
Practical questions matter too. Researchers need suitable preclinical models, a feasible route into clinical development and the ability to identify patients for eventual trials.
Target selection therefore remains a decision based on multiple lines of evidence rather than a simple genetic threshold. A genetically compelling target may be attractive to one company because it has a suitable therapeutic modality, while remaining difficult for another to pursue.
Even after those questions have been answered, uncertainty cannot be eliminated.
Researchers eventually have to make what Lotta described as “calculated bets” about which programmes justify further investment.
Human genetics can make those bets better informed, but it cannot remove the inherent risk from drug discovery.
Finding the variants we cannot yet see
The next major advance may come from increasing the scale at which these experiments of nature can be found.
Large cohorts combining whole-exome or whole-genome sequencing with de-identified health information, proteomics and metabolomics have already generated important discoveries. But Lotta believes much more remains hidden in human populations.
“We’ve only scratched the surface,” he said.
Many loss-of-function variants are extremely rare. Current datasets can identify stronger associations involving variants that occur often enough to produce a detectable statistical signal, but potentially important variants may exist in only a tiny number of people.
The problem becomes even greater for rare diseases, where both the genetic variant and the condition itself may occur infrequently.
Lotta believes datasets will therefore need to move from hundreds of thousands of participants into the millions and potentially tens of millions.
“If we were able to do that, this could be transformative,” he said.
Greater scale could expose genetic signals that are currently statistically invisible, revealing new disease mechanisms and potential therapeutic targets. Proteomics, metabolomics and other emerging approaches could add further biological context to those discoveries.
The challenge will be building datasets large and diverse enough to find these rare signals.
For drug discovery, the aim is to identify genes that influence disease, understand what happens when their function changes and assess whether those effects could be reproduced with a therapy.
Genetics cannot remove the uncertainty from drug discovery, but it can provide stronger evidence for deciding which targets are worth pursuing.
Explore the research behind these discoveries:
No comments yet