Beyond PCSK9: the real value of genomics is ushering in precision medicine at scale

Actio Biosciences founders argue for a refocusing of population-scale genomics efforts, with new goals and a bigger vision

Human genetics may never deliver the treasure trove of targets for blockbuster medicines once hoped for, but there are other, barely tapped applications of population-scale genome profiling that could be transformational. Genetic stratification in routine care and in clinical trials would permit more effective targeted care and more successful clinical trials. Nationwide sequencing would go a long way toward making this vision a reality.

The waves of excitement that have swept through our industry over the last three decades based on the promise of genomics to identify targets that lead to blockbuster medicines have left us with precious few blockbusters and few exciting prospects.

There are sound evolutionary reasons for this, so it’s time to stop expecting genomics to deliver a bonanza of targets that support indication-wide medicines for common diseases. Instead, genomics should be seen as one of the key tools enabling precision medicine — a paradigm that stratifies patient populations into subgroups most likely to benefit from modulation of specific targets. 

That means there needs to be a refocusing of genomics efforts. At the top of priority lists should be initiatives to provide universal access to sequence data. To be most effective these sequence data will need to be seamlessly integrated into patient health records. The hurdles that such initiatives must overcome are likely high. The benefits, however, would be well worth the effort, with the potential to usher in precision medicine at scale by offering a step-change in our ability to match therapies to individual patients’ disease biology, both in clinical trials and clinical care.

The limits of genomics-led target discovery

After thirty years of mostly failed attempts, the fundamental tenets of evolution may be telling us that genomics is the wrong place to expect to find large numbers of drug targets for broad populations of prevalent diseases. One of the key reasons is almost certainly the simple reality that most common diseases are in fact collections of very distinct underlying conditions grouped together on purely clinical criteria.

 While the buzz started even before the draft sequence of the human genome was delivered in 2001, the modern incarnation of this genomic exuberance is best exemplified by PCSK9. Activating mutations had been shown to cause hypercholesterolemia, but it was the identification of loss-of-function mutations that associated with reduced cholesterol and protected against cardiovascular disease that super-charged interest.

Approved inhibitors of PCKS9 have since helped lower cholesterol levels in patients refractory to other treatments. And there are a few other therapies that can be considered both genetics-led and broadly applicable (that is, helpful well beyond a genetic condition caused by mutations in the targeted protein), including sclerostin inhibition for osteoporosis. But other genetics-led targets have proven challenging to translate into broadly useful therapies, including famously SCN9A and ANGPTL3 for pain indications and hypertriglyceridemia, respectively.

More concerning is that despite the availability of hundreds of thousands of human genomes and exomes with linked phenotypic information, we have only a smattering of new associations that look PCSK9-like. In a recent analysis of UK Biobank data, most of the associations reported were for genes already known to cause rare familial diseases.

Despite the many billions of dollars spent in countless genomics projects from companies and countries, it seems fair to conclude that the hoped-for genomic bonanza has not materialized.

There appear to be sound evolutionary reasons why the field has not discovered PCSK9-like protective effects in many more human genes. If a mutation in a human gene leads to a generally beneficial effect without serious adverse effects — as appears to be the case for PCSK9 loss-of-function mutations — that variant would be expected to sweep the human population and become the dominant or only form. The gene in question then would no longer be polymorphic for that type of variant. In consequence, these kinds of universally beneficial variants pointing towards broadly useful drug targets should be as exceptional as they now appear to be.

If that is true, and simply sequencing a whole bunch of people will not deliver us targets that can be beneficially modulated in the general population, then why generate population-scale genome information?

The compelling reasons to do so are quite different from the search for PCSK9-like targets. Work initiated by GlaxoSmithKline plc (LSE:GSK; NYSE:GSK), for example, has clearly shown the relevance of genetic support to successful clinical trials. GSK found through retrospective review that the progress of targets through pipelines strongly correlates with genetic evidence connecting the target to relevant phenotypes. It concluded that targets with genetic support are twice as likely to succeed as ones without.

The aim, therefore, should be to define goals that match genomics efforts to the real deliverables. That existing projects often do not have clearly identified goals is perhaps most simply illustrated by how often projects have a stated numerical goal, such as a number of genomes they intend to sequence.

Genomic sequence data is one of the most powerful tools we have for targeting treatment to disease biology. The fundamental goal of genomic sequencing should be molecular stratification of disease to directly facilitate precision medicines, not finding the next broadly used blockbuster medicine.

Population-scale genomics to support precision medicine

With genomics making limited contributions to the identification of targets of wide utility, it is time to think hard about how to make effective use of genomic information. There are two intertwined priorities for genomics that existing efforts are poorly suited to address: incorporating genetic causes of disease into routine clinical care and into stratified clinical trials. These priorities recognize that precision medicine is, above all, about tailoring treatments to precise underlying causes of disease in individual patients.

 Outside of cancer, however, such targeting is rarely performed in care or trials. Yet in nearly all therapeutic areas, there are already clear indications of the importance of genetic stratification.

Work we performed at Columbia University in partnership with AstraZeneca plc (LSE:AZN; NASDAQ:AZN), for example, showed that 3% of patients with heart failure and 10% of adult patients with chronic kidney disease (CKD) have single gene causes of disease. The percent of patients with strongly genetic forms of disease can be even higher in some neurological diseases, such as some forms of epilepsy.

Moreover, these Mendelian forms of what are generally considered complex diseases often escape clinical notice. And most clinical trials in common diseases continue to be run without reference to underlying genetics, despite the remarkable progress in their genetic dissection.

In these and other therapeutic areas, high locus heterogeneity is the rule, and different causal genes often represent distinct biological drivers of disease, often with a profound consequence for drug response. Moreover, the underlying cause of disease can not only inform about what treatments work best, but also about what treatments to avoid as ineffective, such as steroids in monogenic forms of nephrotic syndrome. 

It is time for such information to be systematically incorporated into both clinical trials and the delivery of care.

A genomic underpinning for precision medicine

If the utility of genomics in drug discovery is more about targeting medicines to those likely to benefit, as opposed to uncovering broadly useful new targets, many existing genome projects are largely not fit-for-purpose. Instead, the field needs global genomic data that are fully integrated with patient health records.

There are logistical, financial and ethical challenges to generating and making available such data. The biggest one, however, may be that with the generation of large-scale genomic data, it is inevitable that expensive and often unnecessary clinical interventions will be triggered.

Overcoming these challenges will not be trivial. But the return to drug development and clinical care would be profound.

Consider a scenario where whole-exome sequence data were generated for all U.S. residents and these data were available alongside clinical record data wherever patients received their care. The example of CKD provides an estimated 37 million affected people in the U.S. A conservative estimate, taking account of the diagnostic yield in different forms of CKD, suggests that upon sequencing, at least 1 million of these individuals would be determined to have mutations in one of the more than 600 genes that cause Mendelian diseases that include kidney dysfunction. It then becomes immediately possible to run large trials in CKD that are stratified by the major genetic forms of the disease. Given the kind of numbers involved, it even becomes straightforward to rapidly run trials that focus entirely on single genetic forms of kidney disease.

Beyond clinical trials, the generation of genome-wide data would enable the clinical and research communities to explore correlations between genetics and both prognosis and treatment responses.

This pairing of genomic data with health records on a massive scale would allow human genetic research to move beyond the predisposition studies that predominate today, permitting powered investigation of genetic modifiers of disease course.

This may prove the most valuable path to new, useful targets, and even to saving drugs that are efficacious only in genetically defined subgroups of patients. This path does not necessarily lead to drugs that are universally useful in all patients with a given disease, but rather ones that are relevant in specific subsets. But that is, after all, what precision medicine is really about.

In conclusion, the genomics advocates that say large-scale genomic studies are critical to drug development are right, but the vision needs to be broader. These studies need to be performed on a much larger scale than current initiatives and their results need to be integrated into clinical care. Such an approach would finally create the long-awaited data commons that is required for genomics to realize its potential.

David Goldstein and John McHutchison are the scientific co-founders of Actio Biosciences Inc., and Goldstein is the company’s CEO.

Signed commentaries do not necessarily reflect the views of BioCentury.