Genomics, Biotechnology's Oldest Next Big Thing

By Elizabeth Silverman

The economist Edward Fiedler famously said "He who lives by the crystal ball soon learns to eat ground glass," a statement to whose veracity I can readily attest having written widely on the potential of genomics during my tenure on Wall Street. Although much has been written over the last 15 years about the power of genomics to transform medical practice, much has also been written about its failure to produce tangible results. So who's chewing ground glass these days?

First, let's look at how the hottest thing to emerge from the lab since Genentech and Amgen ($AMGN) cloned blockbuster protein drugs morphed into a widely dismissed investment bubble: In the 1990s, genomics companies believed that they could discover newer and better drug targets by identifying the variants of genes that caused disease. The underlying assumption was that the frequency of disease gene variants was high enough in patient populations to make them visible when compared to gene variants in healthy populations. Since the cost of gene sequencing was prohibitive at the time, genomics companies developed a multitude of clever shortcut techniques to identify these deleterious gene sequences. Except that they didn't.

As the cost of gene sequencing plummeted, the resulting research revealed that disease gene variants were too rare to be easily plucked out from the background variation without an enormous application of sequencing power and bioinformatics wizardry, none of which was available to the genomics pioneers. Complicating the situation, it is now believed that genes can act in concert to cause or prevent disease (the most likely explanation of why not all individuals carrying BRCA breast cancer mutations develop breast or ovarian cancer). So rather than finding one needle in the genomic haystack, these companies may have needed to find a bunch. Fortunately, the continuing development of faster, cheaper DNA sequencing instrumentation ushered in a genomics reset: Genomics v.2, if you will.

So where are we today? 

Pharmaceuticals: Applications are running back end to front end rather than the predicted reverse. The first fruits of genomics have not been at the front end--target and drug discovery--so much as at the back end of disease subtyping, clinical trial stratification and the companion diagnostics that make it possible. Look for this work to continue and move beyond cancer, where it has had the biggest impact. Target discovery does also proceed--particularly in cancer where new generations of oncology drugs are being developed based on the specific molecular vulnerabilities of a variety of tumor types. Also look for an increase in allied applications, such as a reexamination of failed compounds, to determine their efficacy in patient subpopulations.

Diagnostics: Prenatal health and cancer applications have had the biggest impact. There has been a big push in the prenatal health area, most notably in noninvasive testing for chromosomal abnormalities. Insurance companies have climbed on board both to reduce the cost of care and to provide a competitive edge in their corporate product offerings. Where the money goes, technology will follow. Keeping a close eye on what insurance companies are reimbursing or planning to reimburse is a good way to avoid ingesting some of that ground glass. Look for a continuing expansion in prenatal testing from chromosomal abnormalities to defects in the genes themselves and also to testing prospective parents for their genetic disease carrier status.

With respect to cancer, diagnoses based on the specific genetic abnormalities in the tumor rather than on traditional pathology and tissue of origin is leading to more effective treatment regimens and to the development of new, targeted compounds--so-called precision medicine--as well as to the potential redeployment of older compounds.

Prognostics: Again, cancer applications lead, with the use of molecular tests to more accurately predict tumor aggressiveness and to drive treatment decisions.

But the commercial news is not uniformly positive, and there are issues that may substantially hinder the speed at which genomics can impact medical care. As with many developments in science, the leap from the lab to the clinic is a chasm that should not be underestimated. 

First, it seems obvious to say that genetic information must be correlated with clinical outcomes. Among other things, this requires a level of accuracy that, while mandatory in clinical labs, is not as urgent in scientific ones. The clinical community needs confidence that results are not due to sequencing errors or to artifacts, both of which can occur.

Next, sequencing and sequence analysis requires computing storage and power to a mammoth degree--a single human genome consumes 3 gigabytes of storage, and complex bioinformatics programs are notorious for their voracious IT appetites. This is one of the many reasons that, at the risk of sounding like a Luddite, old-style sequencing arrays should not be counted out of the clinical equation just yet. Many of the operating parameters of newer sequencing technologies are as much a part of the problem as of the solution. Challenges include the clinical laboratory environment itself, where instrumentation and technology must be forgiving of human error and must generate consistent and reproducible results regardless of the operating technician. The sample preparation protocols must be easy, efficient and not overly time-consuming, and the cost of equipment plus labor must generate a price point that is not prohibitive. Additionally, time to result should match the clinical urgency. It also goes without saying that the tests and protocols must pass regulatory muster and not require frequent updates that could trigger burdensome and costly resubmissions. On the positive side, these tests are not your father's diagnostics, and pricing is consistent with the value of the information obtained, which is to say high.

And finally, on a note that is both practical and philosophical, with respect to the human genome, we do not yet have a handle on what constitutes normal. With a scant 30,000 or so human genomes sequenced among the billions on the planet it is not yet possible to uniformly determine which genetic changes--DNA base substitutions, deletions, copy number variations, etc.--are of actionable clinical significance and which are simply random variations. These issues are just the tip of the genomics iceberg of what we don't know about who we are.

Returning to the question of who is eating the ground glass these days, it would seem that both the genomics enthusiasts and the naysayers have their share to consume. And whereas early predictions of the genomics miracles to come were a bit premature, perhaps it is because genomics is proving to be not so much a revolutionary technology as an evolutionary one.

Elizabeth Silverman covered genomics and wrote extensively about the field as a biotechnology stock analyst at Punk Ziegel and Robertson Stephens. She is the author of Genomics: An Investor's Guide. Silverman has a master's degree in genetics from Columbia University, a BA in biology from the University of Chicago, and an MBA from NYU. She runs a consulting firm, Prism Biomedical Research, with clients in the genomics/genetics field.