Unlocking the Secrets of Gene Networks

Harnessing the information from a person’s genome and using it to make individual treatment decisions – so-called ‘personalized medicine’ – has become a hot topic in healthcare research, and Eric Schadt knows what it’s like to be at the forefront of the action. Two new studies on which he is senior author have shown how genomic techniques could be used to understand the complex changes at the root of common diseases, and could also provide evidence for genetic susceptibility to obesity.

The studies, carried out by researchers at Merck Research Laboratories (MRL), where Schadt is Executive Scientific Director of Genetics, support the theory that common diseases are the result of genetic and environmental disturbances in entire networks of genes, rather than being limited to mutations in several specific genes.

As Schadt explains, “The studies were intentionally designed to complement the efforts of top genetics researcher Francis Collins, who promoted the carrying out of large-scale, genome-wide association scans. Those scans deliver signposts in the DNA that tell you where something is happening that will increase your risk of disease. The innovative step we took, and it is a new approach to uncovering the causes of disease, was to look at how those variations in DNA affect the molecular network that actually causes disease.”

What the researchers found was surprising: it isn’t one or two or 10 genes that are affected; it is hundreds of genes, and they were all communicating very closely with each other to increase susceptibility. “Rather than disease being the result of a single signpost in the DNA,” says Schadt, “you could think of it as a light switch, where you flip the switch and you get disease. You turn the gene on or you turn it off, and you get disease or you don’t get disease. But it’s not just a single light switch: it’s a whole network of light switches that are communicating with each other in a much more complicated fashion to increase your susceptibility to disease.”

Gene networks

The studies found that there are hundreds of genes operating together in a network, and it is the state of all those genes together that defines the disease. According to Schadt, it was this increase in complexity that was the most striking result of the studies.

He points out that the main challenge of these results is that if hundreds of genes are causing disease in these complex networks, how, from a pharmaceutical perspective, do you target the disease treatment? “What biomarkers do you use to identify the disease state? Once you have these networks that are predictive of disease, you can analyze each of the genes in this network and start to identify what the key genes are, and what the key information nodes are that you can target to maximally affect the network with respect to the disease.

“Some people might think, ‘Oh, it’s so complex, there’s no hope.’ But we would say, ‘It is complex, but the method we use and the networks we come up with enable you to identify the right nodes to target to maximally impact the disease and also the right nodes to target to treat as a biomarker or diagnostic to assess what particular subtype of disease you have, and then match that with a drug that would most benefit you as an individual.’”

Schadt says the main advantage of this approach will be in interpreting all of the genetic data flooding out of these genome-wide studies that are associating changes in DNA with disease. “It will enable us to fill in the big gap between changes in DNA and the onset of disease. Changes in DNA alone don’t drive a disease, they drive changes in the molecular states of the system that go on to drive disease. This approach will be able to fill in that big gap that isn’t filled in by the genome-wide association studies, to get at the network for almost any complex disease that is manifested in the human population.”

Identifying targets

Once the molecular networks that define the disease state have been identified, it then may be possible to target the best genes to treat the disease. The immediate benefit for the pharmaceutical industry will be the identification of the very genes that could have the biggest impact on disease data.

“This stands in contrast to genetic studies, where you are restricted to genes that harbor DNA mutations that associate with disease,” Schadt explains. “They give you a much more constrained, narrow view, whereas our methods cast a broader net, so that you can consider many more of the genes that are involved in disease and then figure out the best way to target those.

“That also leads to diagnostics that are biomarkers for different disease subtypes, so that you can match the right drug to the right person at the right time. We can track these networks and identify what state a given individual has, what state their network is in, and then match a given treatment to those individuals. That will probably be the most immediate benefit. Over the next several years, you’ll see some of that come to fruition. This process will have the potential to be applied to almost any disease.”

New diagnostics

Schadt’s view is that it may be 10 years before drugs that have been developed using these targets are in the marketplace. But over the next two to three years, we should see diagnostics come out of this work that predict, for example, whether you are going to progress to diabetes or if you have a type of obesity that is best targeted with a given therapy.

“Those will be the kinds of things you see come sooner, and they will be specific to the individual. They’ll be using the networks in a given individual to identify the subtype of disease that individual has and what kind of therapy they should receive. The revolution this will bring is defining these predictive network models for disease that will potentially enable us to get the right drug to the right patient at the right time.”

So is personalized medicine really the way of the future? Schadt thinks so: “There is a serious commitment to better understand the biology, so can we reconstruct models that incorporate things like DNA variation information and molecular state information to predict disease states, and so that we can stratify patient populations to ensure the right drug is given to the right person at the right time. That will definitely be the aim, and our study shows what it will take to get there.

“There has been a lot published on these genome-wide association studies, where the aim is to associate changes in DNA with changes in disease states. The question is whether that will be enough to get at personalized medicine, to understand the type of disease you have and the type of drug you should take to treat that disease.

“What our work shows is that more information is needed; that these diseases are more complex than we anticipated, that they involve hundreds or maybe even thousands of genes and that you need to get at these molecular networks to define what the different types of disease subtypes are and then use that information to stratify the populations and get at treatments that are specific to those subtypes.”

About Dr. Eric Schadt

Dr. Schadt is Scientific Executive Director of Genetics at Merck Research Labs. He has a background in molecular biology and applied mathematics, and brought those two disciplines together to inform complex diseases at Rosetta Inpharmatics, which was acquired by Merck in 2001. His responsibilities in the genetics department include overseeing basic research, target discovery validation and clinical pharmacogenenomics.