Following column paints a bleak picture for RNAi or even some siRNA attempts at drug discovery...
+++++++++++++++++++
Nature's Many Quirks
Limit the Possibilities
Of DNA-Based Drugs
November 11, 2005; Page B1
If only the B-Raf gene weren't microscopic, scientists would have painted a big, fat bull's-eye on it.
This gene, researchers discovered in 2002, is mutated in some two-thirds of all melanomas. The mutation activates a protein that triggers a dastardly biochemical cascade, apparently leading to this deadly skin cancer. A company discovered a drug to knock out the rogue protein, engaging in the sort of "gene-based drug discovery" that is supposed to characterize 21st-century biomedicine. But the compound has been disappointing, to put it politely, doing essentially nothing against melanoma.
In June, scientists figured out why. Yes, the B-Raf mutation shows up in loads of melanomas. But while it may initiate the cancer, once the malignancy takes hold, B-Raf just seems to be along for the ride.
"If it's not important once the cancer develops, targeting it with a drug may have little effect," says Paul Khavari of the Veterans Affairs Palo Alto Healthcare System and Stanford University, who led the study published in Nature Genetics.
Last week, I questioned claims that the new HapMap of DNA misprints would lead to reliable ways to match genetic profiles with the risk of disease. Unfortunately, the chances that data like the HapMap's will lead to new drugs for treatment or prevention of those same diseases are even slimmer.
The concept is brilliant. Start with the bits of DNA mutated in a disease, as the HapMap is already doing. Work out what the rogue gene does and the biological pathways to illness. Identify molecular targets where that pathway can be blocked. Presto! New drugs that, blocking the pathway, treat or even cure disease.
It didn't work with B-Raf because this gene apparently isn't part of a crucial pathway in melanoma, even though the mutation shows up in more melanomas than belts do on pants. If the pants are being held up by suspenders, cutting off the belt won't make them fall down.
"Fewer and fewer diseases feel like they can be traced back to one thing wrong with one gene," says medicinal chemist Derek Lowe, who works in drug development and writes the blog "In the Pipeline." "Stepping into a disease pathway with a drug, unless the target is a chokepoint for the disease, won't affect the disease."
Not knowing whether a mutation is crucial or incidental, belt or suspenders, is only one obstacle looming between genetic data and new drugs. Basing drug discovery on genes is dicey for another reason. There are only 25,000 or so human genes. There are more than 10 times that many proteins, the workhorses of biochemical reactions.
"Because the number of proteins is so much greater than the number of genes, there is a lot of stuff happening downstream of the gene," says Dr. Lowe. That means if you target a gene and (you assume) the disease-causing protein it makes, there is a good chance you will be messing with more than you bargained for.
Consider a nongenetic example. Last month, the Food and Drug Administration called a new diabetes drug "approvable," but declined to okay it for sale. Named Pargluva, the drug targets two nuclear receptors, which shuttle in and out of a cell's nucleus. Once inside, the (inaptly-named) receptors act as switch throwers, binding to DNA and switching genes on or off. Based on years of data on the receptors, Pargluva was expected to reduce both blood glucose (for diabetes) and blood lipids (heart disease).
The problem, Dr. Lowe said, is that each receptor affects hundreds of genes, different ones in different cells, under different conditions and at different stages of growth. You think you're targeting one pathway, but (as economists say) you can't change only one thing. Pargluva had cardiovascular risks worrisome enough for FDA to send it back for more work. Messing around with a pathway expected to be good for heart disease did the opposite.
Even if a gene and pathway truly seem crucial to the disease, you're not home free. "If you block one disease pathway, a cell can sometimes circumvent that," says Vipin Garg, CEO of Tranzyme Pharma, Research Triangle Park, N.C., as happens with some cancer drugs. "The gap between understanding the genetics of a disease and drug development is actually growing."
Still, several companies remain optimistic about gene-based drugs. deCode Genetics, of Reykjavik, has analyzed the genomes of more than 100,000 Icelanders, half the adult population, and correlated genotypes with common diseases. They have discovered 15 disease-related genes, analyzed the pathways that have gone astray and identified drug targets to block or repair it. deCode has drugs in clinical trials for, among others, asthma and heart attack.
"When you start with the genetics," says CEO Kari Stefansson, "you can be sure the pathway is not just part of the disease process, but is key to the cause of the disease."
One can only hope that translates into drugs. It's hard to escape the feeling that, when it comes to using genes to predict or target disease, things are way more complicated than anyone thought. As Dr. Lowe blogged last week, "Nothing plays defense like nature can." |
