PRINCETON, NJ--The way Dale Pfost sees it, single-nucleotide polymorphisms (SNPs) represent the most commercially and scientifically compelling opportunity of the genomics revolution. Associating SNPs to diseases is what Pfost's company, Orchid Biocomputer, has set out to do.
"It's not a question of whether these associations will be found, it's who is going to find them first," Pfost said, comparing Orchid's mission to the early days of genomic sequencing. "Back in 1996 it was obvious that the human genome project would be done. It wasn't a question of if, but of when it would be done," he added.
Orchid, which has roughly 140 patents pending, allowed, or filed, was founded in 1995 by Sarnoff, the former RCA research lab that now conducts for-profit high-tech product development for Princeton University. Last week, the company was awarded a patent for a technology it calls primer extension genotyping. Marketed as Genetic Bit Analysis, the technology enables highly accurate analysis of genetic sequences necessary for clinical and SNP association studies.
Orchid's other patents are split between microfluidics and SNP technologies. "Our technological strategy from the beginning has been massively parallel array reactions," said Pfost, who formerly presided over Oxford GlycoSciences and served briefly as chairman of Pangea Systems. "Other people have looked at DNA sequencing using electrophoresis, but we've always looked at massively parallel arrays and reactors and that's the stuff of high-throughput drug discovery," he said, noting, "That's what attracted me to the company. It looked like the next generation of biotech."
Pfost met with BioInform recently to discuss Orchid's SNP-scoring technology and its mission. Part one of the interview appeared in BioInform's May 24 edition.
BioInform: How will pharmaceutical companies utilize your SNP-scoring technology?
Pfost: If we provide an additional set of tools to drug company scientists in the preclinical stage, they have a better set of correlations to decide which molecule they want to go forward with using the full resources of the company.
There are all sorts of strategies: you can continue with a drug and introduce a diagnostic procedure along with it to make sure certain people don't take the drug; you can abandon the drug; or you can go to a fallback compound that might not have as many correlations to the polymorphisms. In most of these cases you'll need the technology to be able to score SNPs. We could provide it on a product basis, a service basis, during clinical trials, or preclinical. Or we could apply it to existing compounds, for drugs that are off-patent, and extend the intellectual property rights by buying drugs that are almost off-patent and extend it that way.
BioInform: How will SNPs also play a role in clinical practice?
Pfost: You or I will go to the doctor and the doctor will say, I could give you this drug or this drug, and what we've found, generally speaking, is as long as you're not genotype 587, then this drug is probably better for you, but if you are 587 then this drug will be better for you. This is going to give a broadly applicable platform to be able to measure a whole series of tests through the DNA that has correlations to how we respond to the drug. This platform--the mechanics, the fluidics, the biochemistry--will look a lot like an infectious disease test.
In clinical practice it makes the most sense to have clinical reference laboratories running samples, in which case our technology will be licensed to diagnostic instrumentation providers such as the reference laboratories all around the world. Throughput for those places will be a few hundred or a few thousand per day.
The throughput we need to find these associations to begin with is even higher. We need millions per day. That's why we're establishing this factory. [Orchid's Mega Snipatron was described in part one of this interview.]
BioInform: Will the information you discover at the high-throughput stage be put into a database that clinicians will then be able to compare their samples to?
Pfost: Yes, there are a couple of steps along the way, but eventually that's exactly right. In a sense, you want doctors to be able to find out in an easy fashion if they want to give these drugs. And they don't want to know the details of the genotyping. That's a database job.
Look at the scenario: We find 1,000 patients who have taken a drug, and we find 200 people who have had a problem with it. The target of that drug is a protein and the enzymes that metabolize that drug are another protein. The disease that we're trying to treat here is comprised of a series of pathophysiologies. So let's look at all these genes--there might be a few dozen--and then while we're at it we better look at a few dozen other genes that have to do with normal stuff that's going on in the body.
Today we only know about 10 percent of them, but time's-a-wasting, so we start with that 10 percent. You ask, What phenotypic information do we know about these patients? Blood pressure? How old were the grandparents when they passed away? You start to correlate a whole bunch of information to what you've pieced together with sophisticated informatics.
At the end of the day, some of the correlations that are not causal might still be useful but just not correlate. What you'd really like to do is have causal polymorphisms that actually are the root cause of the differences between patient response. That's a big challenge for informatics. Our industrial collaborations will be completely geared toward that. [Orchid acknowledged that it is seeking one such collaboration with Pangea.]
Our clinical genetics network is going to provide us with a whole series of samples and information about patient samples that will allow us to feed the informatics engine, and the factory we're establishing will provide the data. It's all coming together.
BioInform: What will you do to ensure patient confidentiality?
Pfost: Patient confidentiality is ultraimportant and the company is completely dedicated to ensuring that. Many states have already passed laws about the use of genetic information.
This phenotyping information is a whole series of tests one could perform on a sample. You actually don't care that it came from John Doe 33, you care about how this phenotypic information correlated to response to the drug. So it doesn't have to do with John Doe 33, it has to do with the correlation. With a database, you blind the identity of the patients and encrypt it so it isn't usable by the individuals doing association studies.
This brings up a different business opportunity, though, which is to provide clinical testing for patients. That's a separate, important activity. As this information becomes known, people will want it.
In my own family we've had two situations where family members have died from response to a drug. This stuff is happening out there. If you're going to go to the doctor to get some drug that 99 percent of people are completely fine with, but that, after eight days of dosing, puts you in the hospital with liver damage, you want to know.
There's an opportunity here to empower patients directly and have them be aware of this
We're looking at the existing pharmaceutical infrastructure, the pipelines of other companies, and looking for ways to add value to it, either by working with them or by patenting our own intellectual property rights around them. That has a completely different kind of timeline. It's months to years, not years to decades.
predisposition themselves. And that's a separate transaction, separate samples, and separate sets of confidentiality and security issues.
BioInform: Do you believe the US Food and Drug Administration will begin to require genetic risk information with new drug applications?
Pfost: The FDA is very ready to start receiving this information. If you can unwind the hidden variables behind heterogeneous response or idiosyncratic response to drugs, everybody is better off. This is not one of those things that you can wish to go away. This is here to stay. Genetic variability is what makes us human. Therefore, FDA would love to have this information, and I would venture to say, within not so many years, will require it.
I believe the clinical trials of the future will at least have the target of the drug looked at if not a lot more than that. So again, this is just an inevitability.
BioInform: As Orchid is about to go through a second financing round, how are you being received by investors?
Pfost: The question is how do you make money in this space in a different fashion? Sepacor is a good example. They have the technology to improve existing drugs. That's beautiful because you go in and say, this drug could be doing twice as well if it didn't have this problem.
That's what we're doing. We're looking at the existing pharmaceutical infrastructure, the pipelines of other companies, and looking for ways to add value to it, either by working with them or by patenting our own intellectual property rights around them. That has a completely different kind of timeline. It's months to years, not years to decades. Therefore we have been well received by the financial community and we think the limelight is on the space.
We think we're going to be oversubscribed in our next round. We're going out looking for $25 million and there's been a lot of interest. It's always hard to predict IPOs, but markets permitting, we could have one in about 24 months.