SimuGen, a computational biology startup based in Cambridge, UK, has wrapped up a proof-of-concept study of its predictive toxicology platform and intends to launch a commercial version in late 2007.
The six-person firm, which raised £75,000 ($148,000) last April from a group of angel investors, is looking to complete another round of funding to help pay for the launch. Several investors have committed funding for the second round, but SimuGen officials said that they are seeking additional funding to complete the round, in which they hope to raise up to £900,000 ($1.9 million).
SimuGen takes a hybrid approach to toxicogenomics that blends experimental and computational methods. The company has developed a gene-expression analysis protocol for human hepatocyte cell lines, associated reagents, and a set of modeling and analysis tools for crunching the experimental data to predict the toxicity of a given set of compounds.
In the recent validation study, SimuGen said that the approach was able to predict liver toxicity with 90-percent sensitivity. According to the company, this is a vast improvement over animal-based toxicity studies, which typically have a sensitivity range of 50 percent to 80 percent.
Quin Wills, chief scientific officer at SimuGen, said that GeneService, a genomics services spinout from the UK’s Medical Research Council, performed the gene expression analysis for the proof-of-concept study and that the results and methodology were validated by the UK’s Laboratory of Government Chemists. The company has not submitted the study for publication in a peer-reviewed journal “for IP reasons,” Matthijs van Leeuwen, director of business development, told BioInform.
‘Reduction to Practice’
SimuGen was founded in mid-2005. Wills said that 2006 “was really our proof of concept year and 2007 is our reduction-to-practice year, taking what we did in 2006 and making a product from it.”
Now that the methodology itself has been validated, he said, “We need to amplify the amount of toxicity that we’re testing for, and that’s what we’re going to be doing in 2007 before we launch the final kit.”
Rather than approach toxicogenomics as a pass/fail problem, SimuGen’s method is being developed to help researchers prioritize compounds based on 10 subtypes of liver toxicity, as well as dosage levels at which toxicity is likely to occur.
“We’re not kicking out chemicals, we’re triaging them,” Wills said.
But in order to refine the predictive power of its method, the firm needs to expand its internal database by testing more compounds — a project that it expects to complete before launching the product at the end of the year.
“One of the questions we get asked is, ‘Are you covering enough chemical space to make a good prediction?’” Wills said. “And it’s really not so much the case of covering chemical space as much as covering toxic space. We just don’t feel it’s worthwhile launching a product yet that only predicts the major types of liver toxicity. We really want to cover a good spectrum of liver toxicity so that you can be fairly sure at an early drug-discovery stage that if this comes out clean chances are it will be clean.”
Wills said that the company’s focus on dose response should differentiate it in the toxicogenomics marketplace, in which companies like Iconix and Gene Logic have already established a foothold.
Most companies in the market, he said, either treat toxicogenomics as “a big bioinformatics exercise” by building databases, mapping pathways, and exploring the underlying mechanisms of toxicity, “or they use it as a big machine-learning tool,” in which chemical “fingerprints” indicate the likelihood of toxicity.
SimuGen’s approach, on the other hand, generates quantitative dose-response models that researchers can use as a gauge to decide whether to proceed with a compound or not. This feature is particularly important for compounds with a narrow therapeutic index, like the over-the-counter analgesic acetaminophen or the widely prescribed anticoagulant warfarin, which are “toxic at a level slightly higher than the level for which they are therapeutic,” Wills said.
The company’s software also enables medicinal chemists to perform structure-activity relationship analysis to determine “what they could be doing to the molecule to make it less toxic,” he said.
SimuGen claims that its approach is also cheaper than animal testing, but Wills was unable to provide specific cost savings due to the large amount of variability in animal-based studies. Nevertheless, he said, “animal house expenses suck up your budget very, very quickly,” so any approach that can eliminate some of those tests should be welcome among pharmaceutical customers.
In addition, he said, because the method uses cell culture instead of animals, companies don’t have to synthesize as much compound to perform a study, which should also help reduce costs.
SimuGen has identified contract research organizations as the initial target market for its commercial kit, which will include protocols, chemistry, and software. While it hopes to sell directly to pharmaceutical firms, Wills said that CROs present a more promising market for several reasons. One challenge in pharma companies, he noted, is that companies selling automated methodologies usually have to pitch them to people who are at risk of becoming obsolete if the technology delivers as promised — a clear disincentive to sign a purchase order.
2006 “was really our proof of concept year and 2007 is our reduction-to-practice year, taking what we did in 2006 and making a product from it.”
“One of the problems we’ve faced when we spoke to pharma companies was, more often than not, the people who would actually say yes to the product are the people who might lose their jobs,” Wills said, “Whereas if you go with CROs, this is something that they put alongside everything else they offer.”
Wills said that the company also sees a large opportunity for its technology beyond the pharmaceutical market. He cited two European initiatives for chemical testing that the company expects will drive adoption of predictive toxicology methods. The first, the REACH (Registration, Evaluation and Authorization of CHemicals) initiative, calls for 30,000 chemicals to be tested for toxicity and registered with the European Union by 2018. The second, the European Cosmetic Directive, calls for an end to animal testing for all “non-essential” chemicals by 2013.
In the case of REACH, “what the industry really needs is a way to screen down those 30,000 compounds to 1,000, perhaps, before putting them into animals,” Wills said, while the Cosmetic Directive is expected to accelerate adoption of in vitro approaches like SimuGen’s as an alternative to animal testing. He noted that the company has been following up on these initiatives “fairly aggressively.”
SimuGen sees growth ahead in the pharmaceutical market as well. While toxicogenomics has established itself as a preclinical tool in pharma, “the real need in pharmatoxicology is to shift that decision to earlier on,” Wills said. “It’s great making the decisions at animal-stage testing, but then you’ve still invested many, many years of work into that molecule. If one can help someone decide whether something’s toxic or non-toxic early on, that’s the way to go.”
Wills cited the ADME field as an example of the likely trajectory for predictive toxicology tools. “For many, many years, drugs have been lost because of poor pharmacokinetics, etcetera, but now that’s far less of an issue because ADME is being done fairly early on,” he said. “It has a far way to go, but ADME really has shifted from a late-stage tool to an early-stage tool, and that’s what we really need to see happening with toxicogenomics, with toxicology, and that’s what we do as a company.”