Virtual organs: cutting drug development risks

Developing a drug remains a hugely uncertain, and therefore expensive, business. Many promising drugs fail further along the development process, even if they have been tested on animals. Furthermore, researchers can be left in the dark if it’s unknown why they failed, and have to return to square one.

“Even when compounds pass all the animal tests and look good after testing on humans, one in 10 actually reaches the market—in other words, nine of the 10 fail,” says Kalyanasundaram ‘Kas’ Subramanian, chief scientific officer at Bangalore-based informatics and data analysis company Strand Life Sciences.

“If the molecule fails somewhere early in the process, it’s not a problem because you always have backups you can test. But if the molecule fails after you’ve tested it on humans, it’s a huge economic loss because of the time you invested and the money spent on conducting clinical trials, which are very expensive,” he adds.

With the cost of developing a new drug in the billions, pharmaceutical companies are looking to change the odds in their favour.

Developing a ‘virtual’ liver

Moving research from the lab bench to the realm of data analytics could be a solution.

One reason so many candidate drugs fail is that they have a toxic effect on the liver. While it is the liver’s job to detoxify substances that enter the body, the detoxification process can often convert a benign molecule into something toxic.

“It’s a very difficult problem to crack, because toxicity is not due to any one particular mechanism that you can see,” Subramanian says. “There are many different ways a drug can be toxic to the liver.”

He suggested creating a computer model of a healthy liver capable of generating the likely outcomes when a drug molecule is introduced. The virtual liver allows researchers to see not only whether a candidate drug is likely to fail but also why it would fail, by simulating the drug’s effect on normal liver function.

“Understanding the mechanism helps you decide whether you can salvage what’s going wrong in the series, or if you have to shelve the entire series. You can do all this before testing on humans,” says Subramanian.

The ethical benefits of using the virtual liver are plain, as it reduces the need to test on animals. Also, drugs having no ill-effects on humans that may have been vetoed for failing during animal tests may get their fair shake.

Pharmaceutical companies have already begun to use the virtual liver as a tool in the research and development process, to develop treatments for conditions where the liver is a vital part of the biology, such as diabetes and cardiovascular diseases.

New era

The completion of the Human Genome Project (HGP) in 2000 heralded a new era for medical science, as the endeavour to map the entire human genome delivered a new understanding of genes’ involvement in disease.

“There was a lot of excitement in biology that the path of medicine would change completely,” says Subramanian.

The HGP paved the way for next generation sequencing, which then introduced a new way to search for drug candidates: highthroughput screening, in which thousands of tests are run rapidly to identify new drugs for disease targets.

With so much data generated by this high-volume method of experimentation, four computer science professors from the Indian Institute of Science decided to develop a platform geared towards licensors to help them make sense of the figures, and Strand Life Sciences was born.

IP protection

Strand decided not to protect its platforms with patents, opting instead to keep some of its technologies and processes as trade secrets. In fact, the company started looking into patenting its creations only after making the virtual liver available in 2010.

“After we started taking the virtual liver out into the industry there were a lot of questions,” Subramanian says.

“While the virtual liver manifests as software, in the sense that all these mechanisms are eventually captured as software, it pertains more to biological systems than pure software patents.”

“We were trying to model biology so people would ask: what is the biology that’s been captured, how you are modelling it, how can you use it and so on. It seemed as though we would have to disclose a lot of what we did, so we felt that we needed to protect it.” The virtual liver is one part of a suite of software-based solutions that Strand offers. Thousands of laboratories around the world use its Avadis Next Generation Sequencing platform that analyses, manages and visualises data from experiments.

“We didn’t go for patenting Avadis, because we didn’t feel that there was any value in trying to protect it via a patent,” Subramanian says.

“A lot of our patents are around understanding mechanisms, with most covering the virtual liver. While the virtual liver manifests as software, in the sense that all these mechanisms are eventually captured as software, it pertains more to biological systems than pure software patents.”

In February, Strand received a US patent covering the virtual liver’s method of predicting organ toxicity; however, landing the patent wasn’t quite as straightforward a process as the team may have hoped.

“There was lots of back and forth with the examiner,” Subramanian says.

Many methods relating to testing for liver toxicity exist in the prior art, although for different methodologies and processes, he explains. Strand had to convince the examiners that its technology is distinctly different.

Progress in the EU, however, has been much faster, and Subramanian hopes that a second virtual liver patent covering a method for identifying liver toxicity using metabolite profiles will be issued this year.

Lending the liver

So far, about 10 pharmaceutical companies are using the virtual liver as part of their drug development processes, although this isn’t on a licensing basis: “Our business with the virtual liver is more of a consulting or services model,” Subramanian explains.

“For the purposes of assessing toxicity, pharmaceutical companies will send us their compounds. We then run the compounds through the assays we’ve developed here in Bangalore, and run simulations using the data that come from the assays, and generate reports for them.

“The virtual liver is too complex to hand over to somebody.”

Subramanian says that some cosmetics companies have been in touch too, to assess their products’ potential toxicity using Strand’s virtual hair and skin models.

Genetic panels

Building on its data analysis expertise, Strand is also currently developing a collection of genetic ‘panels’ to test individuals' genetic predisposition to developing certain diseases.

Whereas genetic diagnostic companies like Myriad and Gene by Gene look for mutations on single genes to make their diagnoses, Strand searches for all the different kinds of genes that are involved in the development of the disorder, putting them together into a panel that can be measured using next-generation sequencing.

The panels are “in a sense” proprietary, Subramanian says, as Strand has already carried out the analysis, determined which genes to examine, figured out which regions are most important and worked out how to interpret the mutations.

“The next step is understanding the relevance of the data,” he says.

“We have a huge team at Strand that has been reading all the literature and curating every mutation-to-phenotype association that is known or reported in the literature. We now have more than 100,000 of these catalogued.”

Strand is starting to offer its genetic diagnostic tests to around 30 hospitals in India, and will begin establishing its own Strand Centers for Genomics and Personalized Medicine in the US and UK.

“There’s going to be a Strand centre at the El Camino Hospital in Mountain View, California, and we are in discussions with several other community hospitals,” Subramanian says.

“Our strategy is not to go after the Mayo clinics and Harvard Medical Schools because they will want to offer their own services. But there are many smaller hospitals that serve local communities that are interested in offering this to their patients, but don’t have the expertise or wherewithal to create it.”

As companies such as Strand work to tackle the risks associated with drug development, one thing is certain: as demonstrated by the popularity of the services provided by genediagnostic companies, the desire to find out what the medical future holds for us is not going away any time soon, as we enter an age of personalised medicine.