Healx raises $10M to help repurpose drugs for rare diseases

The company uses AI and machine learning to find cures for diseases the big pharma cos are ignoring

One of the problems that arises when it comes to for profit healthcare is companies need a financial incentive to take care of patients. Take pharma, for example; their model depends on creating drugs that can be mass produced and taken by a large portion of the population. If a cure isn't going to be profitable for the company then there's no incentive for them to explore it.

That model inevitably leaves out a huge portion of the population that have diseases that aren't as common. Because the number of people who suffer isn't enough to turn a profit, those people are left without relief. 

"Worldwide there are 350 million sufferers from rare diseases. Just to put that into perspective, that’s equal to the population the size of United States. So, it’s a huge unmet need. It’s 1 in 20, but, unfortunately, out of the 7,000 rare disease, 95 percent are without approved treatment," said Tim Guilliams, co-founder and CEO of Healx, a company focused on finding existing drugs to heal rare conditions.

"That’s the result of the traditional pharma model, and the blockbuster model for the pharma companies, where it is not viable to develop drugs that take 10 to 15 years, $2 billion to develop and to do that for small disease populations."

Healx solves the problem by using AI and machine learning to cut down on the cost and time for developing treatments for rare diseases. On Wednesday, the company announced that it raised $10 million Series A funding round, led by Balderton Capital, along with existing investors Jonathan Milner and Amadeus Capital.

The Cambridge, UK-based company was founded in 2014 by Guilliams, a Bio-Chemical Engineer and founder of the Cambridge Rare Disease Network, along with Dr. David Brown, the inventor of Viagra and ex-Global Head of Drug Discovery at Roche and Dr. Andreas Bender, a lecturer and researcher at Cambridge University’s Centre for Molecular Sciences Informatics.

The founding of the company was inspired by a child named Bertrand Might, who was born with a rare disease.

"Bertrand, when he was three years old, was the very first patient to be diagnosed with a completely new disease, one never reported in the scientific literature before. His dad, Matt, was basically told that there were no treatments available, that he should enjoy the few years he had left with his son and that nothing could done. The guy didn’t take no for an answer and so he started by writing a blog about it in the New Yorker and when we read that we actually contacted Matt and asked if we could potentially help to find a treatment for his son who, at that time, was the only patient with this disease," Guilliams explained. 

From there, the founders realized how big the problem actually was, and how many people are actually affected by these diseases that the pharma companies don't have any incentive to treat. And so they decided to create a company that would tackle the problem. 

"Those with rare disease have been let down and so that’s where we were really moved and inspired by Matt and his son, to start with, and then realized that the size of the problem was enormous, and that actually AI tech and expert drug discovery could help and do something about it."

What the company does is take drugs that are already existing on the market and uses AI and machine learning to figure out ways they can be repurposed to fight other diseases as well. Guilliams describes is as "a bit like a giant search engine for rare diseases."

"We look at all of the scientific papers, patents, clinical trials, all the source of information, put this in a huge database, or the technical term is knowledge graph, and then interrogate that with all sorts of machine learning algorithms. Basically we’ve developed something called HealNet, which is looking at all all of the rare disease information, translating that in billions of data points, that will then match to all of the drug and treatment information, and that’s how we find new links and new applications between drugs and rare diseases," he said. 

For Brown, who had spent 30 years in big pharmaceutical companies, the problem, as he saw it, is that "the way those companies operate just isn’t applicable to rare disease. It isn’t, I think, actually applicable to a 21st century technology, either."

"So we tried to apply 21st century technology, and thinking, to the problem of how do you actually fund research on diseases that may one affect 100 people or 1,000 people or even 10,000 people? Because the pharmaceutical blockbuster model needs sales in the billions to repay their investment and their failure rates. What attracted me to working with Tim on this was the opportunity to completely rethink the whole medicine discovery process and try to do it in a 21st century way," he said.

There are three things that typically go wrong with trying to get new drugs on the market, Brown told me. The first is that the wrong target is chosen. Second is the drug molecule itself cannot be optimized effectively. And, third, is drug safety. By repurposing existing drugs, Healx avoids both the safety problem and the molecule problem, leaving only the issue of finding the right target for these drugs. The way the company gets around that issue is by using its technology to do gene expression profiling to see which drugs will actually work for that patient and that disease. 

"We have a totally different approach than the pharmaceutical companies. They are picking a mechanism, making a guess, of which biochemical mechanism might work when they get to the clinic. We’re absolutely not doing that because it’s been very low success rate with the industry doing that. So we’re doing it in a hypothesis free way, where we look using modern genomic and transcriptomic information," he said.

"By transcriptomic I mean the RNA that’s produced from the DNA, so it’s what’s actually happening in the patient, rather than what could happen, which is what the DNA tells you. We look for what is different in that RNA expression between the disease state and the normal state and you see a pattern of overexpression of some RNA, underexpression of other RNA, and low, of course, is the same as normal, but it’s the RNA that is different that interests us. Then we search for drugs using our databases and algorithms that can reverse those differences back to normal. In other word, the RNA that’s overexpressed, we want to drop that back to normal, the RNA that is underexpressed we want to increase that back to normal."

Having a hypothesis-free approach also allows the company to take out potential inherent biases that plague other companies/

"What I’ve observed about most companies using AI for is basically to reinforce their biases. They’re applying it to the old drug discovery model, so they’re just trying to fix parts of the model that are broken but it still comes down to that initial guess of: is this the right target? If that is wrong then it doesn’t matter how much AI you put apply, everything else is going to fail later on. So we’re avoiding doing that and removing human bias as much as possible."

Working with patient groups

Healx’s technology has already been used successfully in roughly 10 projects, and there are two drugs that are currently moving towards clinical trials, though the company could only discuss one of them, a drug for Fragile-X that was discovered by Healx’s platform. The company started working in concert with FRAXA, the patient group for the Fragile-X Syndrome, around a year and a half ago. 

"It cost us about $100,000 to get from started up to where we are now; it would have taken a pharmaceutical company tens of millions of dollars and maybe seven years to get this far. We’ve done it in 18 months for $100,000 and we’re ready to start clinical trails," said Brown.

The patient groups that Healx works with are "a really important part of the Healx model," said Guilliams, and the company works with such an organization for every project. They have true partnerships, he said, one where they both put time and resources in, and where the patient groups get a cut of the profits back.

The real win for Healx in these relationships is that working with the patient groups actually allows the company to do these project at scale, due to the resources that the groups provide.

"If you have 7,000 rare diseases, even if your algorithms go through 7,000 different disease information, it’s actually impossible for us to have disease experts for 7,000 diseases in-house, so by working with the patient groups, the key opinion leaders, we actually bring in this deep disease expertise. So we provide all of the technology and pharmacology expertise, but they bring the disease expertise, and suddenly you can actually do this at scale, because it doesn’t mean we have to have all expertise ourselves in-house," Guilliams said.

"We work with the patient groups very closely and then when the drugs are tested, and shown to be successful, the patients are already lined up for the clinical trial, so things move remarkably quick. Part of the trick for us is to really engage the patient groups and families from the very beginning. So it’s not just we’re going to do things in isolation and when we’re ready for clinical trial we’re going to start talking to patients and families. It’s actually part of the journey, they have a stake in the project and that’s a pretty unique model."

The patient groups and their expertise also helps Healx differentiate itself even further from the big pharmaceutical companies who "probably only work in a handful of diseases," Brown said.

"In a sense, they are providing us what a big pharmaceutical company has to build internally. Now, because we’re working like that, we can work with a number of patient groups in parallel and we can begin to apply our technology in parallel to all these diseases. We’ve worked on about disease 10 so far but we’re going to scale that into the hundreds in the near future, at which point we’ll be doing it on a scale bigger than even the biggest pharmaceutical company. The important massively thing is it reduces costs and timelines so it becomes economic to product drugs for these rare diseases," he told me. 

Using AI

With this new round of funding, Healx now has $12.5 million raised in total, and the company plans to use that to expand its team, which was 12 employees last year, and which has since grown to 25. By the end of this year, Healx expects to have 35 employees, far fewer than would work at a typical pharmaceutical company. 

"Back in the day, Dave used to have 2,000 scientists as Global Head of Drug Discovery at Roche and enormous budgets, but now he basically does the same with 20 techies and can get a drug ready for phrase 2 clinical trials with $100,000. We don’t need to ramp up the team to ridiculous amounts; with 30 to 35 staff we think we can actually develop a lot of potential treatments in parallel for rare diseases, so that’s a big change and sift compared to what was needed in pharma a few years ago," said Guilliams.

The money will also go toward expanding Healx's AI and machine learning capabilities. 

"The real holy grail here is to build a metabolomics capability and we’re quite advanced on that. We’re recruiting staff in that areas to try and model what happens in human cells in the disease states and which points pertivation intervention could be most effective. And we’re building our knowledge base. Just scaling that as fast as possible now so we’re capturing disease information on a much bigger scale than in the past," said Brown.

"The estimate at the moment is that we can probably count sufficient information on 200 to 300 rare diseases based on what’s known out there. That would already make it 10 times bigger than most pharmaceutical companies; they probably understand 10 to 20 diseases, not much more than that. And we’re curating the drug knowledge very carefully and generating our own data on transcriptomic properties of drugs that will give us proprietary advantage. So we’re building some new technologies and we’re scaling the established technologies."

The ultimate vision for Healxis to take pharma shake up a flagging industry with modern technology that can cut costs, and reduce the amount of time necessary to get a drug to market. 

"The vision for Healx is to build Healnet, a scalable AI-based platform that allows us to match treatments to rare diseases, but, more importantly, allows us to partner with other key groups in rare diseases, whether it’s a patient group, a pharma company, so that you can actually have an impact, not around 10 rare diseases, but, potentially, 100 rare diseases, if not more. That’s really moving towards a massively parallel and scalable drug discovery model," said Guilliams.

For Brown, the vision is to get pharma back to the kind of growth rates it was seeing decades ago.

"When I joined the industry in the 1970s, the pharma industry was kind of the Google and Apple and Microsoft of their day; they were growing at15 to 30 percent annually and really couldn’t do anything wrong. That’s gone. The growth rate of the industry is close to zero, I think it’s basically not going to grow this year, and it’s getting worse and worse," he said.

"Really what I’d like to do is completely change the model using modern technology and modern thinking so that we start to build a new industry which can compete with Google and Apple and be growing at that kind of a rate again. And we’re going to do that by what I’ve called democratization of new medicine discovery, where we engage people globally. We’re not just going to do it internally in a closed group; we’re engaging the patient groups globally. We’re a platform in place that they can work with us through, and we’ll show that this kind of hypothesis free approach is far more powerful than the old way of thinking."