Back to biology: how BERG is using artificial intelligence to tackle cancer
Pharmaceutical start-up BERG is using artificial intelligence to drive its back-to-biology approach to fighting cancer. Co-founder, president and CEO Niven R. Narain talks about the start-up.
Ten years ago people thought they were crazy, but today the team behind Boston-based pharma start-up BERG is catching the industry’s attention with its data-driven, back-to-biology approach to drug discovery. A combination of maths and biology, they believe – and are now beginning to prove – is the future of tackling the world’s most complex diseases, from cancer to Parkinson’s.
Since the Human Genome Project (HGP) was completed in 2003, there’s been a widely accepted belief that understanding genomics is the answer to curing cancer. But that’s far from the view held by BERG co-founder, CEO and President Niven R. Narain.
Beyond the genomes
“Genomics is one component of biology,” he explains. “It’s just one part of the full narrative, which also includes proteins, metabolites and lipids and the way they all interact with each other. I’m not saying it’s an unreasonable approach, but what I would say very emphatically is that you need to go beyond the genomes and employ other parts of the biological system to truly capture what the reality is.”
And that’s exactly what he and his colleagues at BERG have done with their proprietary software, which uses artificial intelligence (AI) to process enormous amounts of biological data and thereby reveal what happens in minute detail during the journey from good health to cancer. The idea is that the resulting insights allow for a more informed hypothesis than those used in traditional drug discovery, which in turn enables more efficient drug development.
“[The current methods pharmaceutical companies use to discover new drugs] are just not efficient,” Narain says. “Out of every hundred drugs that go through a Phase I trial, only one becomes a good drug – that’s a 99% failure rate – and I think that comes from the fact that we want everything to fit in the same bucket and we’re going to treat every patient the same way.
“Starting with a hypothesis driven chemical screen approach is like trying to find a certain Tube station in the middle of London if you don’t even understand the Tube exists. You might stumble across Piccadilly Circus because there’s a big station there but you haven’t taken the time to realise that there’s a well-orchestrated and well-organised train system. That’s what the BERG approach is saying. Let’s try to map out what the full system of that architecture is in various diseases and then start the drug discovery process.”
In this way, Narain thinks drug targets can be discovered in less time and at a fraction of the cost.
BERG’s first cancer drug
BERG’s approach has resulted in the creation of the first complete model of how pancreatic cancer functions – as well as cancer drug BPM 31510. In a nutshell, the model showed the team that mitochondria, which help supply energy to cells and also control their ability to die, are absolutely key to allowing cancer to flourish. Cancer cells are able to turn off the mitochondria and generate energy from lactate rather than oxygen (as healthy cells do), which also results in them losing their ability to die. BERG’s drug redelivers high levels of the enzymes that help the mitochondria function, reversing this effect and essentially turning cancer cells back into normal cells.
Initial trials of the drug on pancreatic cancer patients have been promising. Not only have Phase I trials, which began in 2013, shown evidence that the drug is far less toxic than chemotherapy because it’s made from a naturally occurring enzyme in the body, BERG has also observed evidence that the drug is working to extend survival time and decrease tumour size for some pancreatic cancer patients, including those who have been subject to extensive chemotherapy.
The trial has also validated the original cancer model BERG built back in 2010. “We’ve been collecting tissues and samples from patients live on trial, we’ve rebuilt the cancer model, and compared it to the early model we made in 2010 and what we’re seeing is that it actually showed the same elements and trigger points, which was really cool validation not only of the platform’s capacity to predict, but also validation of the drug’s mechanism of action on a clinical, functional level,” Narain says.
“We’re just starting Phase II trials on pancreatic cancer patients and we’re really excited that this drug can offer hope to patients in a much more safe and effective manner.”
Indeed, the team’s ultimate hope for BPM 31510 is to not only improve survival times for patients, but also ensure that time is quality time. “A lot of patients, when they get to end stage cancer are given a toxic drug and experience end of life issues that are just horrible,” Narain explains. “Our goal is to truly extend survival times of patients but do so in a manner where their quality of life is either unaffected or improved significantly.”
Persistence pays off
According to BERG’s senior vice president and chief analytics officer Slava Akmaev, who is the architect of the company’s AI platform, the team certainly won’t stop at pancreatic cancer. “The beauty of BPM 31510 is that it could potentially work on a whole range of highly metabolic tumours,” he explains. “Further trials are planned for brain cancer, and plans are also under development for detailed trials in gastric, oesophageal and bladder cancer.”
Beyond cancer, the company is also working on three drugs for diabetes, one of which will hopefully be ready for human use by the first quarter of 2018. BERG also has a very exciting neurology programme, through which they hope to release one of the first validated drug targets in Parkinson’s in over 15 years. The company is also researching treatment pathways for chronic diseases such as chronic kidney disease and congestive heart failure. The precision medicine model underlies it all.
Looking at the pharmaceutical industry in general, Narain predicts that the fusion of maths and biology, technology and healthcare, is only set to become more commonplace, as companies and patients alike realise its potential. “When we started doing this early on, people thought we were crazy. We were not the popular guys. But we stuck with it because we believed that it would take hold one day,” Narain recalls. “We took the tough path; we’re just really happy we stuck with it!”