Drug development is a gamble. Pharmaceutical companies can spend millions, if not billions, on testing, trialling and formulating candidates only for them to prove ineffective or unsafe after years of perceived progress.
As a result, reducing risk of failure has become a major focus in the pharma sector. The de-risking process begins in the discovery lab, says Professor Kenneth Kaitin, Director, Tufts Center for the Study of Drug Development at the Tufts University School of Medicine.
“The high cost of bringing a new drug to market and the considerable risk of failure during the clinical development phase are factors pushing companies to de-risk early phase of development,” he says. “The goal is to select only those early-stage candidates with a high likelihood of success to move forward into late stage development.”
Outsourcing has emerged as an effective de-risking strategy, according to Kaitin, as drug industry efforts to reduce the chances of failures in early development often involve third party specialists.
He adds that due to the high cost of discovery research and the large number of compounds that need to be screened before selecting those to move forward, more and more large pharma companies have decided to off-load early phase development to academic centers and small pharma and biotech.
“As former Allergan CEO Brent Saunders was quoted as saying in a 2015 Forbes article, ‘The idea that to play in the big leagues you have to do drug discovery is really a fallacy… Discovery had not returned its cost of capital,’” says Kaitin.
Cutting edge candidate selection
Some failures in early development can be useful, or at the very least instructive. Drug candidates that fall before the first hurdle can save companies millions in otherwise wasted R&D dollars and help them focus their research effort.
“It’s important to note drug development is a high-risk endeavour, and failures during the early development phase are both anticipated and valuable, in terms of better understanding drug mechanisms and effects,” says Kaitin.
However, while pharmaceutical innovators acknowledge the risk of failure, most try to do everything they can to increase the chances of success.
“Advanced data analytic tools, such as artificial intelligence and machine learning, are increasingly being employed by companies to rapidly screen large numbers of candidates and select those with a high potential for clinical success” - Professor Kenneth Kaitin, Director, Tufts Center for the Study of Drug Development
Data and analytics are key to the modern candidate screening process. Kaitin says firms are using cutting edge tech to single out the early phase compounds that have the best chance of becoming blockbuster medicines by improving the efficiency and lowering the cost of the selection process.
“Advanced data analytic tools, such as artificial intelligence and machine learning, are increasingly being employed by companies to rapidly screen large numbers of candidates and select those with a high potential for clinical success,” he says. “Also, companies are increasingly engaging in precompetitive strategic partnerships and integrated alliances, in which they partner with government research centers, academic scientists, patient associations, and competitor firms.”
Examples of this approach include the Accelerating Medicines Partnership and Alzheimer’s disease Neuroimaging Initiative. Under such accords participants share resources and scientific brainpower to better understand disease mechanisms and improve the likelihood of developing safe and effective treatments.
In the seven months since SARS-CoV-2 – the virus that causes Covid-19 - was first detected, 141 vaccines have entered development. And – as of June 2 according to the World Health Organisation – 13 have started clinical trials.
The speed with which the vaccines sector - and the pharmaceutical industry as a whole - has reacted to the pandemic is testament to the gravity of the situation.
It is also an indication of the strides industry has made in candidate selection, according to Kaitin, who says tools being used to reduce the risk a potential SARS-CoV-2 jab failing during clinical development have wider application.
“In my opinion, one of the most exciting things to come out of the race to find vaccines and treatments for COVID-19 has been the practical integration of gene sequencing and editing, along with machine learning and advanced analytics, into the drug development process,” he enthuses. “The pandemic has catalysed the incorporation of these tools into early phase development and will enable scientists and developers to better understand the pathophysiology of some disease and rapidly screen for potentially effective candidates.”
Another unintended consequence of innovation in technologies developed to reduce early phase risk is in the field of drug repurposing.
Finding an alternative use for an approved pharmaceutical product does not – by definition – involve early phase development. All the work has already been done. Instead the challenge – and indeed the risk – associated with such activity is the cost and time it takes to find the drug most that has the best chance of being repurposed.
And it is in this area where early phase de-risking technologies could have a part to play.
Kaitin says early phase development does not really apply to generic drugs or, in many cases, 505(b)(2) drugs: “I can say, however, that tools such as machine learning are extremely valuable in drug re-purposing, as we’ve seen in our attempts to find treatments for COVID-19. The ability to rapidly screen huge drug libraries and shelved compounds increases the chances that if a potentially effective treatment already exists, it can be quickly identified. Remdesivir is a good example.”
Candidate safety testing is also an area in which the pharmaceutical industry is working to reduce risk.
The vast majority of compounds coming out of discovery are examined in animal models to determine if they are safe to move into human trials. However, other approaches are used to augment such research, according to Chris Magee, Head of Policy and Media at Understanding Animal Research.
"The animal tests in regulations are largely intended as simple safety tests, rather than detailed studies into efficacy, or proxies for every genetic permutation we might see in millions of people.” - Chris Magee, Head of Policy and Media at Understanding Animal Research.
“Drugs must typically be tested in one rodent and one non-rodent species before the first human trials can take place and animal tests will often occur concurrently with latter stages of testing some drugs,” he says. “However, in vitro or in silico methods must be used instead of an animal where possible under the Animals (Scientific Procedures) Act 1986 , so there are exceptions where one species can be used.”
He adds that researchers also use in vitro methods alongside animals, either to test different aspects of the compound or increase their confidence in the sum of the data: “The animal tests in regulations are largely intended as simple safety tests, rather than detailed studies into efficacy, or proxies for every genetic permutation we might see in millions of people.”
Other de-risking efforts are focused on the development of animal models that are more representative of the target human disease. While there are potential benefits for some diseases, breeding better models will not be possible for all conditions, Magee says.
“People talk about breeding or otherwise creating better animal models for this or that condition and of course ensuring the study methodology is robust, but there are weaker and stronger species for investigating different diseases which make blanket conclusions about ‘animals’ inappropriate, plus diseases are not of equivalent complexity.
“Indeed, a coronavirus isn’t even in the same room as Parkinson’s disease in terms of complexity and a higher failure rate in PPV is most probably correlated to the complexity of the disease the new intervention being tested is hoping to address.”
Magee concludes that to make progress towards reducing the chances of development failure, it is important to look at how candidate compounds are advanced with some granularity: “you’ll make more progress in areas where the disease is simple and the animal model better at indicating a likely human outcome, all in the context of a commercial environment where safe, effective compounds still fail to be developed.”