The most innovative organizations have demonstrated their success stems from creative strategy paired with iterative and efficient execution. It’s all about failing smart, learning fast, and remaining adaptive. Some of the most powerful and ingenious examples of innovation can be observed in the modern pharmaceutical, biotech, and contract R&D spaces.
R&D strategy and processes have evolved significantly over the past decades and continue to become more adaptive today. In the past, drugs were created through a process that, for the most part, resembled volume-based trial-and-error. But now pharmaceutical and biotech organizations are becoming exceedingly efficient and strategic in how they target their assets of interest and position clinical trials to deliver value to patients. The strategy of bringing a drug from discovery to administration to its first patient involves an impressive blend of deep scientific acumen, operational execution, and adherence to evolving patient behaviors. Noteworthy trends are disrupting the R&D model from top-to-bottom, from drug discovery, to portfolio diversification, to patient engagement.
Machine Learning for Drug Discovery
At the beginning of drug development, organizations learn about new biological discoveries that provide additional insight into how the human body or harmful bacteria function. Then, organizations cross-examine various compounds to identify ones that may be relevant to the new science. Selecting which compounds to pursue can be a meticulous jigsaw puzzle where researchers need to understand the overall profiles of countless molecular pathways and how they respond to dynamic activity. A solution that could, for instance, predict a compound’s early-stage toxicity before clinical trials would directly save organizations millions on R&D budget and increase time runway.
While many are still in progress, there have been technological advancements and drug development AI applications that are allowing researchers to be more predictive in their decision-making. Ideally, researchers want to leverage large datasets from past trials, public archives, and experiments to model new situations, however, there is so much raw data available that it can be difficult to filter and organize correctly. At the same time, many of the rare diseases and patient populations these organizations are targeting have, by definition, little information or insufficient data available.
As a result, training a program to learn with only a small amount of available data has become one of the foremost topics in machine learning and deep learning. Methods like “one-shot learning” have been developed by academics and introduced to exploit auxiliary data to improve models with only a few data points. The combination of curated auxiliary data and increasingly refined algorithms has shown massive potential to streamline drug discovery, especially in situations in which researchers are working on novel targets with limited data. This value is evident to many of the top pharmaceutical companies who are now making serious efforts to apply machine learning to drug discovery, leading to a jump in high profile collaborations and deals the past few years.
In 2016, Pfizer announced its plans to leverage IBM Watson for drug discovery, specifically using the AI technology for its immuno-oncology research. In June 2017, Genentech announced its own collaboration with Cambridge, MA-based GNS Healthcare, with the goal of using GNS REFS (Reverse Engineering and Forward Simulation), a causal machine learning and simulation platform, to discover and validate potential new drug candidates. One of the largest drug discovery collaborations is between Sanofi and the AI intelligence drug discovery company, Exscientia, with the objective of assessing combinations of drug targets that could work synergistically. Andrew Hopkins, the CEO of Exscientia, has stated that “compared to traditional methods Exscientia system can deliver drug candidates roughly 25 percent faster and 25 percent cheaper.”
Portfolio Diversification Strategy
After finding a pool of high-potential drug candidates, the question becomes “which assets should an organization pursue, and why?” In R&D, success metrics and performance indicators are complicated because different departments can have conflicting priorities. From the business perspective, is the goal to invest in the drug that can yield the highest market potential if approved? Or is the goal to invest in the drug that has the least likelihood of failure in clinical trials? What about differentiating the portfolio to capture more of a global market?
To provide a framework for these decisions, organizations are using complexity matrices, which are tools that incorporate various parameters and criteria into a numerical ranking system that judges the holistic “goodness” of a drug candidate. The complexity matrix allows an organization to cut out the noise, approach decision-making quantitatively, and visualize the spectrum of potential for their portfolios.
The first three parameters in a complexity matrix can often deal with portfolio positioning.
- Phase – phase is a parameter used to understand the given maturity or familiarity an organization has with the asset in question – whether or not it falls into their core competencies and areas of expertise.
- Financial – the financial bottom-line is also a major consideration, as not all studies are viewed the same way in terms of investment or business priority.
- Labor – organizations consider their people and whether or not the skillset is present on the research team to effectively transition to manufacturing.
The second group of parameters deal with the supply chain.
- Risk – risk involves an assessment of supply chain robustness, as well as timing. Is the finished goods supply readily available? Are APIs available or unavailable to support the study?
- Modality – modality concerns the physical nature of the asset molecule, and whether or not the research team has familiarity with the development process. For example, dealing with small molecules is less complex and involved than creating biologics, whose specialized nature causes them to carry the highest weight of complexity, especially in the case of gene therapy.
- Maturity – the third consideration when weighing supply chain complexity is the overall maturity of clinical supply chain management, whether or not it is optimized, integrated, or linked. A major question organizations are asking is “how much of the supply chain will need to be planned ad hoc?”
Clinical study design is also deeply factored into the complexity matrix. Dosing regimen, patient population, and region are perhaps the most essential parameters to consider. Simply put, the clinical trial won’t progress unless there is a clear path to how the IP can be administered safely to the patient to yield actionable test results.
- Dosing – for example, should the study be single dose, multi-dose, or a dose escalation with multiple cohorts?
- Population – the patient population, itself, must be clearly defined. how large must the population be to produce desired results, and, more pressing, is such a population feasible to recruit for the clinical trials? Recruiting enough requisite patients for clinical trials is notoriously difficult, given R&D can focus heavily on rare diseases and severe unmet needs within populations.
- Region – patients often live far away from testing facilities and require specialized travel services, so the geography of the trial must be considered
In any case, the Complexity Matrix balances these parameters all at once to arrive at the best possible decision. The process is always iterative, and amendments are built into the approach. Depending on the clinical trial and drug in question, these parameters can be adjusted per asset on the fly. For this reason, the process of using the Complexity Matrix is a quantitative tool that requires qualitative insight to leverage.
Decentralized Clinical Research for Patient Engagement
Direct-to-consumer has become the gold standard across various industries for how organizations create channels that align with consumer experiences. How does this trend translate to pharma and biotech? Traditionally, drugs have been developed in strictly controlled and regulated laboratory settings. Patients are monitored frequently in hospitals and treatment centers, far from their homes, as their data is collected. It’s no surprise that patient retention poses a major threat to clinical trial success, as the average dropout rate across all clinical trials hovers around 30%.
In response, organizations have been exploring options to create decentralized clinical trials (DCTs), in which the treatment is better matched and tailored to the lifestyle of the patient. By incorporating digital tools and telemedicine solutions, organizations want to capture a more fluid and holistic view of patients’ true experience living with the disease, as opposed to taking sporadic “snapshots” at specialized treatment centers. In this way, these “hybrid” trials, which are conducted partially at patients’ homes through digital tools, offer a clearer picture of the efficacy of new therapies. The bottom line: greater insight into patients’ lives can allow these organizations to design treatments that not only address the disease but align more seamlessly with patients’ behaviors and habits.
Not only are DCTs a competitive advantage for organizations seeking higher patient enrollment and engagement, DCTs can potentially use fewer resources while producing higher quality data and more effective drugs. Within pharma and biotech circles, there has been plenty of discussion about how organizations can bring their drugs to market faster, and DCTs appear to be one of the leading strategies.
What do DCTs look like from the patient perspective? Participating in a hybrid clinical trial can be as seamless as using one’s smartphone or wearable tracker like the Apple Watch to capture data and stream wirelessly to a doctor’s EHR. Ideally, patients can continue their daily lifestyles, acquire insights into their health data, and realize the benefits of the novel treatment. R&D organizations are still developing strategies to mix and match decentralized and centralized tools to align with study objectives in their protocols, but the future of this trend seems to be both imminent and inevitable. Many believe the data generated by wearable devices will revolutionize medicine, because they truly capture the patient experience and reduce the bias inherent in testing at a traditional site.
The tech giants have long been betting on this decentralized clinical trial trend to encroach into the healthcare space. In September 2018, Apple announced the Apple Watch 4’s updated ability to capture an ECG signal, bolstering the company’s strategy to become the digital touchpoint for patient data collection. Dozens of other tech companies, including Google, Fitbit and Amazon, are adopting similar strategies.
The decentralized approach has also been widely supported by the FDA. For example, the Clinical Trial Transformation Initiative (CTTI) is a public-private partnership co-founded in 2007 by Duke University and the FDA that has been working on a project that analyzes the challenges of decentralization through telemedicine and mobile healthcare providers. At the DPharm conference in Boston this past September, CTTI released new recommendations for organizations to overcome the legal, regulatory, and practical hurdles for planning and conducting decentralized clinical trials (DCTs).
How R&D Companies Can Innovate
All pharma and biotech organizations possess different strengths, weaknesses, and capabilities. For some organizations, the goal is to create as robust a portfolio as possible, while for others, the aim is to invest strictly in high-potential assets. Some organizations generate value on the discovery side of R&D, while others realize value with efficient processes and decision-making. All organizations need to be as adaptable as possible. The key to leveraging new trends is to constantly identify which actions yield the highest return on investment, considering both short-term and long-term initiatives. Is it better to strengthen core competencies or diversify a skill-set?
Before attempting a transformation, organizations should conduct assessments and have discussions about how these new trends fold into corporate strategy and whether not change is feasible. Strategy is an art. There is not set formula for deciding which path to take, but the first step is to be aware of the options to innovate and then approach transformation objectively with a clear vision of how it will truly benefit your organization.
Coauthor and contributions by Adam Kershner