$48 Billion Is Lost to Avoidable Experiment Expenditure Every Year
Pharma’s productivity challenges are widely known. Over the last decade, pharma R&D costs have been steadily increasing, while the rate at which drugs are approved has remained roughly constant. As of 2018, it cost approximately $2.17 billion (all figures USD) to bring a new drug to market, almost double the $1.18 billion cost reported in 2010.¹ In the same timespan, annual forecast peak sales were halved, from $816 million in 2010 to $407 million in 2018.¹ This has resulted in a decrease in returns of 8.2 percentage points, from 10.1% in 2010, to 1.9% in 2018.¹
Drug development isn’t only costing more, it’s also taking longer. The average project length increased from 9.7 years in the 1990s to 10-15 years through the 2000s and 2010s.²,³ Factors commonly cited for these trends include clinical trial inefficiencies, such as patient enrolment competition and more stringent selection criteria, and the industry’s shift into more difficult, high-risk, high-reward research areas like oncology.
But there’s another, lesser known culprit. Our data suggests that Avoidable Experiment Expenditure (AEE) is an overlooked source of unnecessary spend, significantly contributing to the increase in project length and fall in returns. Considering that ~42.9% of overall spend on drug development goes toward preclinical R&D, this issue deserves more attention.⁴ Avoidable Experiment Expenditure refers to all inefficiencies and productivity challenges in designing and carrying out preclinical experiments. Experiments are the foundation of preclinical research and development, however, irreproducibility rates in preclinical experiments exceed 50%, costing the industry nearly $48 billion annually.⁵
All of the news, delivered with full-text to your inbox. For professionals discovering, developing, and marketing biopharmaceutical drugs.
Inappropriate reagents account for $17 billion in Avoidable Experiment Expenditure annually
There are multiple factors that contribute to AEE, and are inherent challenges of any drug discovery effort. These include inappropriate reagents, poor experimental design, unreliable or variable protocols, lack of details in reporting, and lack of transparency in internal data. However, inappropriate or faulty reagents are one of the most significant individual contributors.
Over a third of AEE, or more than $17 billion, can be attributed to the ineffectiveness of biological reagents or reference materials.⁵ By resolving reagent-related AEE, life science organizations could potentially recoup $17 billion in unnecessary spend while streamlining operations, improving R&D efficiency, bringing assets to clinical trial faster, and generally accelerating pipeline progress.
Commonly used biological reagents, including antibodies, recombinant proteins, RNAi, and CRISPR, as well as model systems like cell lines, are sourced from biological origins, such as animals and patient samples. As a result, the performance of biological reagents can vary depending on the biological systems in which they are both created and used. Reagent selection is challenging—experimental context must be taken into account, data for which is scattered throughout many sources. This is exacerbated by unreliability in reagent data, as well as in reagents themselves; it is estimated up to 50% of reagents do not work as intended.⁶
In fact, over 36% of all unproductive experiments in preclinical R&D can be attributed to ineffective reagents.⁵ Current standard processes for selecting reagents are flawed and unreliable at best. However, the general perception is that the exploratory nature of preclinical R&D makes it inherently inefficient, thus the selection and use of inappropriate reagents is deemed an unavoidable and necessary part of the scientific process. In fact, however, reagent selection can be significantly improved with emerging technology.
Machine learning offers new hope for improved reagent selection and reduced Avoidable Experiment Expenditure
The inefficiencies in preclinical experiments have a direct and measurable impact on an organization’s productivity, resulting in longer research times, more spend on drug development, and increased downstream risk. However, with access to better information, improved reagent selection, more effective experimental protocols, and more informed experimental design, issues such as unproductive experiment rates and proportions of material waste could be reduced.
The first step to solving a problem is finding a way to quantify it. As of now, very little is being done to track productivity in preclinical experiments. We suggest it would be wise for the industry to align on standards to measure this, as well as the impacts of AEE, beginning with reagent-related AEE. Once we know the scope and scale of the problem, we can measure the impact of solutions that address it by helping scientists make more informed decisions.
The second step is to empower scientists with information to design more successful experiments. Data to guide decisions about experimental design, such as selecting reagents, exists but is buried in disparate sources of public and proprietary information. Scientists often spend days searching through literature for reagent references, then weeks to months validating multiple reagents for their experiments. Yet still, at least half of the reagents they purchase can fail them. It’s impossible for scientists to sort through and organize all of the data to inform their reagent selection manually.
Machine learning can enable both the quantification and reduction of reagent-related AEE. When trained to analyze scientific data as a human researcher would, machine learning algorithms can find patterns at scale that no human could identify. For quantification of reagent-related AEE, we can use the algorithms to analyze an organization’s reagent data to measure the utility and cost-effectiveness of reagent purchasing. For reduction of reagent-related AEE, we can use these algorithms to help scientists select the most effective reagents for their specific experiments in a fraction of the time it would normally take.
BenchSci is the only company in the world currently pursuing this avenue. Our discovery of AEE was informed by our work with over 40,000 scientists at more than 4,300 institutions, as well as 15 of the top 20 pharma companies. To learn more about Avoidable Experiment Expenditure, access the full whitepaper here.
References
1. “Ten years on, Measuring the return from pharmaceutical innovation 2019.” Deloitte.
2019.
2. Norman, Gail A. Van. “Drugs, Devices, and the FDA: Part 1: An Overview of Approval
Processes for Drugs.” JACC: Basic to Translational Science. Elsevier, April 25, 2016.
3. Pammolli, Fabio, Laura Magazzini, and Massimo Riccaboni. “The Productivity Crisis in
Pharmaceutical R&D.” Nature News. Nature Publishing Group, June 1, 2011.
4. DiMasi, Joseph A., Henry G. Grabowski, and Ronald W. Hansen. “Innovation in the
Pharmaceutical Industry: New Estimates of R&D Costs.” Journal of Health Economics. North-Holland, February 12, 2016.
5. Freedman, Leonard P., Iain M. Cockburn, and Timothy S. Simcoe. “The Economics of
Reproducibility in Preclinical Research.” PLOS Biology 13, no. 6 (September 2015).
*Freedman et al. report figures for the US only. We have extrapolated their data to reflect the fact that the US accounts for 58% of global pharmaceutical R&D. See:
6. Baker, Monya. “Reproducibility Crisis: Blame It on the Antibodies.” Nature News.
Nature Publishing Group. May 19, 2015.
https://www.abpi.org.uk/facts-and-figures/science-and-innovation/worldwide-pharmaceutical-company-rd-expenditure-by-country/
