The Global Bioproduction Summit held in San Diego California from February 5-6, 2018 detailed the most pressing issues in biologics manufacturing. It featured 35 speakers from several leading drug development firms. A dominant theme at this conference was drug makers’ focus on reducing the time to market for new drugs.
Maximizing time on the market by reducing drug development time
Dr. Rakesh-Dixit, VP at MedImmune started the conference with an overview on the changing state of competition in the biologics drug market. More biologics drugs are being approved than any other time in the last 10 years. This means intensified competition between drugs that treat the same diseases. For example, more than 50 drugs are in development for some orphan diseases. The first one to market wins everything.
This intensified competition is not just for orphans drugs. Gone are the days when a new drug could count on patent protection alone to protect its market position. Biosimilars and biobetters are poised to cut into the sales of the industries biggest drugs even before the end of patent protection.
Thus the industry is re-focusing its efforts to bring new drugs to market faster. But this is no easy task. Industry leaders are rethinking conventional drug development processes, utilizing single-use technologies to add speed, efficiency, and flexibility to process development and production, and utilizing machine learning and data science to aid decision-making.
Rethinking development processes
David Wood, Professor of Chemical and Biomolecular Engineering at The Ohio State University, remarked that the amount of revenue from one month of sales from a blockbuster drug can exceed the salaries of everyone who worked on that drug. As such, focusing on cost efficiency for product manufacturing at the expense of a faster launch represents a significant loss in revenue and patient impact (material costs are less than 5% of the price). Dr. Wood advocated for a standardization of unit operations with known cost and efficiency to avoid the time it takes to optimize a process.
Josh Grieco from Genentech illustrated how his team has fundamentally re-thought their Chemistry Manufacturing Controls (CMC) development cycle. There were no tricks—just a singular focus on defining and optimizing manufacturing processes early, including (i) integrating first-in-human (FIH) and product and process qualification (PPD) into clinical manufacturing campaigns, rather waiting until after phase III pivotal trials; and (ii) prioritizing achievement of Breakthrough Therapy Designation (BTD) for accelerated FDA approval.
Single-use technologies reduce cost and add flexibility
Drug manufacturers are realizing cost savings through the use of single-use bioprocess technologies. These cost savings come from three areas: (i) reducing sunk costs associated with traditional stainless steel plant designs, (ii) eliminating the need for built-in sterilization equipment (needed to prevent contamination and health risks), (iii) and capacity expansion through faster changeovers to new products. In short, single-use technologies enhance flexibility and reduce operational overhead. This is a fundamental re-think on how to produce biologics drugs especially within manufacturing organizations that support multiple products.
Adam Goldstein from Genentech walked us through an analysis to determine the impact of single-use technologies. They use a vendor-supplied modeling application, populated with their historical operational data, to evaluate the cost tradeoffs between traditional stainless steel factories and facilities using only single-use technologies. The results show that mixed stainless steel / single-use equipment may be the best solution depending on the type of products supported.
Robert Baffi, EVP at BioMarin described the development of their new manufacturing facility in Novato, California to support their first Gene Therapy drug for hemophilia A. This plant was designed from the start for production flexibility and for mitigating contamination risk (which can have disastrous ramifications to smaller-sized drug makers). It makes significant use of single-use technologies to allow capacity to be ramped up or down easily. It is also flexible enough for multiple products to be integrated into production. This design allowed them to eliminate many of the large, fixed assets tanks, cleaning systems, purification beds, etc. They were able to complete the facility 11 months ahead of schedule.
Machine learning and data science take center stage
To harness the new-found flexibility of single-use technologies, and realize the benefits of compressed CMC practices, companies are turning to process modeling and machine learning to drive more agile decision-making. These methods have delivered new insights on bioprocess operations resulting in fewer development iterations, and more precise control over critical process parameters during development and manufacturing operations.
Thomas Jostock, Head of Cell Line Development for Novartis showed how his lab has reduced clone screening time by over 40%. They analyzed historical results of thousands of clone screens from multiple cell line programs and identified a common set of genes associated with positive cases. They used this information to shorten time-consuming high throughput screening cycles and by pre-selecting cell lines based on genetic traits.
Dr. Neel Sengupta from BD Life Sciences presented his success at troubleshooting cell culture challenges for their commercial partners using a machine learning algorithm. He was able to identify the impact that cell culture media components have on fermentation performance by using a series of variable selection algorithms. He then used this information to devise a suitable remediation plan.
Deisy Corredor from Pfenex introduced their scale up/down process model for their bacterial protein expression platform. Using data from ~200 fermentations, Pfenex narrowed down the most important drivers for fermentation control as being agitation power (motor power / volume) and oxygen transfer rate. Pfenex can reliably deploy their protein expression fermentation platform from very small volumes (2L) to large volumes (1500 L) through tight control of these variables. Their platform has been so greatly enhanced through using this process model that they can take customer fermentations from 0.5 g/L/hr to 15 g/L/hr with right-first-time performance replication.
One common challenge for these methods, however, is the amount of time and effort required to conduct these analyses. Though the mathematical and statistical methods are readily available, the collection and formatting of the underlying data sets for analysis remains laborious and often a blocker.
Towards the future
We're in the midst of a significant shift in the biologics drug industry. Winner-take-all orphan drug opportunities, advancing biosimilars, and rapid emergence of biobetters are intensifying competition. For the first time, a drug’s utility and market dominance may wane before patent expiration. Developing drugs faster has become the driving force in biopharmaceutical R&D, and manufacturing needs to be flexible, focus on smaller batch sizes, and support fast changeovers within multi-product facilities—all without compromising quality and GMP practices.
It's clear that data science is a lynchpin to this strategic shift—it’s necessary to make data-intensive decisions on a compressed time scale. Data analysis and process modeling have already impacted most of the efforts highlighted at the Global Bioproduction Summit. Yet all have struggled with the time and effort needed to collect the necessary data. We (Riffyn) have heard from many bioprocess organizations that it can take 2 to 3 months to collect data for a single batch. Data itself has become a new bottleneck. And that is why we created the Riffyn SDE—to give you integrated, analysis-ready process data in seconds.
 Designating an Orphan Product: Drugs and Biological Products,
 New Drugs at FDA: CDER’s New Molecular Entities and New Therapeutic Biological Products, http://www.fda.gov/Drugs/DevelopmentApprovalProcess/DrugInnovation/default.htm, Downloaded April 2016.
 EvaluatePharma®, Orphan Drug Report 2015, 3rd Edition – October 2015, http://www.evaluategroup.com/orphandrug2015, Downloaded April 2016.