From Cadillacs to Camrys: Parallels from the automobile industry to CGTx manufacturing

Date: 24-July-2023

Introduction

On 14th September 1990, the first approved gene therapy was administered to four-year old Ashanthi De Silva for the treatment of severe combined immunodeficiency (SCID) by Dr. William French Anderson and colleagues at the National Institutes of Health (NIH). This was achieved by transfusing engineered T-cells carrying corrected gene copies for adenosine deaminase (ADA). This marked the dawn of genetic medicines, perhaps going towards fulfilling the promises of the Human Genome Project. Fast forward to 2023, there is now more than 1,500 cell, gene, and nucleic acid therapeutic (CGNT) trials and 27 approved CGNTs [1]. Yet these life changing medicines are not always accessible to patients given the enormous price tag associated. The cheapest CGNT, Yescarta treatment for large B-cell lymphoma, comes in at $373,000 (USD) [2], while Hemegenix for the treatment of hemophilia B has a price tag of $3,500,000 (USD) [3]. These costs are recognized to be a severe limitation, and for some geographies have led to the denial of payment for these therapies as the costs outweigh the benefits.

Here we will discuss some of the reasons these treatments are so expensive, how other industries have reduced costs, and how we can apply that knowledge to the manufacture of CGNT.

Why is CGNT so expensive?

Some of the main drivers for the relatively high price of CGNT are the cost of goods sold (COGS) in manufacturing cell and gene therapies and the cost of clinical trials [4]. Unlike conventional small molecules or monoclonal antibodies (mAb), cell and gene therapies remain challenging to manufacture in a standardized fashion given the immense complexity of the modalities be it an adeno-associated virus (AAV) or chimeric antigenic receptor (CAR) T-cell. Let’s focus on AAV.

By one estimate, each dose of AAV costs around $100,000 (USD) to manufacture [5]. Several factors contribute to this high cost including the high number of empty or defective capsids associated with recombinant AAV production and the high number of analytical tests. Almost 70-90% of AAV viral particles produced do not contain the transgene. Various publications and developers claim higher numbers. However, many of these studies count defective, partially filled capsids as “full”, an obvious inflationary view of the performance of the method. During purification, these unwanted empty AAV particles are removed. As a result, significantly larger batches of viruses must be made. By one estimate, we would need a billion liters of cell culture and almost 400 Olympic-size swimming pools worth of media to treat a thousand patients, making the current methods in manufacturing CGNTs not economically feasible for most patients.

In comparison to other types of drugs CGNTs require many more analytical tests for release. For example, a typical mAb drug product requires 10 to 15 assays to determine its quality and safety. A CGNT may require upwards of 35 assays to determine the overall quality of the viral vector or cell. These assays can represent between 10% to 40% of COGS.

How have other industries dramatically reduced costs?

To understand methods to improve CGNT manufacture, it helps to look to analogs of manufacturing that deal with highly complex systems where failure rates in manufacturing drive up costs enormously. The automotive industry is a prime example! In the evolution of the automobile manufacturing industry, mass availability preceded mass production of cars with uniformly high quality, a difference that for obvious reasons doesn’t apply to medicines of any kind. However, it is important to recall that quality standards in the mid-50s and 60s were not as high as they are today.

The transition of the automotive industry towards mass production of high-quality products offers methodologies and frameworks that can be applied to the production of genetic medicines. Just as Toyota took inspiration from supermarkets and created The Toyota Way, the genetic medicine biomanufacturing industry must take inspiration wherever we can find it and leverage lean manufacturing processes to drive down the cost of production.

In this series of posts, we will focus on the manufacturing of Genetic Medicines, starting with viral vectors. We are not dogmatic in our stance, but rather would welcome discussion. Innovation is a team sport, and in this case just like the Toyota Way has fundamentally transformed automobile (and other) manufacturing, we believe Genetic Medicine Manufacturing will be transformed.

Here is a brief refresher on The Toyota Production System (TPS). It focuses on designing out excessive burden (muri), inconsistencies (mura) and waste elimination (muda) [6]. The TPS does this by empowering workers to stop the line when a problem is detected, standardizing the process, and eliminating waste through continuous improvement (kaizen). Using TPS, Toyota has achieved high levels of efficiency, quality, and customer satisfaction, and become one of the top automobile manufacturers. In short, the TPS works because it is focused on the customer, and it is a flexible system that can be adapted to different industries and different products. If you are looking for ways to improve your manufacturing process, the TPS is a great place to start.

How can we apply TPS to reduce the cost and improve the overall quality of CGNT?

In addition to low yields and large numbers of analytical tests CGNT utilizes specialized processes and small batch sizes. These drive-up costs and increase failure rates [7]. Currently we are still operating in learning mode in the industry; yes, therapies are getting to patients, but the costs are too high to make therapies available broadly and this frankly jeopardizes the future of the industry. If we as the industry want to move from rare and ultra-rare disease treatments delivered in the richest nations in the world (and sometimes not even there!), then we must aggressively improve our manufacturing capabilities.

As it currently stands, most AAV processes are designed from the ground up using unique components and process variations such that they satisfy the needs of the therapy. There is momentum to standardize components and processes, but to create consistency, we must move towards implementing a robust and flexible manufacturing approach that reduces cost / time of tech transfer to manufacturing. Manufacturing of AAVs requires the interplay of two very different disciplines – biology and engineering. Defining conditions for consistent yield and quality of viral vectors produced is challenging. All departments within a company need to embrace the engineering mindset with a healthy respect for COGS.

This starts with having a scalable process to have it Right First Time (RFT) and this often begins with the cell culture system of choice. Most gene therapy developers currently favour the use of HEK293 cells for expression of AAV and these cells can come in two flavours – adherent and suspension. Adherent cell lines are more common and have been used successfully in numerous applications with low barriers to begin work due to ease of culturing, as evidenced by T75 flasks being a staple in many labs. However, adherent cell lines are hard to scale up. Conversely, the use of a suspension cell line allows for better scalability and cost savings. Suspension cells can also grow at higher densities reducing the number of vessels needed.

Another muda to address, waste of making defective product, is currently inherent in the manufacturing process of recombinant AAV. Improving the percentage of viral particles containing a functional transgene is critical. Currently, there are emerging technologies with promising results that have been shown to shift the full to empty ratio of AAV particles during cell culture. This will greatly reduce the volume of culture needed and number of purification steps, further reducing muda in the manufacturing process. We will explore these technologies in a later article focused on cell line engineering for better AAV production.

As previously mentioned, to release CGNT upwards of 35 analytical tests may be required. One approach to reducing both the number of tests and the quantity of material required for testing, is adopting multi-attribute methods (MAM) for analytics to reduce the total number of assays needed; another one is the increase in performance in key assays, specifically characterizing performance of the assay as predictive performance on Critical Quality Attributes. And of course, by now you could have guessed it: most quality control assays tend to also be time-consuming, adding to the muda of wasted time.  As ICH (The International Council for Harmonization of Technical Requirements for Pharmaceuticals for Human Use) guidelines are being updated such as Q5A, we can expect an increased adoption of rapid molecular techniques such as nucleic acid and next-generation sequencing methods. These methods also tend to be MAM, reducing both time, and number of assays, thus allowing for further cost savings. For more on this, see our post on MAM for viral vectors.

Current models of CGNT are specific to the therapy and the manufacturer making it difficult to scale up or adapt to a new transgene. Using the processes of TPS/lean manufacturing we can improve CGNT manufacture and reduce cost. Outsourcing of AAV manufacturing can be beneficial; one can leverage proprietary manufacturing platforms starting from off-the-shelf plasmids, standardized cells line from single colony picking to well established analytics. This would drastically reduce the time to clinic as the timeline for process development can be reduced significantly. By working closely with a manufacturing partner, it will be possible to begin with the end in mind and produce proof-of-concept viral vectors to be used in clinical trials that are comparable with the final commercial product; thus, reducing mura along the way and de-risking the path to commercialization.

A closing thought: Capacity alone doesn’t solve the problem!

We as an industry must exponentially – not just incrementally – innovate to make CGNTs more accessible to patients.

References

[1] M. Cooper, “How Do Cell & Gene Therapy Requirements Differ Between FDA & EMA,” 24 January 2023. [Online]. Available: https://www.cellandgene.com/doc/how-do-cell-gene-therapy-requirements-differ-between-fda-ema-0001.

[2] A. Weintraub, “Fierce Pharma,” 18 October 2017. [Online]. Available: https://www.fiercepharma.com/regulatory/kite-and-gilead-join-novartis-red-hot-car-t-market-early-axi-cel-approval.

[3] M. Naddaf, “Scientific American,” 9 December 2022. [Online]. Available: https://www.scientificamerican.com/article/3-5-million-hemophilia-gene-therapy-is-worlds-most-expensive-drug/

[4] E. Harris, “Cell and Gene,” 18 June 2018. [Online]. Available: https://www.cellandgene.com/doc/breaking-down-pricing-of-cell-gene-therapies-0001.

[5] J. Plieth, “Evaluate Vantage,” 25 June 2018. [Online]. Available: https://www.evaluate.com/vantage/articles/analysis/vantage-points/100000-problem-gene-therapy-companies-would-rather-not.

[6] “Toyota Production System,” [Online]. Available: https://global.toyota/en/company/vision-and-philosophy/production-system/.

[7] A. Laboratories, “Access Data, FDA,” December 2002. [Online]. Available: https://www.accessdata.fda.gov/drugsatfda_docs/label/2002/adalabb123102lb.htm.

[8] S. Zoratto, “Molecular weight determination of adeno-associate virus serotype 8 virus-like particle either carrying or lacking genome via native nES gas-phase electrophoretic molecular mobility analysis and nESI QRTOF mass spectrometry,” Journal of Mass Spectrometry, vol. 56, no. 11, p. e4786, 2021.

[9] R.-M. Comisel, “Lentiviral vector bioprocess economics for cell and gene therapy commercialization,” Biochemical Engineering Journal, vol. 167, p. 107868, 2021.