I imagine that by this time the FlaskWorks unit would be second, third, or fourth generation, but at that time the idea of FlaskWorks units being distributed is one which could be accepted for it's use in many of the hospitals where patients are being operated on for tumors found to respond to the vaccine. Control of the disposable cassettes would assure that NWBO is properly paid for each use of the device, and of course the central computer would register each time every unit was operated. Every cassette would be tracked from it's creation until it's destruction after use.
I'm not suggesting that this will happen overnight, it won't, but if as many of us believe DCVax-L, and ultimately Direct are found to work on many cancers, and if the one FlaskWorks unit can make both of them, the shear volume of cancers being operated on will make having units distributed all over the world make production of the vaccine fare easier than having to transport human products to a limited number of production sites. Some production sites will still be utilized for many of the hospitals, etc. that don't have all the equipment needed to properly support the units, but for the many that do have the labs and all the rest that's needed, all will be done at the site where the patients tumor is removed and leukapheresis is done.
Gary, Here is an interesting viewpoint.
Distributed manufacturing models for ATMPs: can they work: do we understand the pitfalls? Cell & Gene Therapy Insights 2023; 9(2), 159–164 DOI: 10.18609/cgti.2023.024 PUBLISHED: 2 MARCH 2023
VIEWPOINT
Mark Lowdell
The field of advanced therapy medicines, advanced therapy medicinal products (ATMPs) to use European terminology, arose from the long history of hematopoietic stem cell transplants (HSCT) and adoptive immunotherapies invented by academics. The explosion of interest in these therapies began with Rosenberg’s earliest trials of IL-2 stimulated ‘lymphokine activated killer cells [1] and his use of tumor-infiltrating lymphocytes [2]. The proof of the ability of allogeneic lymphocytes came 5 years later, with Kolb’s demonstration of resolution of relapse of chronic myelogenous leukemia after T cell depleted allogeneic HSCT by infusion of small numbers of lymphocytes from the original HSCT donor [3].
In both of these examples and in hundreds of other similar trials, the cells used for treatment were manufactured near to the patient, sometimes a matter of meters away in the same facility. In Europe, it was not until the publication of the Medicines Directive in 2001 (2001–83-EC) that human cells were first included as medicines which required manufacture to the standards of Good Manufacturing Practice (GMP) within licensed facilities. The subsequent ATMP regulation in 2007 and the revised medicines directive in 2009 (2009-120-EC), plus parallel regulations around the world, cemented this position and, arguably, allowed for the commercialization of advanced therapy medicinal products (ATMPs).
However, the unique manufacturing challenges of both autologous and allogeneic ATMPs and the need for biological starting material from patients and/or donors in most cases, means that hospitals and academic centers often remain essential partners in the commercial delivery of these medicines. Autologous chimeric antigen receptor (CAR-T) cell medicines have been the ‘poster child’ of the field, with rapid progress to marketing authorizations and biologic license application as licensed medicines. However, these require contracts with hospitals to provide the starting materials, and often, with hospital ‘stem cell labs’ for the initial processing of the patient cells to a cryopreserved mononuclear cell suspension for shipping to the centralized manufacturing site for the final CAR-T drug product.
The costs and complexities of centralized manufacture of ATMPs, especially autologous, has led to interest in establishment of decentralized manufacturing models aimed at reducing costs and easing the supply chain problems. Recently, the United Kingdom’s Medicines and Healthcare Products Regulatory Agency (MHRA) has released a consultation on point of care manufacturing [4] which highlights the enthusiasm for regulatory changes to facilitate the most ambitious aspect of distributed manufacture.
Until last year, I was the Director and Qualified Person of the Centre for Cell, Gene & Tissue Therapeutics (CCGTT) at Royal Free Hospital & University College London (UCL), London. This GMP facility was established in 2001 and licensed to produce ATMPs for clinical trials and compassionate use. CCGTT has supported manufacture for academic and commercial clinical trials and has allowed UCL to be the fifth largest center for CAR-T trials in the world. It has never sought a commercial manufacturing authorization for marketed products. CCGTT is typical of the hospital-based GMP facility envisaged for distributed manufacture.
In 2013, I spoke about distributed manufacture at an ISCT conference and showed in Figure 1.
This envisaged two types of ATMP: those which are so patient-specific that they could never achieve a marketing authorization, such as allogeneic mesenchymal cells for treatment of acute graft-versus-host disease (GvHD); and those which could go through full clinical trials and proof of efficacy leading to marketing authorization application (MAA). I have always emphasized the importance of the protection of products with MAA from competition from similar products made under the European Union Hospital Exemption Scheme or UK Specials, where proof ofefficacy is not required.
Here I will share my opinions and concerns about the models for distributed manufacture of ATMPs. For the sake of brevity, I will assume that the challenges of distributed supply of starting materials is solved, although that too remains a significant challenge. I will also restrict the subject to ATMPs that will have obtained marketing authorization, since these are the ATMPs which have proven efficacy and, hence, the greatest burden of CMC control.
These are my opinions and not, necessarily, those of any of the companies for whom I work or advise.
Models of distributed manufacture
Hub & spoke where the spoke is owned by the holder of the marketing authorization
This is the easiest model to imagine and to operate. Here the manufacturing process is wholly owned by the pharma company and the spoke sites are controlled under a comparable or identical quality management system. There will be different manufacturing authorizations for each site but the processes can be identical and owned by the company. Although this is the most simple of the models, it is not without challenges. The fact that distributed manufacture is needed implies that the drug is licensed for use in many countries and thus, more than one regulatory domain. There are many examples of manufacturing reagents that are acceptable in one country, but which are not acceptable in a second. Most recently, I have come across a human albumen solution, which we use for ATMP manufacture, and the same supplier distributes different versions for European Union (EU) and United States (USA). Neither can be bought in both territories so, inevitably, the drug product will be made with different reagents in two countries. This means two supply lines, two validations, two procurement lines, etc, etc. None of this is difficult but it all adds cost. This seems a trivial issue but even in my limited experience, we have multiple reagents that we use in ATMP trials in the EU and the United Kingdom (UK), which US Food and Drug Administration remains unwilling to allow to be used for the same products made for the USA market.
Hub & spoke where the spoke is owned by a third-party company contracted to the holder of the marketing authorization
This is the typical scenario when a company uses a commercial contract development and manufacturing organization (CDMO) to manufacture their product. It has the limitations described above, but with the added complexity of use of a third-party quality management system and staff who are employed by the third party for manufacture. Each site of manufacture of a licensed medicine has to be named in the marketing authorization applications/biologic license application (MAA/BLA) and comparability of production has to be demonstrated. This includes all in-process quality control (QC) and release assays. In the European Union, each batch must be qualified person (QP) released, but the QP in this scenario is employed by the CDMO and the legal responsibility for the product rests with the holder of the manufacturing authorization. The contractual obligations of the third-party CDMO will require very high definition since the quality of the product falls under the CDMO, yet the reputation of the product remains with the holder of the MAA. This model already works for biotechnology products such as recombinant proteins, but those are batch manufactured and a failed batch can be remanufactured. The same is rarely true of autologous products and the reputational risk profile for the holder of the MAA is thus far greater.
Fully distributed manufacture where the product is manufactured at the hospital site:
This is certainly the most challenging model. Many academic hospitals across the world have embedded good manufacturing practice (GMP) facilities which manufacture ATMPs for clinical trials. The regulatory oversight required differs across geographical regions, however, it is fair to say that the level of regulatory control of facilities making ATMPs for clinical trial is leagues away from that imposed on commercial sites manufacturing licensed products with MAA/BLA. Every aspect of the GMP process is affected and the number of quality assurance (QA) and QC staff required is, in my experience, ten-fold greater as a minimum. These staff must be employed and managed even if the facility is not manufacturing the drug product, and the concept of dual manufacture of investigational ATMPs alongside licensed ones is a very great challenge for inspectorates. Alongside the challenges of manufacture to commercial standards are the parallel challenges of running a QC laboratory to the same standard; validated assays and equipment, back-up equipment and staff, secure storage of reagents and data, etc, etc, etc. I do not believe that the level of activity in any hospital for the manufacture of a specific licensed ATMP can justify the cost of obtaining and maintaining a commercial manufacturing authorization (MA). This has been recognized in the consultation document issued by UK Medicines and Healthcare Products Regulatory Agency (MHRA), where the commercial MA is held by the ‘hub’ and extended to each ‘spoke’. The details of how this can work and the levels of compliance needed at each spoke remain to be determined but, as ever, the Devil will be in the detail.
One of the highest risks associated with licensed drug manufacture is label control and it is difficult to imagine how this can be managed adequately with this level of distributed manufacture.
In the real world, there is also the challenge of how the contract between the holder of the MAA and the hospital is maintained. The hospital must agree to maintain the GMP facility and staff to meet the demands of the MAA holder, but the hospital’s first duty is to treat patients and resources are inevitably skewed to that purpose. Essential equipment breakdowns in hospital GMP suites are never treated as the priority they would be in a CDMO or pharmaceutical company. This puts patient-specific ATMP manufacture at risk and presents very high reputational risk for the holder of the MAA. If batches fail and patients die through lack of treatment, it is reputational suicide for a pharmaceutical company to sue the hospital responsible. In fact, in my experience, it is very unlikely that the hospital would accept that level of liability for a commercial medicine.
Fully distributed manufacture where the product is manufactured at the patient bedside - ‘black box’ manufacture’:
This is the dream of many ATMP developers and many more engineers. Closed system, automated manufacture on the basis of quality By design, with no QC release assays, and a ‘virtual’ QP appears to address all of the challenges I’ve presented above. This may come to pass but the ‘black box’ will be very large, and will still have to be managed under a complex quality system covering the QA of stock control, monitoring of stores, training of staff who load and unload the device, management of data trails, reporting of adverse events. The hospital will need to employ coordinators who can ensure that the patient-derived starting material is procured at the right time that the ‘black box’ and staff are available to manufacture.
In this scenario, under whose manufacturing authorization is the drug made and released? Who is liable for drug failures when such an event could be due to reagent control, staff error, equipment error, or simply the quality of the starting material that couldn’t be tested prior to manufacture in the ‘black box’ model?
Conclusions
Technologies are advancing in our field, which give some hope to models of distributed manufacture with the greater availability of closed manufacturing systems and semi-automation of processes. QC testing remains a challenge, not least, design and delivery of suitable potency assays. Digitization of batch manufacturing records is now available for ATMPs and has become affordable even for hospital-based facilities. This is an essential step for distributed manufacture of licensed ATMPs and, in some cases, can truly facilitate automated analysis of QC data and reduce the QP role to ‘release-by-exception’, which could allow distributed product release.
The manufacture and delivery of autologous ATMPs is challenging and will always be expensive. Distributed manufacture is inevitable but is not the panacea that it is often presented as being. Hybrid models of those described above are being developed and championed, but each will face the same issues I have highlighted and will need real-world business models to determine the actual savings that could be achieved. In 25 years of ATMP manufacture, I have learned that contractual issues are often the hardest to resolve, followed by staff retention, and facility management/maintenance. All of these only get greater as the manufacture gets more distributed.
References 1. Rosenberg SA, Lotze MT, Muul LM et al. Observations on the systemic administration of autologous Lymphokine-activated killer cells and recombinant IL-2 to patients with metastatic cancer. N. Eng. J. Med. 1985;313, 1485–1492. Crossref 2. Rosenberg SA, Spiess P, Lafreniere R. A new approach to the adoptive immunotherapy of cancer with tumor-infiltrating lymphocytes. Science 1986; 233, 1318–1321. Crossref 3. Kolb HJ, Mittermüller J, Clemm C et al. Donor leukocyte transfusions for treatment of recurrent chronic myelogenous leukemia in marrow transplant recipients. Blood 1990; 76, 2462–2465. Crossref 4. UK Government. Consultation outcome Point of Care Consultation. Medicines and Healthcare products Regulatory Agency (August 12 2021). Crossref
Affiliation
Mark W Lowdell Professor of Cell & Tissue Therapy, University College London, UK and CSO INmune Bio Inc, Boca Raton, Florida, USA
Authorship & conflict of interest
Contributions: The named author takes responsibility for the integrity of the work as a whole, and has given his approval for this version to be published. Acknowledgements, disclosures and potential conflicts of interest: Lowdell M is co-founder and CSO of INmuneBio Inc., an innate immunotherapy company developing ATMPs in clinical trials in UK, EU and US. He is also co-founder of Achilles Therapeutics Ltd, developing neo-antigen specific, autologous TIL therapies, and of Autolomous Ltd, a company providing software solutions to cope with the challenges of commercially delivering autologous immunotherapy products to thousands of patients. He is an advisor to multiple biotech companies and CDMOs and sits on the SAB of Avectas Ltd, Bio-Recell Ltd, Cytoseek Ltd and Sartorius GmbH.
Funding declaration: The author received no financial support for the research, authorship and/or publication of this article.
