Introduction

In Part I the rather wide spectrum of medical treatments referred to as Cell Therapy was introduced.  This diversity includes the proximate source of the cells, the cell type, any activation or engineering required, the mode of cell culture, and the therapeutic indication. Processes in the early days of adoptive transfer therapies, such as Rosenberg’s in-house produced TIL were described.  That outline presented current practices, general facility requirements, equipment currently employed, and examples of newly commercialized cell therapy (CT) equipment including those providing semi-automated and locally closed processing [ 1 ].

Impending Initiatives

The future promises a continuation of increased filings and significant approvals, as well as the expectation of exciting new equipment, process monitoring, production technologies, and facility design options. There are many evolving aspects of CT− such as the anticipation of moving from autologous approaches of rather “orphan drug” indications, to allogeneic products for more main-stream diseases [ 2 ].  Advances in vector production include next generation viral and non-viral vectors, and continuous producer cell lines.  Suppliers, sponsors, and dedicated consortia are providing rapid advances in many areas, for example, ARMI’s BioFabUSA program includes Tissue-Engineering and Regenerative Medicine (TERM) to advance the tissue manufacturing industry [ 3 ].

Cells and Sources

We’re all familiar with bone marrow transplants, but there are a number of other CTs gaining much attention.  The applications for CAR T and stem cell therapies are finally achieving functional success.  Here, advances in the treatment of retinal degeneration, cutaneous wounds, heart infarction, and skeletal muscle regeneration are showing promise.

T Cell Receptor (TCR) T cell approaches are demonstrating progress, but challenges remain, including concerns for graft versus host disease (GvHD).  Nevertheless, their overall safety profile is considered to be lower due to e.g., lower rates of cytokine release syndrome (CRS). They are in development for many reasons, including their ability to target proteins unique to specific solid tumors, and more. CAR technology is being applied to other immune cells, such as Natural Killer (NK) cells potentially providing such benefits as reduced CRS in autologous and GvHD in allogenic therapies, multiple mechanisms for activating cytotoxic activity, ease of targeting diverse antigens, and increased infiltration into solid tumors. Concerns such as transfection and expansion techniques are being overcome, yet obstacles do remain. Another alternative to T cells is CAR-carrying macrophages (CAR-M). In fact, the FDA has cleared an investigational new drug application, CT-0508, which targets solid tumors that overexpress HER2.

Finally, there are developing sources for cells to be employed in CT. A growing number of cell types can be efficiently harvested from patients for autologous approaches.  Cells from e.g., cadavers and supernumerary human embryos for allogeneic approaches raise legal and ethical issues, and therefore much work is being done to finalize human induced pluripotent stem cells (hiPSCs) as they have such therapeutic potential without the social concerns.

Process Monitoring

Essential to the commercialization of many technically successful cell therapies is the capability to provide timely information on critical quality attributes (CQAs) throughout the manufacturing process and QC release. Real-time process feedback is essential for process control to ensure manufacturing success, but in the CT arena such feedback for some values is often delayed due to lengthy testing times.

Initial values of importance include real-time results on such culture values as DO, pH, temperature and even viable cell density which can be provided by in-line via probes.  Such parameters as cell percent viability, glucose and lactate, and electrolytes can be reliably reported in near-real time by at line by automated analyzers.  However, such information as cell biomarker and media components must be supplied, with delay, by off-line analysis.

Because cells are the product, many MAM (multiple attribute methodology) approaches could not be applied directly to cell therapy, and more specialized (and often time consuming) analysis are supplied by flow cytometry, liquid chromatography, and immunochemistries.  But, MS-based proteomics and metabolomics and even new capillary electrophoresis and mass spectrometry (CE/MS) instrumentation is supplying new at line data [ 4, 5 ].

Many experimental and bioinformatics approaches have been developed to characterize the human cell repertoire, but applications can be limited by the poorly understood functional relevance of biochemical profiles to cell function. 

Digital

The pharmaceutical industry in general had been slow to incorporate many advanced data science applications for such reasons as the industry’s complexity and highly-regulated nature. There is also a desire for evidence that the emerging applications will be practical, powerful and robust. But, we have been seeing the gradual institution of such technologies as paperless operations and systems, digital intra- and inter-process communication and the use of programmable logical controllers (PLC)  in automated production. Industry 4.0 and digital biomanufacturing applications are now affecting every aspect of both process and facility development [ Table 5 ].  Those familiar with newer digital health technologies often think of such applications as  telemedicine, wearable technology, and computer aided diagnosis− but computational simulations, modeling, and machine learning are now emerging to model both cell therapy-based production processes and health-related outcomes. These advanced simulations can be replicated and tailored to particular flavors of CT and are affecting the very nature of a sponsor’s product development, operations, and quality systems. They are supporting such goals as process analysis and automation, reduction in personnel census and clinical decision support systems (CDSS).

So-called “smart manufacturing” will provide cell therapy with a digitized integration of the manufacturing process from supply chain management, to operations control, to final product track and trace.  Its enablers include machine learning (ML), the internet of things, real-time and integrated big data analytics; automated, autonomated and cyber-physical systems; augmented reality; cloud based ERP and edge computing; and the wide-spread application of digital twins in modeling both equipment and operations. As we have seen in the electronics and automobile industry, we anticipate the impending transformation of the bioproduction process, and in fact are seeing process development and facility design stakeholders moving in this direction [ 6 ].

3D computer aided drafting (3D CAD) and building information modeling (BIM) feeding virtual design and construction (VDC) and discrete-event simulation (DES) software support integrated multidisciplinary performance models, which are changing the face of facilities design and construction. These technologies digitally model buildings and suite structures,  process flow and building construction procedures, while incorporating site, process, materials, personnel flow, schedules, budgets, and other information to digitally plan out all aspects of a construction project.

These models support tracking, optimizing, and validating the construction progress while minimizing waste and validating designs. We are even seeing 4D planning with construction method dynamic simulation and schedule visual simulation in construction projects, alerting to e.g., time-related clashes in advance, instead of encountering them on the construction site.

Edge and Multicloud

We anticipate that EDGE (Enhanced Data for Global Evolution) analytics will provide rapid results from data sets resulting from local sensors and off-line analytics, as well as distally located facilities. They support improving production quality, detecting deteriorating performance, and reducing the risk of failure. For CT particularly, it can enable efficient shipping, logistics and production schedules between remote locations.

Multi-cloud is the future for any manufacturing, and especially for such information dependent activates as autologous CT. It involves exploiting multiple cloud computing and storage services in a single network architecture. One can envision a combination of software as a service (SaaS), platform as a service (PaaS), and infrastructure as a service (IaaS) models orchestrating clinical data, track and trace, production scheduling and control, as well as quality and regulatory compliance. It seems that the only remaining challenges here are data governance and security.

Artificial Intelligence

AI systems (including ML systems) promise to empower many facets of process design and operations.  In process development, beyond speed and efficiency, AI is enabling such novel advances as interpretation of the multiomic analysis of cell samples. AI empowered generative design algorithms produce many feasible solutions or outcomes from a set of performance objectives and constraints within a multi-dimensional design space. These designs can operate on input from very diverse aspects of the process. In cell therapy, they can include such input as material and equipment status, initial sample clinical source and operational performance distinctions, financial, regulatory, and quality constraints, as well as arrival time and schedule/capacity factors.

In operations AI empowered sensor-data analytics (including soft sensors) can detect patterns and deviation from usual process output.  AI-based in silico modeling of the process can predict future process values and product levels, supporting downstream operations and process control decisions. In cell-based operations, AI enabled machine vision can alert when in-process quality is lower than expected, triggering alerts to avoid excursions.

The diverse characteristics of individual patient cell samples creates process control challenges.  Currently, decisions regarding the response to cultures struggling in nominal conditions are made by individual process personnel, based upon their accumulated experience and tacit knowledge  This approach suffers from e.g., a lack of consistency shift-to-shift, as well as slow and incomplete distribution of wisdom gained in continued activities. AI holds the promise of providing higher levels of analysis while incorporating more monitored variables in a more consistent manner. It can also provide the potential of incorporating new information (and understandings) from any number of sources in making these recommendations [ 7 ].

Even in larger-lot allogeneic design, AI’s ability to discover patterns in lakes of disparate data, while eliminating redundant or unhelpful data, is supporting such goals as rapid and inexpensive quality control processes, as well as digital twin-based prediction and event control.  It can aid in maintenance of privacy and data security, track, and trace in incoming and outgoing shipments, and in the general provenance, transmission, authentication, and verification of operations data.

Automation, Semi-automation, and Robotics

In most current implementations, teams of highly trained workers spend weeks on each patient’s treatment, in highly classified suites, producing volumes of operational data in a costly process that can often only serve a thousand doses a year.  Automation is the answer, and solutions previously seen in the food and electronics arenas are becoming incorporated into biomanufacturing in general, and specific implementations can particularly influence both process development (PD) and operations suite activity.

Initial processes employed many open-culture manipulations and manual handling of samples. These were prone to variations in such environmental parameters as temperature, duration of exposure to harmful physiologic conditions, and variable hydrodynamic forces.  Such variation contributed to inter-batch and inter-operator variability.  Automation can remove the subjective and fluctuating performance of even a highly skilled staff, with precise robotic movements and accurate timings. It can improve both inter-operator and inter-batch comparability with a plant, or across globally located sites.

Beyond, process variability, SOP breaches, and the possibility of cross-sample contamination, manual manipulations posed sources of  particulates and exogenous, adventitious contaminants. We are now seeing stand-alone units providing automation of even multiple process steps [ 1 ], and these functionally closed and automated processing chambers provide much improvement in these areas.  However,  CT-specific factors, such as diversity in patient’s cells, can create workflow steps that are not always integrated.  The existing ‘single box’ approaches to automation provide many values but can present significant limitations in the selection of responses to disturbance variables of the normal process. The more autonomous robotic production being developed can not only increase the reliability and efficiency of a process, but provide the flexibility not provided by current fixed flow-path devises.

Newer automation possibilities support the regulatory goal of safe and efficacious therapies in better-controlling sterility and particulates, reducing inconsistencies and errors, improving record keeping and retention, as well as increasing productivity and performance consistency. It’s true that manual processes provide immediate flexibility and limited training costs, but the long-term value of  biomanufacturing automation is undeniable.

Robotic handling provides continual maintenance of optimal cell-processing conditions over long periods. The smaller footprint of automated systems increases facility footprint productivity.  The indefatigable nature of the precise and accurate performance of robots are clear− but what is less appreciated are the benefits afforded by the increased flexibility and reduction in training provided by robotics.  While programmers are required to introduce a new process, ultimately the number of training personnel, manuals and their governance is greatly reduced.

We see pressures to support such disparate goals as product-based versus platform approaches, standardized designs, and customizable facilities. Advanced design and engineering can in fact address all concurrently, because a single-use material process design can in fact be initially highly product-specific, yet include features making it very customizable. We are now even seeing the development of many scalable factory-in-a-box products for CT applications [ 8 ].  Such developments are enabling end-to-end automation while creating highly individualized cell therapies, including CAR X therapies. However, purveyors of stem cell therapies complain that significant automation for their process seem to be years away [ 9 ].

A single robot can traverse a 3D space that would otherwise require many mechanical components, resulting in significant space savings. A single machine can reliably produce many different products with minimal changeover time/risk between batches. Recipe-driven programming minimizes components required and increased simplicity.  Instead of physical adjustments and tool changes, robotic production lines can be tuned and  guided automatically by embedded sensors and machine vision systems. Robotic production lines also reduce the chance of operator setup and cleaning errors.

Operations Summary

Innovations providing flexibility in suite and facility design can be generally organized into developments in technology, equipment, and facilities. As more entities are approved, sponsors and regulators alike are feeling more comfortable with the idea of adopting new scale-up or -out process technologies and equipment [ 1 ] as well as application of newer approaches to suite design [ 10 ].  This includes not only designing for where the process technology is today, but “future-proofing” it to support process and product evolution, as new equipment, process optimizations and scale considerations emerge.  Regulators are now recognizing this need for such process development flexibility and accepting the need for continued process design even after trials begin.

New approaches being addressed by CT sponsors include those occurring in processes [ Table 4a ],  production suites [ Table 4b ] and in the facility or plant [ Table 4c ].  Because of its power and novelty, digitalization or “4.0” developments are addressed separately [ Table 5 ].

Table 4. Facility features supporting novel CT entities in various platforms

• Process
  • Process development flexibility supports even continued process design even after starting trials
  • Greater automation to improve efficiency and reduce cycle time and personnel requirements
  • Systematizing cryogenic transport, accessioning and provenance of client-specific material
  • Including flexibility and process adaptability in hospital, clinic, and  pharma facilility plans
  • New functionally including closed and aseptic automated equipment and methods
  • Improved (simpler and more efficient) methods of vector production
•  Suite
  • Advanced Micro Methods (real time event monitoring) and CCIT
  • Ballroom suites supporting nested environment classification
  • Commercialization of process-specific single-use systems
  • Continuous processes and continued process verification
  • Advanced connectors including aseptic fittings, and NTT
  • Environmental nesting-supportive systems including modular and docking-integrated cell culture isolators
  • Flexible, modality agnostic suites supporting process enhancement and future scale-out, including
• Plant
  • Efficient support of newer “single-box” cell processing systems
  • Prefabricated and modular facility structure and equipment
  • 4.0 systems design / methods including Deep Learning AI

Table 5. Digitally enabled (“4.0) spaces support

• Networked suites for process research, development, and tech transfer
• Warehouse and manufacturing automation and even autonomation
• Controlled warehousing for incoming/outgoing products
• AI augmented digital twin modeling alerts and control
• Centralized PD, materials supply, operation control
• Packaging, labeling, and storage of vialed products
• Flexible and robotic lines with gloveless isolators
• Automation, autonomation, articulated robotics
• AI empowerment machine vision inspection
• Monitoring, analytics, computing capability
• Improved data governance and traceability
• Improved data security and transmission
• Developing deep learning applications
• Storage and inventory of SU materials
• Real-time assessment and/or release
• Discrete Event Simulation modeling
• Advanced data-heavy MRP and ERP

The logistics and flow of autologous therapies remain more complex than others, yet innovations are supporting the mantra of “simplicity” to be established. Here, commercially available pre-sterilized, single-use and ready-to-use components are providing even process-specific systems that support increased efficiency and safety, while reducing personnel requirements [ 1 ].

The number and diversity of CT therapeutic modalities has been described. But, as in many emerging technologies, for even generally similar therapeutic approaches there is currently significant diversity of materials and operations among researchers. This diversity also exists within the manufacturing processes of therapy sponsors.  It is anticipated that the emergence of optimal and proven manufacturing techniques and process flows will provide regulatory ease and more efficient process design and filings.  Relatedly, the current diversity in hospital-based materials, preparatory activities, and transport is predicted to gradually focus to recognized best practices, which will not only improve performance, but incoming sample consistency and reporting. It is hoped that we may soon move from an epoch of creativity and innovation (supporting advances in design) to more uniformity and standardization (supporting interchangeability and robust procurement).

Facilities

There are a few distinct visions for the future of both cell therapy process and manufacturing facility design in general.  Currently, we still see stick-built facilities employing upstream equipment which was designed for other purposes being adopted to CT production, and handled in standard BSCs housed in familiar grade B suites.  This might continue to be popular, or we could see the same equipment employed more often in facilities constructed from prefabricated units of more modular or podular design employing modular isolators technology in class C or D suites [ 11, 12 ].

Generic aseptic containment isolators and especially the newer, integrated cell-culture isolators can enable, classical, robotic, or entirely closed CT operations.  Their contiguous, multiple modules are configurable and expandable.  Some provide multiple docking stations connecting other equipment and allow for multiple means of sanitization. They empower reduced suite classification, and support greater automation in either a unit operation or across an entire process.  Finally, they can reduce the number of trained processes required and provide greater process safety [ 13, 14 ].

Some designed-for-purpose prefabricated, modular, controlled-environment chambers now combine principles of process intensification, automation, and straight-through processing in self-contained ISO 14644-1 classified operating modules [ 15]. While current offerings are closed for only part of the process, fully closed and automated  “factory-in-a-box” approaches may provide both production and aliquoting capacity that can drastically reduce facility footprint, capital, personnel demand, and operational expenditures.

Conclusion

Cell therapies continue an exciting evolution in treatment modalities, cell types, engineering and culture, and therapeutic indication.  Manufacturing facilities are incorporating more 4.0, factory-in-a-box and flexible facility features.  The future for CT is promising, dynamic and exciting.

About the Author

William Whitford, Life Science Strategic Solutions Leader, DPS Group

Bill Whitford is the Life Science Strategic Solutions Leader for DPS Group, a global consulting, engineering and construction management company. He assists in developing creative strategies supporting the manufacturing of both classical and innovative biotherapeutics. Bill began his carrier as an R&D Leader, commercializing over 40 distinct products supporting biomedicine and biomanufacturing. Most recently Bill has been a thought leader identifying burgeoning biomedical products and processes. Bill has published over 300 articles, book chapters, and patents in bioproduction; is a regular presenter at international conventions; and is an instructor in biomanufacturing.