Industrial ATMP Manufacturing: Digitization’s Role in Aseptic Manufacturing for ATMPs

However, these new opportunities also provide new challenges. Unlike larger-scale biotech and pharmaceutical production processes, ATMP manufacturing involves small, highly personalized batches, diverse product types, and labor-intensive manual processes. The unique characteristics of personalized therapies impose significant demands on manufacturing procedures, systems, and skilled personnel.

This article aims to summarize the specific challenges in executing and controlling aseptic ATMP processes and describe how digital solutions can support manufacturers to overcome them. Building on an optimized digital ecosystem that allows for control of the process and digital data availability, we explore how data management and digital twin concepts can achieve real process control and optimization.

ATMP Manufacturing Challenges

For ATMP manufacturing, the underlying groundwork that has been developed for other biotech processes can be used. However, aseptic ATMP processes provide some specific challenges. These include the following examples (see Figure 1).

Figure 1: ATMP process challenges.

Production Planning

Production planning for ATMP manufacturing may require more effort, as there are more considerations. Smaller batch sizes in ATMP manufacturing lead to more frequent changeover in the production processes and additional process control requirements. Personalized therapy processes increase the process diversity, requiring individual product specifications. In personalized therapy processes, the risk of losing a batch often equals the risk of losing the patient, leading to the necessity of increased safety procedures.

Compartmentalized Manufacturing

With compartmentalized manufacturing, segregated production spaces for small batch manufacturing require an additional process control layer. Further, individual processing of the cells in isolators or biosafety cabinets (BSCs) increases the level of complexity because data from particle sensors are received for every batch within a room over short periods of time.

Process Complexity

The individualized nature of ATMPs means that there is an increase in process complexity, and for many reasons. Patient-specific manufacturing often requires a batch-specific sampling plan. Also, the vast majority of ATMP processes as of today include open system operations, leading to increased risk of contamination. More frequent manual handling of the cell suspension and samples with operator intervention increases the risk of process errors and requires higher training levels and increased operator monitoring. The chain of custody (COC) events in ATMP can be defined as intervention steps, leading to a large number of events that need to be tracked, controlled, and monitored.

Cross Contamination Prevention Checkpoints

Personalized ATMP manufacturing processes require rigid control to prevent cross contamination and ensure patient safety. This process has multiple checkpoints designed to maintain the integrity of materials and data throughout the whole supply chain process (see Figure 2). Next, we summarize the key checkpoints in safeguarding personalized therapies based on the industry best practice by BioPhorum.3

Together, these checkpoints form a robust framework and serve as a strong foundation for preventing cross contamination during ATMP manufacturing processes. By following these checkpoints, manufactures can reach the highest standards of quality and safety in the rapidly evolving field of personalized therapies.

Figure 2: Materials and data integrity checkpoints.3

Collection of Starting Materials

The process starts with the careful collection of starting materials from the patient. Verifying patient identity and obtaining informed consent are fundamental steps. This information is logged into hospital and sponsor systems with accuracy to create a secure chain of identity (COI). Correctly, clearly, and properly labeled containers with unique identifiers, such as a COC ID and donation ID number, ensure that materials are assuredly associated with the correct patient. To maintain material integrity, the collected samples are stored under optimal conditions, such as cold storage, immediately after collection.

Packaging and Handoff to Couriers

It is critical that materials are packaged correctly, clearly, and properly during the transfer of materials from the collection site to couriers. This involves maintaining the “chain of condition” by temperature-controlled packaging and securing sealing to prevent human interferences and tampering. Accurate transfer records need to be logged in all relevant systems, such as hospital, courier, and sponsor platforms. This is to detect and eliminate any discrepancies. Cross-verification of shipment information during transfer helps ensure that materials are handled by authorized personnel and the package is delivered intact.

Transportation and Receipt at Manufacturing Facilities

GPS and environmental monitors are advanced tracking systems that are essential to ensure supervision and maintain oversight during the transit of materials. Critical conditions like temperature and shock levels must be monitored, logged, and recorded by these systems. The materials are rigorously examined to verify their authenticity, identity, and integrity once they arrive at the manufacturing site. Before processing starts, this involves confirming that products fulfill quality criteria by examining and verifying labels, shipment records, and environmental data.

Manufacturing Processes

At the production stage, strict protocols must be followed to prevent cross contamination. Before the manufacturing process begins, multiple independent checks, including barcode scanning, manual reviews, and visual inspections are used to verify and confirm the identity of the patient material. The appropriate segregation between different patient materials can be achieved by assigning equipment and designating specific production work areas or facilities for processing this material. These listed precautions can significantly reduce the risk of accidental mixing or contamination during the manufacturing process.

Final Product Labeling and Dispatch

Once the production stage is complete, accurate data from various systems are collected and combined to generate final product labels. Next, these labels undergo independent verification prior to being affixed to ensure accuracy of labeling. Products are then rapidly cryopreserved under strict timelines to maintain stability of the product. Lastly, during dispatch to the treatment center, critical records must be updated. This step ensures traceability and accountability of the final therapeutical product during transportation to and receival at the treatment center.

Administration at Treatment Center

Upon arrival at the treatment center, the final verification step takes place, which is cross-checking patient identity and details of the produced therapy. At least two identifiers of the product are matched before administration to the patient to prevent cross contamination. This critical step relies heavily on the successful execution of all prior checkpoints discussed in this section. The administration process is documented thoroughly in hospital and sponsor systems to provide a complete traceability record.

Systematic Quality Assurance

Throughout all discussed stages of the supply chain process, systematic involvement of quality assurance is needed to comply with data integrity principles. Regular and routine audits and cross-checks of records across systems make sure that all processes meet regulatory standards and requirements. Here, privacy laws, such as the General Data Protection Regulation and Healthcare Insurance Portability and Accountability Act, must be strictly enforced, as they ensure that patient-identifiable data is very well protected from being compromised to unauthorized personnel across the whole supply chain process.

State-of-the-Art Control of Aseptic ATMP Processes

State-of-the art-cell therapy manufacturing approaches are moving toward fully automated and roboticized platforms that seek to eliminate human intervention. By removing the human factor, such platforms provide repeatability of operations to enhance process efficiencies and improved aseptic technique, without the risk of operator error. This approach also brings with it the ability to process multiple autologous batches in parallel in one system, with full digital tracking of each batch.

Individual cell processors should be highly automated modules that can easily be configured to interface with supervisory software systems for process visibility and GMP data management. In respect to data management, state-of-the-art aseptic ATMP manufacturing is supported by a digital ecosystem, as described next, which supports the collection and analysis of data against specifications.

A Digital Ecosystem: Compartmentation

ATMP processes are made up of several disparate process steps with varying levels of automation in each specific step. This can make material and batch traceability arduous when navigating a complex mix of manual and automated equipment when using paper-based batch records. Compartmentation, or treating each unit operation as its own module of digital manufacturing, offers a practical strategy to assess the digital manufacturing capabilities and successfully implement digital technology in or around each part of the process.

Compartmentation consists of breaking the process into individual compartments. These can be executed either manually by a person or automatically by a piece of equipment. Each compartment’s data capture should be grouped into the following categories: manual, results, and automatic data capture. Manual data capture is for data from manually executed process steps in an isolator and for data from equipment with no reporting or network connectivity. Reports data capture is for data from a report generated by an island of automation. This can either be procedurally or automatically transferred to supervisory software systems. Automatic data capture is for networked equipment that automatically sends data to supervisory software systems, and the data can be read only or fully read/write.

Multiple data capture categories can exist for a specific compartment. A good example would be an isolator that automatically produces real-time environmental data, but the process executed within it is completely manual. To put all this together, let’s look at an example of the ATMP manufacturing line and assign different compartments to it (see Figure 3).

Figure 3: ATMP manufacturing line.

Supervisory Software Systems

Breaking the process into each of these data categories allows picking which supervisory system will integrate best into the process. Two main supervisory software systems are used for the coordination of people, processes, and equipment in ATMP manufacturing: the manufacturing execution system (MES) and the process control system (PCS). Both systems have different capabilities and can integrate with each other. Implementing these supervisory systems will lead to better material and batch traceability by:

• Electronically storing all material consumption and GMP-critical batch data from manufacturing
• Automatically creating electronic batch records with review by exception
• Linking manufacturing data with laboratory and business data for a digital COC

MES

The MES provides the bridge between business and laboratory systems—e.g., enterprise resource planning (ERP), the laboratory information management (LIM) system—and the production floor. At its base level, the MES will electronically capture all data on equipment and operations in manual compartments. This is done through equipment logbooks and digital work instructions. These sources of data are used to compile a fully electronic batch record (EBR) for the manufacturing campaign that can be linked into a digital chain of custody for a patient.

This creates a paperless environment that streamlines activities like end-of-batch line clearance. The line clearance activities are digitally logged and recorded with the completion of each section of the EBR. This information then becomes available in real time from the digital environment. The MES is also leveraged to great effect in compartments with automatically captured data. A digitally mature MES implementation will gather data directly from the PCS, coordinating automatic compartments. This data will be used to compile an EBR with no additional input from the operator.

PCS

The PCS provides equipment coordination and continuous data collection in a process. The PCS is used where visibility and centralized GMP data collection is required across multiple pieces of equipment or systems. More complex requirements include centralized recipe management for an entire process train and plug-and-produce rapid integration of vendor equipment and instruments.

Holistic Process Data Models

One strategy to overcome the high level of complexity and meet strict control requirements in aseptic ATMP manufacturing is to develop powerful digital data models. These digital data models represent the manufacturing process with all components and are developed and optimized using real-world data. Data sources for aseptic manufacturing of ATMP include sensors and equipment, software to control the execution process, environment monitoring systems, and manual data collection. Software considerations include planning and material data from the ERP; process data and an audit trail from the MES/EBR; process data for the PCS; quality control and sample data from the LIMS; and deviations and corrective action and preventive action (CAPA) from the quality management system (QMS).

Data that needs to be collected falls into five categories: environmental monitoring, process data, operator data, product data, and equipment data. Environmental monitoring includes air quality metrics, surface contamination control, and temperature and humidity. Process data includes input material specifications, the operational times of open and closed steps, and sample plan and in-process control (IPC) results. Operator data includes gowning protocol and compliance, critical interventions, training documentation, and operator contamination control. Product data includes sterility and bioburden analysis, endotoxin control, residuals analysis, and the sterility assurance level. Equipment data includes equipment cleaning and sterilization, pressure differentials in cleanroom areas, filter integrity tests, and isolator and glove integrity tests.

Because manual data collection creates more effort and is more error-prone, it is highly beneficial to cover as many components as possible with digital solutions and to develop an integrated digital ecosystem in which data can be received and contextualized automatically and stored in a central data hub. However, the current ATMP landscape does not reflect this optimized scenario. Many of the equipment solutions that are used by industry do not provide smooth integration capabilities, and most manufacturing companies do not have a complete integrated digital ecosystem in place.

These limitations should not lead to abandoning the idea of a digital data model, as even a partial implementation can provide the resource for a very valuable outcome. Instead, it is recommended to mitigate important data flow via manual interaction and build up a holistic data model that is optimized for the manufacturer specifics and where, with development, more manual interactions are replaced by automized data flow.

Traditional data models reflect parts of the process to facilitate analysis of input parameters on output results. For example, the operation time of an open process step can be analyzed for dependent contamination probability to define process specifications. However, the more complex the actual process is, the less this kind of data analysis provides the necessary results to control it. The operation time of an open process step influences the final contamination risk, and this risk is even higher if the operation time of the remaining process is longer. Residuals in the cell suspension can influence risk of contamination in both directions, depending on the solution components. Thus, the actual process is a complex network of independent, dependent, and interdependent parameters that influence each other.

Holistic data models take the next step by reflecting the actual process as it is. They integrate data over the complete manufacturing process and take all parameters into account. The more complex the aseptic manufacturing process, the more powerful the holistic data model is, making ATMP the perfect example. Real control of the aseptic process can only be achieved by understanding the complex network of parameters that define the results of an aseptic ATMP process. In the implementation process, holistic data models usually start with the complete manufacturing process for key parameters and are built up step by step by improving and stabilizing the model to allow for the addition of parameters.

Holistic data models allow for aseptic process optimization by identifying and mitigating risks to increase safety and efficiency. Furthermore, they allow for advanced predictive process development. By analyzing historical data, manufacturers can predict the effects of planned process changes on the aseptic process parameters or predict when equipment is likely to fail or require maintenance, preventing unplanned downtime and ensuring a seamless manufacturing process. The complexity of ATMP manufacturing can be overcome by enabling manufacturers to simulate, predict, and visualize outcomes before they occur in the physical world.

Digital Twins

Digital twins are a virtual representation of the manufacturing process and all its components to allow for real-time interaction with the digital data model. Over the last 10 years, digital twin technology has emerged as an advanced tool in manufacturing industries, offering innovation and optimization to manufacturing processes. This is accomplished through real-time data simulation and actionable feedback on processes and performance. Lessons learned from industries such as aerospace and advanced manufacturing demonstrate the potential of digital twins to integrate real-time data with external factors, driving advancements in product development and operational efficiency.

In ATMP manufacturing, high investment costs and lengthy transition periods for training and process optimization are common, and the precise setup and operation of cell processing equipment are critical to maintaining product quality, regulatory compliance, and patient safety. In the context of ATMP manufacturing processes and facilities, digital twins offer a compelling solution. They may serve as a virtual representation of various operational tasks and enable the simulation in support of process optimization, facility design, training, onboarding, and technology transfer processes more effectively during or before the operational phase.

In the aseptic ATMP manufacturing process, digital twins can be particularly valuable. They offer a real-time view of the conditions in a cleanroom or other critical areas, allowing for continuous monitoring of environmental parameters such as temperature, humidity, and particle count, where alerts can be triggered based on specifications or other critical process variables. By direct integration, the model can be fed in real time with equipment data, allowing for the analysis of parameters like pH or dissolved oxygen, to provide real-time analysis. Should the system detect a deviation, corrective measures can be initiated instantly, preventing potential contamination or failure.

ATMP Manufacturing Roles and Digital Twins

In ATMP manufacturing, digital twin technology offers significant potential, but its value and application differ greatly depending on the perspective of its users. This section focuses on several typical roles in ATMP manufacturing and outlines how each could use and benefit from digital twins.

For Scientists and Operators:

Digital twins provide a dynamic, real-time virtual replica of manufacturing equipment and systems. This enables operators to train, monitor, and interact with processes independently of the actual cleanroom or facility. Digital twins also help build processes and SOPs dynamically. This technology enhances decision-making by offering intuitive insights into equipment setup, performance, processes and workflow, and potential bottlenecks.

Digital twins provide a way to experiment with continuous improvement without directly altering the official processes or equipment. Currently, most of the professional development happens during hands-on operations, and the training remains paper based, relying on static information. Data interpretation will become more important, as digital literacy to interact with digital twins increases.

For Quality Assurance Personnel:

Digital twins provide the ability to visualize and simulate processes with greater precision than traditional paper-based methods, making them invaluable for maintaining regulatory compliance, ensuring full traceability (a digital thread), and safeguarding product integrity. They enhance quality assurance by tracking deviations, validating processes, and enabling predictive quality monitoring. In general, real-time quality processes become integral rather than retrospective, allowing for continuous monitoring and improvement across various stages. This shift implies that the roles within Quality Assurance will increasingly require data analysis skills to interpret and act on real-time data effectively.

For Supply Chain Staff:

A digital twin serves as a critical tool for improving logistics and inventory management by improving data management, ordering from master material records, and providing full traceability of the supply chain. Integrating real-time data helps synchronize supply chain activities, anticipate disruptions or shortages, and optimize resource allocation in personnel and warehousing. This optimization will require skills in modeling, automation, industrial engineering, and optimization.

For IT Professionals:

Digital twins demand a shift in IT expertise. Beyond managing networks and devices, IT teams must acquire deep knowledge of interconnected facility processes, cybersecurity, and data streams. They also must be capable of building and managing digital infrastructure. Their role evolves into ensuring that data infrastructure supports the integration and functionality of digital twins across diverse systems.

Discrete Event Simulation Software

Use of discrete event simulation (DES) software can produce a true digital representation of a new facility as a 3D simulation that can be used to create the best possible layout, before ground has been broken. Development of this type of digital twin can provide the following key benefits. First, it’s possible to run detailed process simulations to accurately determine the equipment requirements based on the required throughput. This helps ensure that conservative equipment estimates do not result in the installation of greater than required equipment quantities. This approach also ensures that valuable space within production suites is not taken up with unnecessary equipment.

The digital twin can be used to map material and personnel movements through the facility to ensure the best possible facility layout, removing pinch points that could result in undesirable crossovers. Running the simulation will allow tracking operator movements with respect to the manufacturing operations based on the specific equipment locations. A high concentration of operators in a specific area within a manufacturing suite can lead to an increased risk of mix-ups, batch contamination, and potentially even contamination of the cleanroom environment through creation of “hot spot” areas. The information provided by the simulation can be analyzed to determine whether adjustments to the layout need to be made to prevent this scenario occurring.

Building Information Management Tools

A true 3D representation of the new facility can be developed using building information management tools, such as the Autodesk Construction Cloud (ACC). A walkthrough of the digital 3D model can allow visualization of the areas around each piece of equipment to conduct maintenance and operability studies. These assessments can ensure that sufficient working area is provided around each piece of equipment to minimize the risk of cross contamination through overlap of operations being carried out at different pieces of equipment in parallel. An augmented reality (AR) interface with the 3D model can also be implemented, allowing the user a more immersive experience when reviewing the digital facility compared to simply reviewing the model on a screen.

Construction and Operation

The “Factory of the Future” will also need to have a significant focus on sustainability, which applies equally to the construction and the operation of a new facility. The use of digital tools such as DES modeling and ACC help ensure the most efficient use of space. As well as optimizing the layout to ensure sufficient operational space around equipment, these tools can also be used to ensure that no space is wasted and therefore no area is unnecessarily oversized. The larger the operational space, the greater the construction material requirements to build and the greater the energy consumption to operate. Efficient sizing of all the manufacturing areas will help ensure that stored carbon and operational carbon emissions are kept to the necessary minimum.

Sterility Analysis

In current ATMP manufacturing, sterility analysis is often only executed after the end of manufacturing. Some processes even allow for patient treatment on conditional release, when not all release test results are available, as required by short holding times of the product. This makes thorough process control even more important, where digital twins for the holistic aseptic process can drastically improve patient safety. By using advanced machine learning models, digital twins allow for the prediction of product quality before receiving the release test results based on real-time data. This proactive approach to quality assurance ensures that products meet the required safety and efficacy standards before they reach the patient.

As with personalized ATMP manufacturing, the variability of the starting material leads to variability of the process, and operator decisions play an important role. Digital twins can support the operators with these decisions, providing a data-driven support system, which leads to improved outcomes but also reduction of the emotional burden, considering that their decisions have such a direct impact on patients’ lives. Taking the predicted rest of the process as well as interdependent parameters into account, the digital twin can facilitate root cause analysis and suggest the best course of action, i.e., based on the results of an IPC analysis.

Impact of Digital Twins

Digital data models and digital twins allow for real-time analysis of data within a holistic model, facilitate root cause analysis, suggest mitigation actions, and predict their outcome. This helps improve process optimization and process control to enhance patients’ safety in aseptic ATMP manufacturing. Digital twins can advance the quality and compliance with respect to the human element of the precision processes prevalent in ATMP manufacturing.

These advancements not only improve therapeutic outcomes but also contribute to the broader goal of making advanced therapies more accessible, scalable, and impactful for patients worldwide. This section explores the potential of digital twins in ATMP manufacturing, with a focus on end users and optimization of standard operating procedures (SOPs), and the commissioning of new facilities to improve or accelerate training and knowledge transfer process before use of the facility.

Interactive SOP Development

Digital twin technology can greatly enhance traditional paper-based SOPs development using a virtual (digital) representation (model) of the equipment or process. The digital replica of the equipment or process/operation can be used to display instructions, providing greater clarity during complex stepwise operations, improving training and reducing the risk of human error in the real world. Another tool that can be generated from the digital twin is AR.

AR is often used to visualize the digital twin as an overlay onto equipment or processes in the real world seen through a set of lenses or goggles. This overlay can be used to create interactive SOPs, incorporating visual workflows and dynamic updates. Changes to process parameters and equipment configurations can be instantly updated in the digital model, ensuring users have the most current information. The automation of SOP updates can mitigate the risk of outdated procedures, which is a common challenge in ATMP manufacturing.

Virtual Training in Real Scenarios

Interactive SOPs can improve the traditional “read and understand” approach with a more effective “do and understand” model. Operators are no longer limited to learning procedures through passive reading. Instead, they engage in hands-on, task-based learning that closely aligns with real-world operations. By aligning interactive SOPs with physical equipment in a classroom or lab environment, operators get training from contextualized instructions.

This merger of physical and digital technology can improve task accuracy and consistency with real-world operations. Virtual simulations can be run to practice setting up, operating, and responding to deviations during processing. This reduces reliance on physical cleanroom facilities for training and accelerates skill development. Digital twins can provide a framework for monitoring task accuracy using data-driven metrics to support targeted training and performance improvement where it is needed most.

Regulatory and Quality Assurance Support

Digital twins can have a positive impact on quality management and regulatory compliance by integrating real-time process data from operations with advanced analytics software to optimize workflows and minimize errors. Through the integration of LIMS or MES systems, digital twins provide real-time, audit-ready logs of operator interactions, creating an audit trail that aligns with the US FDA and European Medicines Agency guidelines. EBRs are generated automatically, capturing task completion and setup accuracy, ensuring traceability, and streamlining documentation processes during audits.

Conclusion

Although aseptic manufacturing has been optimized over decades, aseptic ATMP manufacturing provides specific challenges due to high complexity of the process, small batch sizes, and high frequency of manual intervention. The digital ecosystem can be used to orchestrate and streamline the process execution. Moreover, a sophisticated approach to data management enables manufacturers to gain actual control over the process by building powerful data hubs that allow for holistic process data models, which can be used directly for process characterization or, as a next step, as a template for digital twins.

Digital twin applications in aseptic ATMP manufacturing include real-time understanding of process changes, suggestion and analysis of mitigation actions upon events, and the possibility to support the human component as the most critical ATMP process parameter, taking process control to another level. The benefits of digital twins are clear: the greater the integration of data and processes, the greater the impact. Roles and use cases will evolve, and many are already well-defined and are being used in practical ways. In a rapidly advancing digital world, traditional project approaches must adapt. Conventional methods will be replaced with new ones. The journey of digitalization is an ongoing, iterative process that will require an organization-wide approach with stakeholders from IT, engineering, operations, and quality assurance to ensure a successful outcome.

To conclude, imagine Florence, Italy, at the height of its power—a city banking for Europe’s royalty and fueling the Renaissance. In its center stood the Florence Cathedral (Santa Maria del Fiore), unfinished for 140 years. Planned to be revolutionary, its greatest challenge was the dome. No one knew how to build such a massive structure without collapse, and existing materials couldn’t support its ambitious diameter. The cathedral remained roofless for over a century, waiting for a solution. Italian architect Filippo Brunelleschi is the one to ultimately deliver the breakthrough: a step-by-step method in which the dome was constructed with bricks in a spiral pattern, where each layer strengthened the next. His ingenious design, using a herringbone structure, remains one of the greatest architectural feats of the Renaissance.

The pharmaceutical industry is already world class, but digitalization will be its Florence dome—a seemingly impossible feat with conventional methods. Through a step-by-step approach and the power of imagination, breakthroughs will emerge, enabling cures beyond our current comprehension.