Earlier Decisions, Better Processes Required For Automating Autologous CAR T Manufacturing

By Erin Harris, Editor-In-Chief, Cell & Gene
Follow Me On Twitter @ErinHarris_1

Autologous CAR T manufacturing automation promises greater consistency, lower costs, and faster turnaround times. But as we discussed during our Cell & Gene Live, Automating Autologous CAR T Manufacturing, Lessons from Mayo Clinic, getting those benefits requires more than investing in equipment.

My conversation with Mayo Clinic’s Cynthia Wilkins, Ph.D., M.B.A., Senior Director of Process Development and Assistant Professor of Regenerative Medicine; Snigdha Rai, M.S., Senior Director of Biomanufacturing; and Anthony DeFusco, M.S., Senior Director of Quality Control for the Center for Regenerative Biotherapeutics highlighted a recurring challenge for cell therapy developers. Manufacturing decisions made early can shape process consistency, QC, tech transfer, and commercial viability years later. Indeed, our experts made clear that automation works best when developers understand their processes, account for patient variability, and plan for future changes before they become expensive problems.

Establish a Process Before Automating It

Automation can reduce operator variability, standardize manufacturing practices, and limit opportunities for contamination. It can also streamline training, batch record review and certain facility requirements, particularly when using closed systems. But automation cannot compensate for a poorly understood process.

Dr. Wilkins emphasized the importance of starting with the intended end product and assessing the complexity of the manufacturing workflow before selecting an automated platform. Developers need to understand their critical quality attributes (CQAs) and critical process parameters (CPPs), then evaluate which steps are suitable for automation.

Rai reinforced that point from the manufacturing floor. A process needs to be sufficiently defined, characterized, and supported by appropriate controls before automation can consistently improve its performance.

That may sound straightforward, but early-stage developers often work with evolving protocols, novel constructs, and limited resources. Automating a process before understanding its essential requirements can lock teams into decisions that become difficult to reverse.

They shared that developers should bring process development into the conversation early, assess risks before committing to equipment, and determine whether a modular or integrated platform best supports the program’s needs. Automating individual steps early can also help teams identify problems without committing immediately to a complete manufacturing system.

Account for Patient Variability Before Optimizing the Workflow

One of the most important points raised during our discussion was that automation cannot eliminate the inherent variability of autologous starting material.

Patient medical history, disease state, prior treatments, apheresis collection, and material handling can all influence the cells entering manufacturing. Whether material is processed fresh or frozen can also affect the process and final product.

Standardizing equipment and operator practices helps, but it does not make every patient’s cells behave identically.

Dr. Wilkins and Rai stressed the importance of testing disease-derived starting material during process development and technology transfer. Rai shared that relying on healthy donors alone can obscure important differences that only become apparent when working with material from patients.

This is particularly relevant when developers are preparing for clinical trials. A process that performs well with healthy donor material may encounter unexpected challenges with the intended patient population.

Developers should establish relevant CQAs early, understand how disease biology and prior treatment affect cell health, and use those findings to inform process parameters and release strategies. The goal is to understand the range of variability the process must accommodate, rather than assume that a standardized workflow will produce identical results every time.

Treat Quality Control as Part of the Automation Strategy

Automation can improve consistency and potentially reduce certain manual activities, but it does not remove the need to demonstrate product quality.

DeFusco emphasized that developers still need appropriate analytical methods, specifications, and a clear understanding of the product's CQAs. Teams must demonstrate identity, potency, safety, and other relevant quality attributes, regardless of whether manufacturing is automated.

There may be opportunities to reduce in-process sampling when developers have strong process understanding, reliable onboard sensors and data demonstrating that the control strategy provides an equivalent or better level of assurance. That reduction, however, must be supported by evidence and risk assessment.

The panel also highlighted the importance of coordinating manufacturing improvements with QC capabilities. Shortening a production run does not necessarily accelerate product release if testing remains a bottleneck.

For developers, this means designing manufacturing and analytical strategies together. Identify where testing is essential, determine which measurements could be performed in line, and establish how results will support process decisions. Faster manufacturing only delivers its full benefit when quality testing and release workflows can keep pace.

Make Manufacturing Data Useful, Not Just Available

Predictive manufacturing offers an opportunity to anticipate problems and adjust processes before a batch fails to perform as expected. But useful predictions depend on understanding the biology behind the data.

Dr. Wilkins described the potential for disease-specific algorithms that account for patient variability and identify patterns associated with cell health and product performance. Measurements of metabolic activity, including glucose and lactate levels, may help reveal how cells behave during culture.

However, developers need to establish which indicators correlate with product quality for their particular process. A sensor reading is not automatically a reliable predictor of potency or final product performance.

Rai also raised an often-overlooked data challenge. Experienced operators may recognize subtle changes in cells, media handling, or process behavior that never make it into formal batch records. When teams automate a workflow, that practical knowledge can disappear unless they deliberately capture it.

This is an important opportunity for manufacturers. Batch records and digital systems should capture relevant observations alongside instrument readings. Teams can then review those observations for trends and use them to improve process controls.

DeFusco added that automated equipment and connected sensors need appropriate qualification, validation, and data-integrity controls. Developers should be able to access raw data, trace sample identity, and reconstruct what happened during a manufacturing run. Vendors need to support those expectations, including integration with quality systems and electronic batch records.

Build Flexibility into Platform and Technology Transfer Decisions

Automation can make technology transfer more efficient, but it can also introduce rigidity when a process depends on equipment that is unavailable at another site.

Dr. Wilkins described Mayo Clinic’s interest in modular platforms that can accommodate different therapies and manufacturing needs. Modular approaches can allow teams to automate selected operations, adapt components, and build capabilities without committing every program to the same integrated system. Partnerships can also provide access to equipment or expertise that an organization does not maintain internally.

Rai noted that using the same automated platform across manufacturing sites can simplify training and process transfer. When platforms differ, developers must assess whether equipment, process parameters and documentation differences affect comparability.

DeFusco emphasized that analytical transfer is a particularly important consideration. Differences in instruments, reagents, reference materials, or software can affect results, even when the manufacturing process itself appears unchanged.

Developers should establish a strong baseline using data from multiple runs, retain suitable reference material where possible, and develop potency assays early. When feasible, running the same starting material at two sites through a split-batch approach can help assess comparability. When that is impractical, historical data from the originating site becomes especially valuable.

The goal is not simply to meet release specifications after a transfer. Teams need to understand whether the new process continues to perform within the established range of expected quality.

Prioritize Investments that Solve Real Manufacturing Problems

For developers working with limited budgets, the panel offered a useful reminder that the most sophisticated technology is not necessarily the best first investment.

Rai identified automating the wash and concentration step as a potentially high-value opportunity. This operation can be open during manual manufacturing, creating opportunities for operator variability and handling risks. Automating and enclosing it can improve consistency in cell recovery and final formulation.

Dr. Wilkins encouraged developers to start with the end in mind, engage process development early, evaluate disease-derived material, and develop analytical methods alongside manufacturing. DeFusco highlighted another important purchasing consideration. Organizations should retain ownership of their raw data, be able to export it in a readable format, and confirm that automation software can integrate with systems such as laboratory information management systems and electronic batch records.

Far more insight was shared on this topic, as our discussion was laced with detail and nuance. Be sure to tune in to our full Cell & Gene Live conversation on demand to hear what else our expert panelists from Mayo Clinic had to say.