Getting a therapy to work is only half the battle. Developers must also establish whether a candidate can be manufactured reliably, consistently, and at the scale required to meet clinical demand.
Compared with many small-molecule therapeutics, advanced therapies present distinct manufacturing challenges, making it essential to understand the factors that influence product quality throughout development. For example, cell and gene therapies can involve sophisticated manufacturing processes, patient-specific production, and short shelf lives. mRNA therapeutics and other complex biologics have their own “quirks”, from maintaining RNA integrity and controlling product-related impurities to understanding how raw materials and process conditions influence the final product.
The US Food and Drug Administration (FDA) has highlighted several of these difficulties, including product complexity, patient-specific manufacturing, limited manufacturing runs, and the need for extensive product characterization and analytical testing.
As Jessica Dunn, PhD, associate director of the Office of Compliance and Biologics, explained during an FDA webinar on chemistry, manufacturing, and controls (CMC) flexibilities in cell and gene therapy manufacture, short product shelf lives can also leave manufacturers with a narrow window between production and administration.
For Dr. Aron Gyorgypal, a postdoctoral research fellow at Harvard Medical School and Massachusetts General Hospital, these considerations should influence therapeutic development much earlier than many developers might expect.
Gyorgypal completed his PhD in chemical and biochemical engineering at Rutgers University, where he studied antibody glycosylation in the context of bioprocessing. He now applies this expertise to research in immunology, consults for biopharmaceutical companies developing biosimilars, and is a member of Technology Networks’ Scientific Advisory Board.
His view is that manufacturability should not be treated as a problem to solve once a therapeutic candidate has been selected. Instead, it should help determine which candidates are worth the hard graft in the first place.
The manufacturing challenge starts earlier than you may think
“Personally, I do not think of manufacturability as a development activity, but rather as a selection criterion,” said Gyorgypal. “By the time you have chosen your candidate, you have already made most of the decisions that determine whether it can be made, and candidate selection is the last point at which changing your mind is cheap.”
Leaving these considerations until later does not necessarily mean a candidate will fail. But developers may find that their molecule takes much longer to bring to market because manufacturing “mishaps” only begin to surface during technology transfer.
“Unfortunately, manufacturing problems are slow to walk back once a program is committed,” Gyorgypal explained.
The antibody field has already experienced this challenge, prompting developers to bring developability screening into the discovery stage. As Gyorgypal explained: “You can see the result in the data, where approved antibodies look measurably different, biophysically, from the ones that stalled.”
He believes that the same thinking now needs to be applied to newer therapeutic modalities. “When we think about this outside of monoclonal antibodies, surprisingly, nobody has done that work for cell and gene therapies yet,” he noted.
In Gyorgypal’s opinion, the questions developers should be asking are relatively straightforward: “Which attributes of this product matter? Can I measure them yet? Is anything in my supply chain single-sourced or loosely specified? And what does a dose cost at the scale this indication genuinely requires, rather than the scale my bench happens to run at?”
These questions can help identify potential problems before a program becomes too difficult or expensive to change.
Building in manufacturability from the beginning
One of the most significant challenges often arises when developers move from a laboratory-scale process to one that operates under good manufacturing practice (GMP) conditions.
“[W]hat makes that transition tedious is that the decisions causing the trouble were taken much earlier,” said Gyorgypal.
He uses the production of viral vectors as an example. Producing the same construct in two different host systems does not necessarily result in the same product with a different yield. The resulting vectors can also differ in potency and impurity profile, with the effects varying between serotypes.
“And the effect is serotype-specific, so you cannot reason your way to it; somebody must measure it,” he said. “That is entirely foreseeable work. It must happen before the platform is locked, not after.”
The mRNA field has its own version of this problem. Gyorgypal explains how an impurity in a lipid raw material can form adducts on RNA, affecting its stability over time. Whether developers detect this can depend on the analytical method they have chosen. “… one method may show the damage, and another may be completely blind to it. A raw material specification and an assay choice, both made very early on and both easy to treat as routine, predetermined a stability failure that surfaced long afterward.”
The FDA also emphasizes the importance of building an effective manufacturing control strategy throughout development. Kimberly Schultz, director of Division 2 in the FDA’s Office of Gene Therapy, explained that, while manufacturing controls are important at every stage, the implementation of current good manufacturing practices (cGMPs) can be tailored to the phase of development.
She added that the FDA encourages “sponsors to use the information gained during early development to implement an effective process control strategy which will ultimately facilitate consistent performance of the commercial manufacturing process.”
It’s important to understand which variables dictate product quality, rather than trying to micromanage every step of the process.
Accounting for variability and adapting strategically
Biological variability adds another layer of complexity, especially in cases where the starting material comes from an individual patient. Gyorgypal separates variability into two areas: differences in what enters the process and those arising from the process itself.
“For CAR T[-cell manufacturing], with an autologous product, you do not control the input because the input is a patient, so the answer is not a tighter specification on something you do not govern. It is a wider operating window and more measurements. And the input variability that matters is often not the variable you would guess,” he explained.
He highlights a case study published in Blood Cancer Journal that examined CAR T-cell manufacturing data. In that study, the more obvious markers measured at apheresis were not the factors that predicted manufacturing failure: “What predicted it was a therapy the patient had received earlier—the chemotherapy bendamustine—sometimes months before anyone was thinking about cell therapy at all, which tells you something about how early these considerations could really begin.”
Gyorgypal described this as a cautionary example of how far back developers may need to look when considering the factors that influence manufacturing outcomes.
Variability also affects more established manufacturing. In his words, “a more everyday version of the same problem is raw material variability in antibody manufacturing.” Cell culture media, for example, may be less defined than developers assume. He uses trace metal content as an example, “which can drift from lot to lot, influencing cell metabolism and final product quality.”
Even when process conditions appear stable, the product itself can continue to change. For example, continuous antibody manufacturing, where product characteristics can drift despite process parameters being held constant.
“Antibody glycosylation is one of the quality parameters known to do this,” noted Gyorgypal. Glycosylation refers to the addition of sugar or glycan molecules to a protein or lipid, which can influence its characteristics and function.
“I have seen changes in glycan patterns over weeks, and without intervention (even at steady state), glycan profiles have fallen out of specification,” he said. “Steady state in the process is not steady state in the product!”
This understanding also becomes important when manufacturing processes need to change as a therapy progresses through development. Up-scaling production, or changes to equipment or process conditions, can affect product characteristics, meaning developers must determine if the resulting product from the latest run is sufficiently comparable to the one made previously.
Schultz emphasized the FDA’s risk-based stance on CMC modifications: every manufacturing change requires an assessment that accounts for study timing and clinical impact, but for investigational drugs, “limited comparability data may be sufficient to support implementation of low-risk change.”
Gyorgypal agreed with this approach but also highlighted the financial and practical implications of timing. “The same process changes costs very little before a pivotal trial and an enormous amount after one, so the initiative should be to make your significant changes earlier rather than later,” he said. “Although it also requires a high level of process knowledge and familiarity with your product.”
He uses the development of the enzyme replacement therapy Myozyme® (alglucosidase alfa) to demonstrate the point: “When Genzyme increased production scale from 160 to 2,000 liters, differences in glycosylation were sufficient for the FDA to require a separate biologics license. The larger-scale product had already been approved by European regulators, but entered the US market under a new name— Lumizyme®.”
Gyorgypal also notes the importance of analytical methods in assessing advanced therapies: “You can only demonstrate comparability for attributes you are able to measure, so your analytical capability sets a ceiling on how much process change your program can absorb.”
He recommends qualifying methods early, retaining reference material and enough batches from each development stage for comparison, and deciding in advance which product attributes will determine comparability.
Applying established principles to newer therapies
Although manufacturing platforms differ between therapeutic modalities, Gyorgypal believes that some of the underlying approaches used in established bioprocessing can be applied more broadly.
“What transfers most readily is the methodology, not a plug-and-play manufacturing platform,” he noted. These approaches include quality by design, designed experiments, fit-for-purpose scale-down models and analytical methods that link process conditions to product quality. “The lesson is to use those tools early, while there is still room to change the product and the process,” Gyorgypal advised.
“But I would resist treating newer modalities as one manufacturing problem,” he continued. “Viral vectors share a recognizable cell culture and purification architecture with antibodies, although the products impose different constraints. With mRNA, raw material quality, RNA integrity, purification, and formulation become central; it is not simply a matter of increasing titer or reactor volume.”
The distinctions are also clear in CAR T-cell therapy manufacturing. Autologous products require patient-specific batches to be produced, meaning manufacturers must scale out production, “whereas allogeneic products create an opportunity to scale up multi-patient batches, provided expansion preserves cell quality,” Gyorgypal said.
Even mRNA-based CAR-T therapies do not follow one manufacturing model. Ex vivo approaches still require cells to be produced outside of the body, while in vivo approaches shift the manufactured product towards an RNA delivery formulation.
Focus on the principles, not the platforms
The FDA has also highlighted opportunities to improve manufacturing efficiency through analytical innovation. In particular, Schultz encouraged manufacturers to consider technologies that can provide faster detection during lot release. Alternative analytical methods, where appropriately justified, may offer more efficient microbiological control and assurance than some traditional methods.
Lot release testing is used to establish whether a manufactured batch meets the required criteria before it can be supplied or administered. With therapies that have a short shelf life, reducing testing times can be especially beneficial, provided quality and safety are not compromised.
Gyorgypal sees plenty of lessons to take from established bioprocessing, but concludes that really it is about applying the same principles, not assuming the same manufacturing approaches will work for every type of therapy.
“The transferable lesson is therefore not a particular production platform, but the understanding that early identification of which process variables determine the quality and function of the actual product is key,” he said.
“The common lesson is to develop the therapeutic and its manufacturing process together. If manufacturability informs the early design, you have a chance to build in robustness rather than trying to retrofit it after the clinical program is already committed.”
Source: www.technologynetworks.com



