BIOPHARMA SAYS BYE BYE TO BLACK MAGIC
Michael Kowolenko, PhD, refers to the early days of biologics manufacturing-the late 1970s and 1980s-as the "black magic" era. "When we were doing things in roller flasks and hoping to God we could get our growth factors out of a fibroblast, we knew we had to have certain media to grow certain types of cells, but we never understood why," says Dr. Kowolenko, senior vice president, biotech operating unit, technical operations and product supply for Wyeth Pharmaceuticals.
As the industry evolved, biopharma companies knew they had to improve their methods of controlling processes to make them more predictable at a larger scale; they just didn't know how.
"Young companies like Centocor, which had two products approved in the 1990s, had to put a supply chain in place and grow a biopharma development group quickly," says John Dingerdissen, vice president of drug product development in the biopharmaceutical development and marketed products support group within Centocor research and development. "We had to grow up to become larger pharmaceutical entities, without a lot of experience with multiple products being developed at the same time, and all the while challenged by the complexity of the science we deal with in biomanufacturing."
"Young companies like Centocor, which had two products approved in the 1990s, had to put a supply chain in place and grow a biopharma development group quickly. We had to grow up to become larger pharmaceutical entities, without a lot of experience with multiple products being developed at the same time, and all the while challenged by the complexity of the science we deal with in biomanufacturing."
-John Dingerdissen, Centocor
So, in the midst of biopharma's exponential growth in the 1990s, it was fraught with bottlenecks and technical challenges. If you walked into a biotech facility in those days, says Dr. Kowolenko, it was like walking into a suburban housing development. "There was a left-handed split and a right-handed split, and they were all the same. We were familiar with certain types of cells and expression systems and optimized that-but the problem was it turned all of these facilities into one-trick ponies. We had this tremendous investment in infrastructure, with a typical plant costing around $600 million to produce a product, but they were only producing one thing."
But thanks to several factors, ranging from a vastly improved understanding of cell culture production to new systems approaches borrowed from other industries, biopharma is beginning to overcome the hurdles of its youth.
For years, cell culture production was the most difficult and costly step for biopharma. "Ten years ago, cell culture production represented about 70% of our costs. Now it's about 30%," says Dingerdissen. "Industry has mastered these steps, a lot of which had to do with better selection of the original cell line or replacement of weaker and less productive cell lines during process development. Our titer productivity per cell has gone up from a half gram to three grams per liter several years ago to a norm of three to six grams per liter today. That was a major limiting factor-how much protein was in a batch-and it's significantly changing our paradigm on productivity."
As biopharma began making more optimal use of its assets, the focus started to shift toward the increased yield coming out of cell cultures. "Way back when, when we were using fetal bovine sera as a supplement, you had to hope for the right lot harvested at the right time of year for cells to grow," says Dr. Kowolenko. "Now, we can define what the cell needs to grow and supplement it and at the same time modify the biochemistry of cells to produce an antibody or fusion protein to survive in cell culture."
As yields became larger, so did reactors, and companies began investing in how cells behaved in the milieu of a 20,000-liter reactor. "The joke is that we've never gotten it right. We either have too much or too little capacity," says Dr. Kowolenko. "Now we're starting to see that we can take facilities designed for one product and maximize output by reengineering to make our facilities much more flexible."
In this new environment, drug developers can stop the expansion of cells into larger reactors by letting them grow to a certain density where their growth stops. By increasing concentration, researchers can harvest cells early and produce more material with shorter cycle times.
Today's bottleneck exists more on the purification side; companies can produce product much faster than they can purify it. Back when serum was still a part of the process, purification was even more of a challenge. "The drug was basically a contaminant in the whole milieu of the media," says Dr. Kowolenko. Getting rid of serum was a great leap forward, and today, biopharma scientists get a little media with their drug instead of a little drug with their media. But the fluid generated still exceeds the capacity of a given chromatography column.
"So, we're looking at improvements on that side, such as differential precipitation, the combination of filtration with chromatography media," says Dr. Kowolenko. "Basically, it's anything that will either get your volumes of liquid down or increase your surface area, since when you do chromatography, it's all a surface area phenomenon."
Reducing purification process steps from an average of three to five down to one or two is a hotly pursued goal for most biopharma companies, says Dingerdissen, who adds that Centocor is touting orthogonal chromatography steps as a means to this end. "We're also using newer technologies, like more specified resins that may be more expensive by unit cost but accomplish two or three of our process needs rather than just one."
Biopharma companies are also focused on reducing the complexity of their overall process-something that got short shrift in the early start-up days. "In the 1990s, and even from 2000 to about 2003, we were focused on getting the product out, at whatever cost, to get it to the patient as fast as possible," says Dingerdissen. "In the last four to five years, reducing the number and complexity of the steps it takes to get to a high-quality product has become the focus. It's become a huge competitive advantage to be able to produce these molecules in fewer steps, particularly in downstream manufacturing."
"We've moved away from having the process by the product to understanding the levers-what the qualities of the product are that give it that activity. This allows us to build those sorts of analytical tools right into the process, so we always know if the batch is progressing in a fashion that will produce a product that has the right attributes."
-Michael Kowolenko, PhD, Wyeth Pharmaceuticals
For Centocor, a division of Johnson and Johnson, that has meant a significant reorganization. The company has brought its pharmaceutical process development people and the manufacturing technical experts who start up and maintain a plant together under one umbrella-biopharmaceutical development and marketed product support, the group Dingerdissen helps to lead. The changeover, which is ongoing, began in early 2008.
"One of the biggest hurdles to successfully shortening the time cycle in process development for biomolecules is transfer of knowledge from the discovery group to the manufacturing floor," explains Dingerdissen. "There's a lot of engineering and scientific adaptation that has to be done to send a product into a plant and make it cost effective, and that's generally been done by shoe-horning the product in and then working for several years afterward to improve the process."
By combining several hundred people from development and manufacturing under one leadership, Centocor hopes to bring much of that process to late-stage development and eliminate the handoffs in tech transfer that slow down manufacturing start-up. "We're already seeing benefits where we're finding pockets of redundancy in technical capabilities that were put in place under that different paradigm, where you moved a molecule quickly into manufacturing and fixed things later," says Dingerdissen. This is all part of a strategy borrowed from Toyota for parallel rather than serial product development. Dingerdissen cites Michael Kennedy's book, Product Development for the Lean Enterprise: Why Toyota's System is Four Times More Productive and How You Can Implement it, as an inspiration.
"Looking forward, everyone in the industry wants to shorten drug development time from seven to 10 years to five to seven years," he says. "Well, how did Toyota do it? How did they take [the] car development cycle down to about one-third of what it was in time and probably in money as well? Parallel development is a big part of that. Now, you can only do some of that with a drug, because you need a certain amount of early phase work before you can move forward, to minimize risk to the patient, but by having a closer relationship between discovery and manufacture and a more enhanced relationship with the supply chain, you can eliminate a lot of inefficiencies."
A PAT ON THE BACK
Process analytical technology (PAT), the pharmaceutical industry term du jour, is another key component of biopharma's evolution, says Dr. Kowolenko. Improvements at the cell culture level have made this possible. "We've moved away from having the process by the product to understanding the levers-what the qualities of the product are that give it that activity," he says. "This allows us to build those sorts of analytical tools right into the process, so we always know if the batch is progressing in a fashion that will produce a product that has the right attributes."
So, from what was largely a guessing game in which some product was thrown away and some was kept, biopharma is moving, in a fairly random way, toward the kind of data-driven integrated decision-making that much of the drug development world is also pursuing, retrofitting old facilities and designing new ones.
"For example, if you were now to do a process change or a process transfer from one facility to another, you'd know the culture conditions that the cell would require, you'd have a whole series of control data that you would analyze to determine whether or not cells were growing properly and producing protein, you'd know the byproducts cells were producing, and so on," says Dr. Kowolenko. "If things start to drift, you know what to change in order to bring that cell culture condition back into the normal operating range so that the appropriate protein with the appropriate attributes is expressed. It's not a black box anymore."
These gains in knowledge have had dramatic effects on productivity. According to Dr. Kowolenko, instead of having to budget for a write-off rate of 25% of lots, most biopharma companies now run at a 90% to 95% success rate.
Dingerdissen characterizes the current approach to biopharma as "a holistic view of drug delivery from the beginning of process development." For a selected few products in its portfolio, Centocor is planning to invest more work up front on drug delivery efforts and presentation to the patient to ensure that the right molecule for tech transfer is selected. In the past, many biotechs selected molecules for development with efficacy as the primary-sometimes the only-parameter. Developability was an afterthought.
"Now we're formalizing a more rigorous process to determine the developability of selected NMEs [new molecular entities] to move forward, to improve their probabilities of success in meeting patient needs more effectively, and giving them a competitive advantage," says Dingerdissen. For example, Centocor now has products under review by the Food and Drug Administration that at first presentation will be given by a health care provider. "Eventually, perhaps in the second or third year after approval, we hope to get the product into an autoinjector and make it easier for the patient to use. The way we're doing things now, we want to get to the autoinjectable delivery before the approval process and launch with a better presented product for the patient out of the gate."
That means that instead of just considering things like Ph and solubility, Centocor-and no doubt its competitors-will be broadening the parameters for its molecular entities by considering more product characteristics, such as scalability and ability, to be delivered by more patient-friendly application devices, including pre-filled syringes, autoinjectors, nasal inhalers, and patches. The plan is to develop the delivery device with the product whenever possible.
Too many biotech companies have been "bulk-centric," Dingerdissen says. "In small molecules, that would be API [active pharmaceutical ingredient]-centric. The focus has been 'if you can make bulk, you can produce product,'" he says. "To be more effective and meet more patient needs, we need to change that paradigm to deliver a product that's more focused on the need in mind-what does the drug product look like, not just the substance." ¦