Amgen’s Jerry Murry on the Next Steps in Bioprocessing and Manufacturing

By Allison Proffitt

August 27, 2024 | Jerry Murry, senior vice president, process and development, at Amgen, gave the Monday plenary presentation this week at CHI’s Bioprocessing Summit. With an eighteen-year tenure at Amgen—and experience at Merck and Pfizer before that—Murry has seen nearly two decades of advances in the world of bioprocessing and manufacturing.

Harnessing the power of biotechnology to serve patients every day is a “humbling responsibility,” Murry told the standing-room-only crowd at the event. That has certainly always been the case. The challenge now, he said, is scale. Amgen has been dedicated to scaling its bioreactor and bioprocessing systems for several years. Murry walked through the various manufacturing improvements the company has adopted over the past decade to improve the scale of its processes.

One of the challenges, he said, is the heterogeneity of Amgen’s current products. Ten years ago, Amgen’s product list included 12 commercial products: some endogenous proteins and two monoclonal antibodies. Today, there are 36 commercial products: “A lot of complexity, a lot of differences in how we manufacture them, and a big variety in terms of needs and scale,” Murry said. “We have products that are dosed at below microgram levels. And we have products that are dosed in the hundreds of milligrams, if not grams, on a monthly basis. So we need to be agile and to develop platforms that allow us to deliver against this very heterogeneous set of products for the company.”

There are serious commercial drivers as well. Murry reported that year over the year, “most medicines” are reducing their prices by 5% to 7%. “Meaning that just to break even in a year over a year basis, you have to demonstrate 5% or 6% productivity in your overall manufacturing processes—just to break even!”

Those sorts of sustained productivity gains are not going to be made with just iterative engineering advances, Murry argues. “There are things you can do from OEE [overall equipment effectiveness] and optimizing timing and run rates, etc. But you will very quickly run out of those continuous improvement ideas if you don't have new technology to catalyze those ideas.”

Automating Visual Inspection: AI Use Case 

Murry outlined the steps Amgen has taken in the past decade to make significant process improvements—some engineering and some biological.

To advance biomanufacturing, Amgen uses continuous manufacturing for producing drug substances, has miniaturized robotic systems to fill each unit, and has begun to use automated visual inspection to confirm quality for syringes and vials. 

Automated visual inspection—done with AI—was an initially challenging process, he said. All product receives visual inspection. Initially this was done manually, but the percentage of visual inspection that has been done automatically has been growing. “We found that to be very limiting,” he said. “In some cases, we get rejection rates that are in the double digit. And you can't be throwing away 10% of your product at the very last step during inspection!”

Amgen worked on refining AI and machine learning computer vision algorithms to increase the percentage of product that can accurately inspected, but it was a long process, that “caused a lot of concern in the overall regulated industry,” Murry said. “Validation and learning are two things that need to be balanced. If you have a validated process and you’re allowing the machine to learn, how do you keep that in validation?” Answering that question took several years and conversations with FDA, but today, Murry reports, 95% of syringes and vials are released through automated visual inspection.

(Murry spoke more about Amgen’s AI approach in a fireside chat after his presentation.)

Decade of Bioprocessing Advances

But to achieve the productivity improvements needed to keep up with pricing pressures, changes need to be more universal. Over the past decade, Murry said, Amgen has invested heavily in improving its drug substances network. One of its largest sites in Rhode Island, for instance, has 20K stainless steel bioreactors. “The efficiency of that plant! When that plant is running, it is unstoppable. It’s an amazing feat to watch,” Murry said. However, he estimated that adding similar capacity would cost $2 billion today. “We were faced with that challenge. We were looking for a way that we could get the same level of productivity without a much, much smaller footprint.”

Amgen launched a next generation facility in Singapore making use of single-use bioreactors and perfusion, saving footprint, taking half the time to construct, and costing about 20% of the legacy site, Murry said. The new site, “allows us, in a 2K [bioreactor] format with perfusion, to get to the same level of productivity that you could get out of a 15K vessel,” he said. The next iteration went to even smaller bioreactors—500L—and was built in less than a year. Both properties are making commercial products now, Murry said. 

“Basically, what really allowed this to happen was the intensification using single-use bioreactors, using perfusion, and the way that we staged that so that the volumetric productivity of a 2K bioreactor can give you about 50 kilos of drug substance within two weeks' time,” Murry said.

Manufacturing can go even smaller depending on the properties of the product—to a space about the size of a closet, Murry said. The Amgen manufacturing facility that size, “manufactures some very potent products, but in a way that is very, very productive.” Rather than using a traditional fed batch process and running those every two weeks, perfusion gives much higher viable cell density and continuous manufacturing lets a batch grow for 40-45 days. “The same set of cells continues to produce product. Every three or four days, you’re siphoning product off and sending it to the downstream [purifying and harvesting] while you’re continuing to make product in the bioreactor.”

The continuous manufacturing approach requires a good deal of automation, “or you’ll find yourself there on a Sunday night wishing that you had taken the time to set up your automation ahead of time,” Murry said. “I can tell you that’s a true experience.” 

Amgen’s latest generation of bioreactor resides in a new facility in North Carolina. Amgen built an entire manufacturing process around the 10K bioreactor nicknamed FleX Batch. The facility in North Carolina was built very quickly for about two-thirds of the normal investment and about half of the water consumption, Murry said. “We’re expecting that we’ll be producing via 10K bioreactors several hundred kilos of protein for the products that I mentioned that are really going to be used globally.”

Really, Really Productive Cell Lines

The two newest approaches that Amgen is exploring for productivity gains Murry calls “intensified fed-batch” with “really, really productive cell lines.” 

“Now, in the entire 40-year history of Amgen, we’ve only ever had two commercial cell lines,” Murry said. But about a decade ago, Amgen started to work to develop a new commercial cell line that could be used by both the commercial manufacturing business and the discovery research business—a cell line that had very high yield for the commercial business and was very fast for the discovery business.

“We’ve never had this before because while our commercial cell line is very robust, he or she is very slow to develop and much too slow for what the discovery research group needs,” Murry explained. “They need something that is much faster that can give them two to four grams of protein very quickly so that they can move on because they’re making hundreds if not thousands of molecules at that stage.” But to go to the trouble of replacing the current commercial cell line, Murry said, Amgen wanted to see 3x productivity gains. “Under yesterday’s standards, I guess I would say we were getting eight to 10 grams per liter fairly routinely; we were expecting that we would get to 20 to 25 grams per liter with a new cell line,” he said. “We need to make it worth our while.”

In addition, the team wanted seamless workflow between discovery and commercial businesses, and to save at least 40% off the time to first-in-human (FIH). 

“And finally, they can’t just work for a standard monoclonal antibody,” Murry added. “What we’re seeing now, and many of you are seeing the same thing, is we have almost no standard monoclonal antibodies in our early pipeline. Almost everything that we’re working on today are bi-specific in nature in some way or another. And for a long time, those were always the outliers. Those were always the single digit gram per liter products that didn’t really fit the mold. And what we needed was a cell line that could be developed and vectors optimized for those types of Frankenstein molecules that were going to become more of the norm.”

The resulting cell line didn’t happen overnight, Murry emphasized. All of these developments have been a long time coming, reflecting a great deal of work. But the returns are rich. The new cell line is 3x more productive in manufacturing and does not require expensive media for those productivity gains. Murry confirmed that Amgen has products in patients developed and manufactured with the new cell line.

“We find that the time from vector identification to clone is almost half of what it was taking us before. And because [discovery research is] using the same cell line that we’re using, that allows seamless transfer straight into FIH and GMD production.” Research and manufacturing are also together developing robotics and automation solutions.

Highly productive cell lines again impact the design of the manufacturing facility. “We find… that there’s really not much of a need to go beyond 10k scale,” he said. “Instead of going to 20k bioreactors, we scale that at 10k, we nest those at a 3-day run rate where every 3 days we’re harvesting out of a 10k bioreactor. And if you just do the simple math, that’s a lot of protein coming out of that.”

Network Optimization

All of these advances—from bioreactor improvements to new cell lines—has generated a huge amount of new data, Murry said, so in addition to making process improvements Amgen has invested heavily on monitoring the efficacy of those improvements. “You can just imagine the amount of runs that we do, both in process development laboratories, but then also as we’re transferring [those improvements] to manufacturing and doing very large qualification and validation runs.” 

These data are complex, varied, and global, Murry said, so Amgen has also been investing significantly in data analytics and network monitoring, primarily at its hub in Cambridge, the NOT—Network Optimization Center.

“This is an enterprise data fabric, we call it. It has many, many, many input systems and those systems are big and clunking and don’t like to talk to each other,” he said. Amgen took an open data approach. “Basically everybody has access to all the data all the time,” he said, “and that helps us curate the data and look at ontologies of the data to make it useable.” In one room in the NOT, data analysts watch data from bioreactors from all over the world: from Singapore, Rhode Island, Ireland, Puerto Rico and more.