BioPharm International: Welcome to the BioPharm International expert interview on analytical considerations for successful upstream process development. This interview, conducted independently by the editors of BioPharm International, is sponsored by Cytiva. Cytiva provides bioprocessing products and services to the biopharmaceutical industry for the development and manufacturer of therapeutics and vaccines. The company uses its scientific knowledge and technical expertise to help biomanufacturers increase speed to market, reduce costs and improve productivity. Thank you for joining us.
We are very pleased to have Dr. Gregory Lane with us today from Bristol Myers Squibb. He started in the year 2000. He's a Ph.D. organic chemist in small molecule API process technology department. During this time, he's developed scalable synthesis and utilized process analytical technology to monitor reactions and crystallization from the laboratory to the pilot plant. In 2016, he moved to the drug product science and technology department at BMS.
Today we're talking with him about upstream processes, continuous manufacturing, and just in general how the industry is moving and where it's heading to. The Fort Devens facility is a plant that you've had hands on experience with. Can you relate that or tie that into why the money was spent, why the investment was made and, overall, where sort of the industry is going with real- time analysis?
Dr. Gregory Lane: I don't work there every day, but I have worked with them for periods of time; and what I've observed and seen papers published by us and others is the savings in time is where it's really at because every sample that has to go to an analyst, whether it's a small molecule or a protein or a biologic, takes time to run on a system. It takes a dedicated person to run that sample. The PAT instrumentation is built into the process, and i's there real-time, and then people can look at the data as they see fit. So, it's a time savings versus offline analytical results.
The other benefit, of course, is the trending. You see things that you may have been missed if you're only pulling one sample a day. Something could have happened in those three hours between samples one and two, and the instrumentation will pick up those types of hiccups in a process. Ideally, you want everything smooth and running right down the middle of your intended process parameters, but if something doesn't go that way, it shows up in the trends of the process analytical technology.
BioPharm International: During scale up, because I know that part of the point of the Devens facility was to increase capacity during scale up, have the kinds of things that go wrong changed?
Dr. Gregory Lane: Scale up is all about robustness. It really boils down to having rugged instruments that perform well in an environment that's being sprayed down with water and part of a cadence—that's the whole point of doing stuff in production and operations, and I've seen that in my pilot-plant experience. The emphasis there is ruggedness. And when it comes to PAT implementation, it will go through a lifecycle where it's developed in the laboratory, tested in a scale-up laboratory, and then eventually, it works its way into a larger commercial manufacturing environment. Of course, as you go up, you have a lot more people involved, and a big part of it is the automation teams and the communication and networking teams, because once that data is in a pilot-plant environment, it has to find its way onto a historian or some other process control and monitoring system within the commercial or development laboratories where they're collecting all this data. That's a big part of it, which maybe folks like me don't always appreciate in a development laboratory. We quickly learn that there's a lot more to it than just setting an instrument into a process.
BioPharm International: Sounds very complicated, but I think cell perfusion is becoming more popular, and it is sort of dealing with how to cope with metabolites and other normal parts of the cell microenvironment. So, can you talk a little about that?
Dr. Gregory Lane: Chris, the reason I moved into the drug product space was to get more involved in biologics and protein production, and upstream is a big part of that. As you mentioned with the perfusion reactors, that's an area where we've seen a lot of growth in the last five to 10 years, because people are generating large quantities of cells in a very short period of time and then using that to either inoculate larger bioreactors or run in a continuous manner. That's where PAT comes in handy.
Raman is the primary application because it's in the tank monitoring the cell culture, which is the perfect example of the feed rate not matching the cell growth rate. You can starve those cells in short order because the populations are enormous, so the feed has to match the consumption. By monitoring that in real-time with Raman, for instance, you can keep track of the glucose ingress into the system, as well as the lactate production. All those materials are well balanced to keep the cells healthy and growing. Some companies do this for days or weeks for a continuous process in perfusion, where they take off material and process it. Others are just growing the cells in what's called an N-1 fashion. That means you're going to use those cells about three or four days later to inoculate a larger bioreactor. The use case there is that you want to make sure if you have three or four perfusion reactors running simultaneously that all three and four of those reactors are hitting the finish line at the exact same time, so, the cell density, as well as the quality of the cells you're putting into the larger bioreactor, are the same and you're not going to handicap yourself in any one of those large bioreactors when you go to inoculate.
BioPharm International: Is the critical process parameter one set of priorities, and it then bumps into the critical quality attribute pyramid of importance? Can you comment on how to balance those things, or how you suggest other people balance those things?
Dr. Gregory Lane: I've seen it both in my work, but I've also seen good examples in the literature from Biogen and Amgen in particular. Specifically, what we've seen is trying to control glucose levels to ensure that the protein glycosylation is well behaved. So, the quality is the protein, the critical process parameter you're monitoring is glucose, or even lactate, in the system in the upstream environment. Raman, again, is a perfect tool for that, but what they've really done, and I've seen this done in the literature, particularly with some Biogen examples, is to monitor both glucose and lactate and ensure those levels are kept within very careful ranges.
Therefore, the protein is ensured to be high-quality and not either overglycosylated, or they may be looking at acidic or charge variants.
Again, those are critical quality attributes. The only way you can do that is to monitor in real time because the typical process has a large injection of glucose every day. That may be overfeeding, or if you wait too long, under feeding the cells, and that's going to impact the quality of the protein. It's one of those classic examples, Chris, where you can't make good wine from bad grapes. It's the same thing. If you have an upstream process that is generating material that has too many impurities or too much variation in the glycosylation pattern, then folks are going to say, “I can't clean this up with chromatography in the downstream step. We have to go back to the beginning, get better control over the process.”
Again, that's why you need the real-time analytics, including spectroscopy. I also forgot to mention dielectric spectroscopy, which is capacitance, which is another tool that's made a big ingress in the upstream space for monitoring cell growth, but it also can tell you things about cellular health.
BioPharm International: How real is the real-time aspect of PAT? What kinds of delays within the system can a normal setup tolerate?
Dr. Gregory Lane: For upstream, since most of those typical batch bioreactors are two weeks if you're only using a single probe, you can get measurements as quickly as every 15 minutes, but typically you're monitoring four vessels at a time and once per hour. But once per hour is still 23 times faster than taking one sample per day. That is still an order of magnitude or 20 times more data that you would normally collect by pulling samples. Pulling samples takes a lot of time. I've been in a laboratory where we've had 15 or 16 bioreactors, and it took two hours of running around, grabbing samples from all of those vessels, just to get the analysis for a single point in time in a day. And you can do this every hour with spectroscopy or even faster with some other techniques, but that really accelerates the quantity of data, and it lets you generate the nice little trends that you see changing in real time.
BioPharm International: For continuous manufacturing, does that create a different burden on the types of analytical tools that you're using?
Dr. Gregory Lane: In the small molecule space, continuous manufacturing revolves around monitoring with a near-infrared spectrometer most of the time. It's a dusty environment and there's a lot of material flowing through the system at any given time. But the huge benefit, whether it's small molecule or even large molecule, is if you're operating a continuous manufacturing environment, you tend to have a lot lower variation because there's not somebody constantly shaking something or putting it into a bin and then rotating it around. It's all fed in through an automated feeding system. What we’ve seen, at least what I've picked up from other companies as well as perhaps in my own experience, is you lower the variation. If you lower the variation, the chances of having to do an investigation or dig deeper after a process is run because something didn't quite go according to expectation goes away and variation, or at least understanding variation, is very time consuming from a scientist and human resource standpoint, so having something continuous lowers that burden. It does create a larger organizational commitment because you have one person expecting material in a continuous manner from somebody else, as opposed to the typical batch process, whether it's large molecule or small molecule. So, it does require an entirely different way of looking at things.
BioPharm International: Ironically, the amplitude of the trends is decreased under a continuous setting, so that in some ways, it's easier. How do you then integrate that with somebody else's processes?
Dr. Gregory Lane: Exactly. In continuous, there's always a start-up phase, a running phase, and then a shut down or end-of-process phase—and pharmaceutical has kind of been slow to this game. Let's face it, there are plenty of companies out there, from oil to paper production and food production, that do continuous processes all the time, and they look at our industry and wonder, why haven't you guys jumped on this bandwagon? It's always a learning process, but I see the trends shifting in that manner. You will see many pharmaceutical companies touting some of their continuous processing, both in the small molecule and the biologic space, and it's all around the world. It's not only in this country; it's in the Far East and in Europe as well. So, it has evolved slowly, never as fast as I would like these things, but I'm the person who likes playing with the technology.