Most new drug molecules today are chiral, so developing asymmetric technology that will work at production scale can be difficult
Shaw is a freelance writer based in Marlboro, N.J.
If you’re making spaghetti sauce for eight and the recipe serves four, no problem: just double the tomatoes, onions, and spices, and dinner is served. If you happen to be feeding a battalion, however, simply multiplying the recipe by 100 or 200 might not work. A recipe scale-up that vast often involves changes in cooking time, ingredient proportions, and the size of your saucepan, for starters. The more complicated the recipe, the more likely it is that if you try to serve it on a gigantic scale by simply multiplying ingredients exponentially, your dinner will wind up a disaster.
So it is, more or less, with pharmaceutical compounds. Taking the reactions that work perfectly at the laboratory level and scaling them up to the manufacturing level is
click the image to enlarge
In situ images from a crystallizer. The size and number of crystals can be tracked as they change over time. (Source: Mettler Toledo)
not a simple, linear process. “Nothing ever scales linearly. You never get to take a product at 5 g and then do it at 50 g and have it work the same way, and all the more so when you’re moving from 50 g to 5 kg,” says Paul Vogt, PhD, assistant director of chemical development for Albany Molecular Research Inc., Albany, N.Y., a chemistry-based drug discovery and development company. “New problems always arise that you cannot anticipate. No process chemist ever gets an easier process when they take it up in scale; there are always things to be overcome.”
Vogt takes the analogy back to the kitchen. “Let’s say you had to boil 10 cups of water. It wouldn’t take you all that long to do it on a stovetop. Now imagine you had to boil 10,000 gallons of water. First of all, it will take you a significantly longer time to get up to that temperature. If you’re running a reaction in that water, you don’t know the effect that being exposed to all that extra heat for that additional time could have on a reaction until you actually run that reaction. It could lead to new impurities, different impurity profiles, or changes in polymorphic forms, for example.”
To some extent, these are challenges that have remained the same for as long as there has been mass manufacture of pharmaceutical compounds. John Ragan, PhD, an associate research fellow with Pfizer Global Research and Development, recently published a review of the types of reactions scaled up at Pfizer from 1985 to 2001 in Organic Process Research & Development [R. W. Dugger et al., vol. 9, pp. 253-258 (2005)].
“We wanted to determine the nature of the chemistry we run on scale; has it changed significantly over this 16-year time period?” Fundamentally speaking, the answer was no, but Ragan and colleagues Robert Dugger, PhD, and David Ripin, PhD, did observe some trends. “For example, there’s been an increase recently in metal-mediated cross-coupling reactions, which can present special challenges. A lot of them are catalytic in the metal-ligand complex. We love catalytic chemistry, because it reduces cost and waste, but it can also present challenges in terms of dealing with low levels of heavy metal contaminants.”
|Scaling It Safe
Taking chemical processes from lab scale to manufacture scale can involve proportionately greater safety concerns. Particularly as discovery scientists continue to develop highly selective, high-potency compounds, containment and other safety issues become ever more challenging. “We’ll sometimes have a compound that has, for example, less than 1 nM potency,” says Pfizer’s John Ragan, PhD. “That’s a relatively small subset of our portfolio—the standard potency is around 25 to 100 nM—but these highly potent compounds can be particularly difficult to work with at scale.”
When something does go wrong in the scale-up of a new drug compound, as in most other industries, public relations concerns often keep the event out of the public eye. “That’s one of the biggest problems with safety in the pharmaceutical industry,” says Albany Molecular’s Paul Vogt, PhD. “When incidents are disclosed, it’s often years later, and not necessarily in a forum that chemists would read, so we don’t have the opportunity to learn from our mistakes.”
Aiming to fill that information vacuum, Vogt joined with David am Ende, PhD, with the chemical research and development group at Pfizer, in writing an annual safety review for Organic Process Research & Development. The editorial “Safety Notables: Information from the Literature” looks at literature that chemists normally wouldn’t have the time or inclination to read, like environmental health and safety journals, and compiles an annual accounting of “incidents and accidents” in the pharmaceutical scale-up process.
“Repeat types of accidents seem to occur every 10 or 12 years, because the institutional memory of an incident is lost and so are the lessons learned from it,” says Vogt. Based on this annual scouring of the literature, the most common lessons lost to posterity in scale-up safety are those dealing with chemical incompatibilities. “Certain reagents don’t work well with a solvent, for example. People don’t take the time to research it, or the interaction has been lost to our collective memories. We hope that the review provokes chemists to look at articles that they might not otherwise have noticed, in journals they don’t usually read, and focus on these safety issues.”
Even more significant in the scale-up equation is the fact that most of the new drug molecules coming out of discovery these days are chiral. “Developing asymmetric technology that will work at scale is one of our biggest challenges today,” says Roger Bakale, PhD, executive director of process research and development at research-based pharmaceutical company Sepracor Inc., Marlborough, Mass., “A lot of the asymmetric technology developed in pharma, as well as in academic labs, is quite difficult from a scale-up perspective, because the performance is affected by scale-up itself.”
Many times, scientists will default to classical techniques such as enzymatic resolutions to produce chiral molecules. “More pharmaceutical companies run classical resolutions because they’re easier to scale up, more robust and more cost-effective in the long run. Catalysis can be quite expensive, especially if you can’t recover the catalyst,” says Bakale. Asymmetric chemistry is preferred, but it takes more development time to scale up because so many factors affect an asymmetric induction to produce the chiral molecules in question, such as temperature and mixing effects.
Yet another big scale-up challenge involves polymorphism. Scientists must produce a consistent particle size for the formulated product; if it’s irregular or different, that can change the drug manufacturing process. If the active pharmaceutical ingredient in the compound is not consistent in its particle size distribution, it can become a significant problem. “You want to develop a crystallization process that gives you the same particle distribution and polymorphs at scale,” Bakale says.
Polymorphs are different crystal packing structures, and it is very important to always produce the same one, as they have different bioavailability characteristics. Getting inconsistent forms can change the drug-manufacturing process.
One of the better-known examples of polymorphs causing scale-up problems occurred during the development of the protease inhibitor rotonavir. “They’d been producing that single form for six to nine months, and then it was discovered—unfortunately, during manufacturing—that another form was also being produced, and they had to shut down production for six months to solve the problem,” Bakale says. After production of the slurry, milling can be used to achieve a desired particle distribution, but that adds time to the process; ideally, manufacturers want to develop a crystallization process that leads to the same particle distribution and polymorphs at scale, without the added step of milling. “During the process R&D activity, you have to run extensive crystallization studies verifying that you have good control over what polymorphic form you have.”
All this, of course, has an effect on the bottom line. The longer it takes a company to address its scale-up problems, the more expensive it becomes and the longer it is before a drug is successfully brought to market. “That’s the big challenge in scale-up today, which has also been the biggest challenge for the past 30 years and probably will be 20 years from now as well,” says Vogt. “How do you scale up production as quickly, cheaply, and safely as possible?”
A PAT on the back
One increasingly popular approach involves in situ analytic tools that monitor the progress of chemical reactions. Known by the catchall term “process analytical
click the image to enlarge
Typical results of in situ mid-infrared measurements which follow reaction changes on a second-by-second basis. (Source: Mettler Toledo)
technology” (PAT), these tools allow scientists to better understand what’s going on in a reaction while that reaction is still taking place. Although the concept of PAT has been around for some time, it’s recently begun to take off as a scale-up tool.
“These are starting to become powerful technologies for people to use,” says Vogt. One particularly promising tool is the ReactIR 4000 from Mettler Toledo, Columbus, Ohio, which provides a real time, dynamic picture of chemistry using mid-infrared spectroscopy to follow the changes in the reaction on a second-by-second basis. “We use the ReactIR quite a bit, not only for reaction monitoring but reaction screening, to help us make decisions about what kind of chemistry we are going to use.” ReactIR allows process chemists to see, for example, what impurities might be forming at what point in the reaction timeframe. “If you knew that four hours into the process you start to see a particular impurity, you could avoid that altogether by figuring out how to get the reaction complete in three hours.”
Complementary to ReactIR, which examines the liquid or gaseous phase of a reaction, is Mettler Toledo’s Lasentec, which provides similar real-time information on particulate-based processes and crystallizations.
“You are basically using a laser beam, if you will, to look at the particle distribution of the slurry in situ,” says Bakale. “You can have a particle analyzer probe inserted anywhere in a process stream to give you a shot of what’s happening. [Infrared spectroscopy] shows you what’s happening in solution, while laser probes like Lasentec show you the physical chemical characteristics of growth outside of solution.”
Other PAT methods
There are other methods that fall under the PAT rubric as well. “You can also certainly use UV (ultraviolet) technology, which lets you look at the amount of moisture in the dryer, for instance. You can use Karl Fischer methods, which are not inline, but offline, measurements. Basically, you’re taking a small piece of the stream outside and analyzing it—which isn’t in situ, but it’s still PAT,” says Bakale.
Yet another such technique is focused beam reflectance measurement (FBRM), also offered by Mettler Toledo. This laser probe offers a precise, sensitive method for tracking changes in both particle dimension and particle population in suspension. “It’s used to determine the average particle size in a reaction mixture when the product is starting to crystallize and you have solid particles suspended in a liquid,” says Ragan. “This can be very important for formulations.”
In addition to PAT, says Vogt, there has been more work lately focusing on chiral asymmetric synthesis using new ligands for chiral hydrogenations that can perform at scale. “Our group, Merck, other manufacturers, and academics are all at work on developing new ligands and new ligand-catalyst pairings to do asymmetric hydrogenation. That makes chiral molecules without forcing you to throw half of your result away. Some really good examples have been presented in the literature and at conferences.”
Automation and robotics are also contributing to the enhancement of scale-up. “When you’re trying to make the process more robust before taking it to scale, there’s a lot that’s been developed in recent years in robotics that enables you to get more chemical information per experiment per gram of material,” Vogt says.
“If an automated system allows one person to run, say, 16 reactions at once, collecting data on what the temperature or concentration reaction should be run at and how long it should take, you can generate much more information much more quickly,” he says. From that, you can design experiments that should lead you to the most optimal conditions for scale-up. Used in combination with PAT’s real-time analysis, you can hopefully generate a tremendous amount of data to carry forward into the scale-up.”
Filed Under: Drug Discovery