Feliza Mirasol is the science editor for 

Feliza Mirasol

Biopharma quality control is being revolutionized by advanced analytics and digital transformation through a synergy that boosts quality and scalability.

Drug Digest: Closing the Gap in Modern Process Control and Automation for Quality

Digital transformation and advanced analytics are revolutionizing biopharmaceutical quality control (QC). By synergistically combining formulation and analytical methods, the industry enhances quality, scalability, and innovation. This crucial shift is driving R&D investment, making advanced biotech equipment essential for scalable drug production. This episode of Drug Digest will explore this modernization of QC and examine the role that advanced analytics and digital transformation play in this unfolding revolution.

, Fellow, Regulatory Compliance Associates

At Regulatory Compliance Associates (RCA), Schniepp is responsible for training and advising RCA staff and clients on current trends and issues facing the pharmaceutical, biotechnology, and medical device industries. Her background includes current good manufacturing practices, audit readiness, change control, investigations, supplier quality, and batch record review. She has more than 40 years’ experience in quality assurance, serving in leadership roles at Allergy Laboratories, Oso BioPharmaceuticals, Searle, Abbott, and Hospira. She has been on the Parenteral Drug Association (PDA) Board of Directors from 2011–2013 and from 2016–2019 and currently serves as the board’s chair. Schniepp has served on numerous planning committees, including the PDA/FDA Joint Regulatory Conference Planning Committee since 2002.

, PhD, Vice President, Technical, Parexel International

At Parexel, Dr. Schmitt provides consulting services to the healthcare industry on all aspects of regulatory compliance, particularly the design and implementation of quality management systems. The services cover the entire product lifecycle, from the clinical to the commercial phase, for all aspects of the good practices. His areas of expertise include all aspects of quality and compliance for systems, processes, facilities and operations for drug substances and drug products. Specifically, he consults on compliant quality systems that integrate industry best practices with future-looking compliance approaches. His career spans over 35 years in industry. He is an accomplished editor, author of nearly 200 publications, trainer and presenter.

® editors, who interview industry experts to discuss the emerging opportunities, obstacles, and advances in the pharmaceutical and biopharmaceutical industry for the research, development, formulation, analysis, upstream and downstream processing, manufacturing, supply chain, and packaging of drug products.

Editor's note: This transcript is a direct, unedited rendering of the original audio/video content. It may contain errors, informal language, or omissions as spoken in the original recording.

MJH, pharmaceutical technology presents drug digest, a tech talk with the pharmaceutical technology editors all about emerging opportunities, obstacles and advances in the pharmaceutical, biopharmaceutical and biotech industries. Join us as we discuss with industry experts the research, development, formulation, analysis, upstream and downstream, processing, manufacturing, supply chain and packaging, as well as business strategies and regulatory issues.

Hello everyone, and welcome to the joke digest video series. I'm Felice Mirza, science editor for pharmaceutical technology, pharmaceutical technology Europe and BioPharm International. In this episode, industry experts will discuss how digital transformation and advanced analytics are revolutionizing biopharmaceutical quality control and the challenges that the industry faces undergoing this transformation, joining me today will be Susan schneep from regulatory compliance Associates, and Dr Siegfried Schmidt from pyrexel International. But first a few words from our sponsors. This episode of drug digest is sponsored by LGM pharma, Nelson labs, the parenteral Drug Association, Thermo Fisher Scientific and beltek Associates. LGM Pharma is a US based cdmo, providing integrated drug substance sourcing and drug product development and manufacturing from early stage programs through commercial production. Nelson labs, a sotera health company, is an industry leading global provider of laboratory testing and expert advisory services that performs more than 900 rigorous microbiological and analytical laboratory tests across the medical device, pharmaceutical, protective barriers and tissue industries, the parenteral Drug Association is the leading global provider of science, technology and regulatory information, and since its founding in 1946 as a nonprofit organization, has been creating awareness and understanding of important issues facing the pharma community and delivering high quality, relevant education to the industry, as well as scientifically sound, practical technical information and expertise to advance pharma, manufacturing, science and regulation. Thermo Fisher Scientific is the world leader in serving science, enabling its customers to make the world healthier, cleaner and safer. Bell tech Associates is a single source supplier providing solutions for pharmaceutical or biotechnology contamination control requirements. And now, without further ado, let's get to our speakers.

Hi everyone. My name is Sue Schneep. I'm a Distinguished Fellow with regulatory compliance Associates, a division of Nelson laboratories. I've been in the industry about 40 years, so I guess I'm old and crabby, but I've learned a lot, and we're here with Siegfried Schmidt to impart some of our wisdom and knowledge onto the audience here with that, I'm going to turn it over to Siegfried to introduce himself.

Thank you. So my name is Siegfried Schmidt, and I'm Vice President technical at PAREXEL, and I've got over 35 years in industry and over 18 years with powerexcel, and so my expertise is really from clinical to commercial, all matters of compliance any regulatory system in the world. So yes, and it's it's always fun and a pleasure to share some of our expertise with the rest. Thank you.

Now for my first question, I'd like to throw this to Siegfried first, and then Susan, so Siegfried, how are advanced analytical techniques changing the definition of analytical best practices in biopharmaceutical quality control today?

Well, you see these techniques on the one side, they represent the state of the art, and on the other side, they help expedite analysis. And why do I mention state of the art? Because state of the art is actually a regulatory requirement. This doesn't necessarily mean that all the methods must be cutting edge, but you need to use the methods that are established by industry. So we are no longer talking about thin layer chromatography. We're talking. Talking about really modern analytical methods, HPLC, UHPLC, hyphenated techniques and so on. So from the perspective of the analytical laboratory, you want fast and accurate technologies because they help an increase of throughput of the analysis. It supports the fast creation of certificates of analysis, because these you want to have as early as possible, because you want to release your product or the product for the next step and so on. And being faster also means you reduce cost and you increase revenue, very important, particularly in R and D and so these technologies, they typically run on highly automated equipment, which, again, is an important factor, because Electronic Data Management helps, or is actually essential for data integrity. I am sure sue you have additional information you want to share on this?

Well, yeah, Siegfried, I think you bring up some excellent points what this new analytical technology is doing, which is, to me, very important, is driving to real time release in a lot of cases, because you've got in line monitoring procedures and processes you've got, you know, the rapid microbiology results that come along with this, and I think driving to real time release. And, you know, like, as you're making it, you're releasing it on the back end, you're manufacturing, you're putting it together on the front end, and you're shipping it out on the back end, all within, you know, 24 hours. I think it's important, because as we get into the biologics and the atmps, I think we're dealing with very sensitive products that may or may not have, you know, a stability, a longevity of stability, you may have to get it out the door relatively quickly in order for it to be viable with the patient, right? I mean, that's kind of where I think we're driving. And I think these are going to get more and more sophisticated, and some of our traditional thinking around compliance to the Compendia is going to need to go away, because the Compendia is not going to be able to keep up, at least under their current operating processes, right with the rapid changes going on with these techniques.

So you're, you're very right there to go and combine not just the thinking about the analytical method, but about the processes, because you need to look at it holistically, at both what is happening in operations and what is happening in the laboratory. You can't think of things separately. They are integral to each other.

For this next question, let's start with yusu. What specific impact has the use of innovative tools had on improving quality control?

I think the impact has been we've been able to get results faster and evaluate the state of the product faster. This becomes important. If you have like a defect or investigation deviation during manufacturing, you can get a result or resolution to that quicker, hopefully, which will allow you to longer you keep that subjective batch running through your process. Hopefully, you know, you don't get to the end, because that's a waste of money. If you have to abort the batch, you want to do it sooner in the process, rather than later. So I think it allows us to kind of close deviations, maybe investigations a little bit quicker. But I think there's a downside to it too, in terms of we know more about the product. In terms of, like, if we did impurity profiling, we probably are picking up with these pieces of equipment, a lot more impurities than we ever picked up with the traditional HPLC. And so now we're chasing, I don't want to say chasing our tails, but chasing our tails, trying to figure out whether the impurities are harmful, you know, and and. At some point we almost know too much about the product that almost negates it from being safe, right? It's kind of that catch 22

that's very interesting what you say, because the more precise, the more detail an analytical method can detect, the more you can find, but we clearly also have limits defined what is acceptable, what is not acceptable, and just because we can dig deeper doesn't mean do we always have to, but perhaps another point you already brought up is the speed of things. And I think these tools not only help improve the analytical methods to be faster and better, but it's also the processes that they support. They become faster, more reliable, more efficient. And what I recently saw was absolutely astonishing. Some of the tools that big pharma are developing and using, they use, for example, ultrasound, to move samples of meniscus volume in the laboratory so there is no more manual handling. It's ultrasound driven. Sample goes from a to b, phenomenal. And of course, when we talk about tools, we also have to think about, for example, the media that may be needed in the microbiology laboratory, and there are modern methods that help you understand the variability of the various media in the short time. So what we're talking about is something that helps us, not just with the actual analytical method as such, but with all the media that we use around them, the methodologies, the processes, and maybe I'm just repeating what I said before, but we need to look at it holistically. We can't just look at one element only, and I think this is where it's also important for the industry. If you want to improve, say, your processes in the operations part, then you also need to look at your laboratories. Can the laboratory support this? Can they not or the other way around? You may have a super laboratory, but a lousy process operations, and what is the value of a fast laboratory? If your process is still slow,

I totally agree with you. Sigma freed what and you what you said raised an issue with me on these new pieces of equipment and moving it with ultrasound. And there's also an element of, we're saving resources, right? We're not using as much. So there's an environmental friendly and sustainability quality about many of these new analytical tools that are coming down the line where we're not using, you know, benzene to or acetone, we're using a lot less resources, a lot less waste, which is, you know, more sustainable.

How would you say companies are leveraging the synergy between formulation development and analytics to enhance product quality, scalability and innovation throughout the entire drug development process. And I'll toss it to whichever one of you wants to start first,

what I've seen, and I find it quite interesting, is that formulation development rather sticks to standard formulations using standard excipients and formulation processes, I see limited innovation. Perhaps this is mostly with intention to save time and provide an acceptable formulation quickly for clinical trials and commercialization. Now, of course, if you have modern analytical methods and tools, you may get faster, more accurate analytical result. So So overall, I think the key driver is speed is not necessarily innovation,

and what core R D investments in advanced equipment are most essential for achieving scalable drug production in the next

decade? I don't know that R and D investment. I think it's going to be in the manufacturing line investments, not necessarily R and D. I think it's the manufacture. Lines that are R and D usually tends to have a lot of creative and clever and cutting edge equipment kind of built into their budgets, and so they're fiddling with, you know, things like the rapid micro. It's what I worry about, is the manufacturing facilities that are a little bit lagging. We tend to not update them on a traditional schedule. You know, we have equipment that's been around for 30 years. It's kind of like one of those cars that you get a little plaque if you get over, you know, 500,000 miles, and you're lucky, it's still running. I think the game to winning this game is going to need to manifest itself into key investments and purchases of new equipment in the manufacturing arena. Now, all is not lost, because there are several companies that are starting to build, you know, Greenfield manufacturing facilities around the world that would accommodate these new biologics that are coming out. But at some point, some of the older biologics are going to be transitioning to contract manufacturing arenas, and they're the ones that are really lagging in terms of updated equipment and facility, at least last time I checked, I'm going to throw it to Siegfried and see if he can correct my misinterpretations on this.

No, no, no. So yeah, you're right there. You're right. It's not the R and D. R and D can use design of experiments and all sorts, and they have a lot of nice tools, is here really is about commercial manufacture and see what we what you really need to get better and more consistent and more reliable manufactures that we need in line, online, nearline analytics, with feedback loops for continuous, if possible, real time analysis, because otherwise we will be stuck, as you say, with equipment that's 30 years old. Now the vessel itself is not the issue. It is the instrumentation. It's the automation. And when I say automation, I also think it's the integration of automated systems. If you don't have the integration, if you don't have the transmission of the data from one system to another, then you may wait, and it's not unusual days weeks for an analytical result. And what happens in the meantime with production? Nothing, it stops. And just to give you an example, and this was an R and D plant, unfortunately. So this plant that I visited, it takes 15 minutes for a person to manually transport a sample from the line to the laboratory. Yeah. I mean, this is not sustainable.

I totally agree, Siegfried, and one thing you prompted in my mind is this continuous loop as we're getting information faster and faster, but we're also getting interpretation. We're adding AI to the whole process, and virtual learning and digital twins and all that stuff that you know we're talking about in the AI. The AI world, and so I wonder how fast we, the human can learn, and we've got the machine learning quicker than we can right? Just some thoughts on that from you.

Well, so AI, everyone mentions AI, and I totally believe that AI will play a role in in the facilities of the future, say the laboratories, and where I think AI really works well is how to manage things, yeah, because these tools are really good at identifying optimal processes. Say, in the laboratory, you have a series of systems available, you have a number of samples waiting, you have a number of analysts and trying to figure out what is the optimal way which sample should go on, which equipment be analyzed by which analyst, etc, etc. We can do it, but slow AI systems are very fast at that, so that's great. And these tools may, may be useful for QA too, not for say, release decisions. But to support the review activities, and again, I have to say it's a connectivity that is absolutely essential to make these systems work. Because again, I see so many companies where analytical results are manually typed into a limbs to create a certificate of analysis. I mean, it's beyond belief. I have to say this is just antiquated, archaic. This is not the way to work.

Now, in both of your experience, what are the primary challenges contributing to the gap between hype and implementation of digital transformation, particularly concerning process control, automation and process analytical technology?

That's a very interesting question, and I'm really going to be fascinated to hear siegfried's take on this. But from my perspective, we hold ourselves back in a lot of ways, because we're we're afraid of the regulatory implications or filings or compliance that we need to achieve when we're adding new and innovative technologies, even AI into the even putting in a new line in our manufacturing facility, people tend to shy away because they're afraid of the regulatory implications or doing something, and then having the agencies Say, oh no, no, you didn't do it right? I don't think that should drive our thought processes here. I think what we need to and this is going to sound real old school and old fashioned, but in my way of thinking, what we need to do is revive our skills when it comes to risk based decision making. I think we have lost that talent in quality and regulatory areas of professionals, and we've driven ourselves over the last 20 years to what I would call a checkbox mentality. And we need to go back to those appropriate, risk based decisions and understand that if I can put in a new manufacturing line, it's going to be more beneficial and good for the patient as well as the product, as well as the company. And how do you do that? You need to look at the regulations and figure out which ones apply and how so let's just take an example. I if you're as as I alluded to before, if you're a contract manufacturing organization, you have multiple clients. You're producing multiple products. The traditional pharma thinking has been that in order to put in a new line, you need to build up two years of inventory of all the product that you're making, then you need to decommission the line, then you need to put in the new line, but you got to get permission, a prior approval supplement from in this case, it would be the FDA to put in that new line that may take you a year. So you may be down on a line for two years while you figure out this interesting dilemma of putting in a new line to improve productivity, reduce scrap and improve patient safety. Now, if you read the regulations in the US, you find out that you can do it relatively easy in four months. If you interpret the regulation. So for instance, if I have an old traditional line with what we called shower curtain barriers, and I want to put in a RABs restricted access barrier system, I can do that as an annual reportable, because there is in that particular guideline something that says, you know, any increase between the product and the human, any increase of intervention, which certainly a restricted access barrier, is a lot better than shower curtain, right? That can be an annual reportable so go ahead and make that change as a positive. So I think in some of all of this improvement, the quagmire is one we've made ourselves, and we need to develop a little bit more risk based decision making. Siegfried, I'd like your take on that that was a lot of ways saying we were stuck in the mud.

We are. We are so and yes, if I hear someone saying, Oh, I don't know, well exactly, why don't you do a risk assessment? Where are the facts? Where the data don't just use good feeling, give us the facts. If it if you say. The regulations don't allow it. Where does it say? Have you spoken to the agency? But I think when we look at these concepts like pa t, they have been around for two decades, and I know facilities where they have been implemented for two decades, and they work, and it's it actually is nothing new. But then we have to ask ourselves, why aren't they everywhere? And something you mentioned is the upgrading and upgrading an older facility to make the facility ready for full digitalization. That isn't trivial. That is really a tricky thing. And again, if you don't have an alignment between analytical and manufacturing technology, this will all be difficult. And so what? What I see, and I'm sure you've seen the same so is a lot of these facilities have absolutely minimal automation and digitalization in there. Yeah, there's very little integration of systems. And there's, there's reliance on laboratories rather than on online, in line analytical systems. And I remember nearly 30 years ago, we put an HPLC right next to the line. We didn't take the sample for 15 minutes to the lab, and perhaps they have some time, some time to give us a result, yeah. And what it seems to me is that the cost benefits of the digital transformation aren't really fully understood by so many companies. Yeah, but when you look at Big Pharma, the picture, to me looks very different, because these companies definitely embrace technology, and they profit greatly from it.

I really like what you touched on to kind of exemplify a little bit of where we're at. And Siegfried said it really well, some of these lines we're talking about, the manufacturing lines are very old and haven't been updated, and it's so it's like taking a 1968 Mustang and then trying to put in all the computer equipment and sensors of 1925 BMW, right? It's going to take a really long time and a lot of work and a lot of money to bring that kind of technology to an old manufacturing facility or place, right? And that's what we're talking about. So in some ways, you have to ask yourself, Do I scrap the facility and go green field, or do I try to do this very expensive upgrade, and maybe it works, maybe it doesn't right. So I think that was a really key critical point you brought up.

Siegfried, now Sue going off your previous comment. Have you actually seen any examples where companies like cdmos have made that decision, in other words, to go Greenfield, with a facility, or go with another strategy?

No, I haven't seen that decision. I think what's kind of happening, although I don't know right, because I haven't been but my guess, or from what I hear in the scuttlebutt, is we're kind of limping along there by adding technology slowly at certain points in the process. You know, so, but the whole transition that really needs to be able to keep up with some of these at Advanced therapeutic, medicinal products that are coming out, it's it's going to be next to impossible, I think, in some ways, to achieve that vision. But Siegfried is much more the expert than I throwing it over to him.

Well, cdmos find it very difficult to change a facility because it usually affects too many clients. So building a new facility, almost is the only solution. And when we suddenly what I've seen, and sometimes you get the facility of the Year Award for this is new facilities, and they're mainly built by large Big Pharma. They are the ones that are really automated, that leverage all what can. Be leveraged from these tools, and so really, they try not to upgrade the old equipment that's already 30 years old, because it's difficult. It's much easier to build a new one. However, on the other hand, I see many new plants, new facilities where the operators on their mobile phone have more computing power than what's on the equipment, and it doesn't make sense in some way. And what Sue mentioned about atmps, selling gene therapy manufacturing that there has to be a much higher level of automation built in right from the start, because, again, as Sue said, You cannot wait two weeks for an analytical result when you have the cells ready and it's either inject into the patient or put in the waste, either or. And you need results there and then. And so this can only be achieved through rapid methods, through fast, automated systems.

Well, thank you both, and I'm afraid that's all the time we have today. Thank you to Susan and Siegfried for being on today's episode and sharing their experience and insights. I'd also like to thank our sponsors, LGM pharma, Nelson labs, the parenteral Drug Association, Thermo Fisher, scientific and veltic Associates, for supporting today's episode. You can catch all our drug digest videos at www.pharmtech.com thank you for watching.

Did you enjoy this episode? If you did, please share this video with colleagues who you think would enjoy the content. Thank you to our editors and experts for sharing their insight. Thank you as well to our sponsors for sponsoring this episode. If you want to stay in touch with pharmaceutical technology team, subscribe to this podcast as well as to our E newsletters. When you sign up for our newsletters, you will be updated about future episodes of drug digest. Receive our magazines and more. Thanks to everyone for joining us for this episode of drug digest, we will see you next time you.

Videos

Drug Digest: Closing the Gap in Modern Process Control and Automation for QC/QA

Sanjay Konagurthu AAPS Interview Part Two

, held from Nov. 9–12 in San Antonio, Texas, 

, PhD, senior director,Science and Innovation, Thermo Fisher Scientific, about the presentation “

AI/ML Model Informed Early-Stage Drug Development.

 Dr. Konagurthu talks about integrating a systematic 

 framework to accelerate decision making regarding 

He says this framework provides insights into a final dosage form, as drug loading can impact both physical and chemical stability—and how much drug substance can be contained in a 

“Drug loading is an important aspect when it comes to physical and chemical stability,” Dr. Konagurthu says. “With some of these modeling tools we have deployed and implemented, we can predict not only the favorable excipients and polymers to screen prior to any experimentation, we can also predict how much drug loading one can achieve.”

Dr. Konagurthu also addresses how molecular dynamics simulations and quantum mechanical calculations inform complex 

“If the drug is bound too tight, it won't release,” he says. “If it's too loose, it doesn't really stabilize in the amorphous form. So, these are the series of calculations and various descriptors we assign in addition to the physical chemical properties ... to be able to do these calculations and narrow down the formulation design space. And all of this is done 

 of Dr. Konagurthu’s interview with PharmTech Group, in which he discusses how artificial intelligence and machine learning can help solve the dilemma of poor solubility.

 for all of our AAPS PharmSci 360 coverage.

Editor's note: This transcript is a lightly edited rendering of the original audio/video content. It may contain errors, informal language, or omissions as spoken in the original recording.

Hello, I'm Sanjay Konagurthu. I'm Senior Director of Science and Innovation at Thermo Fisher.

When we do things like quantum calculations, we look at the confirmer space. So what that does is give us insight into how a molecule folds upon itself. That gives us an idea about the accessible surface area of the molecule and the excipient polymers to interact. So that's one piece of it.

And we look at the various descriptors, like the classical hydrogen bond donors, acceptors, aromatic, hydrophobic. We look at potential energy surface charge models. These are like the sticky points of a molecule that essentially give us insight into how a molecule and an excipient or a polymer are interacting at a molecular level.

And then we look at molecular dynamic. Simulations, looking at interactions between the drug and the polymer. If the drug is bound too tight, it won't release. If it's too loose, it doesn't really stabilize in the amorphous form. So these are the series of calculations and various descriptors we assign in addition to the physical chemical properties, like I mentioned, to be able to do these calculations and narrow down the formulation design space.

. And so we are saving an incredible amount of trial-and-error and unnecessary experimentation and getting to the goal quicker.

Drug loading is an important aspect when it comes to physical and chemical stability. With some of these modeling tools we have deployed and implemented, we can predict not only the favorable excipients and polymers to screen prior to any experimentation, we can also predict how much drug loading one can achieve.

So what that does is, our priority is, you get an insight into what your final dosage form is going to look like because the drug loading can impact both physical and chemical stability, and how much drug loading you have can impact how much drug you can get in a tablet or a capsule. Those are the interplays that we can get insight into based on these AI/ML modeling tools that we have implemented at Thermo Fisher.

The conclusion of this interview with Sanjay Konagurthu focuses on simulations and calculations that go beyond simple screening.

Molecular Dynamics and Quantum Calculations: AAPS Thermo Fisher Interview, Part Two

Sanjay Konagurthu AAPS Interview, Part One

, PhD, senior director, Science and Innovation, Thermo Fisher Scientific, about the conference presentation “

In the above interview—the first in a two-part installment—Dr. Konagurthu discusses the potential of 

 using artificial intelligence (AI) and machine learning (ML).

“The big advantage is time and cost savings and resource savings for the industry and our customers,” Konagurthu says. “Time is money, literally speaking, and what we bring to the table is reducing risk in product development, saving time and resources, minimizing API consumption. It's an ultra-material-sparing approach by leveraging some of these 

 modeling tools like AI, leveraging AI/ML for supporting product development and formulation development.”

How can AI/ML improve early-stage drug development?

In particular, Konagurthu says, AI/ML can help sift through a field of drug candidates in which the percentage that are 

 has increased in the past several decades from 30% to as high as 90%.

“It's a challenging environment and the probability of success of getting a poorly soluble NCE [new chemical entity] to a commercial product is pretty low,” he says. “It's way less than 10%, probably more like 5%. So, by leveraging some of these AI/ML tools that we've developed, we aim and strive to reduce the risk in product development.”

Check back for Part Two of PharmTech Group’s interview with Konagurthu, in which he drills down further on molecular dynamics simulations and quantum mechanical calculations as tools to be deployed beyond simple screening.

Hello, I'm Sanjay Konagurthu. I'm Senior Director of Science and Innovation at Thermo Fisher. I've been in the industry going on now 27 years.

Background, I'm a chemical engineer by education, PhD. My experience, broadly speaking, is in small-molecule oral drug delivery areas like solubility, bioenhancements, spray drying, hot-melt extrusion, nanotechnology, as well as modified control release, complex dosage forms, etc., ranging from early development through commercialization.

I also lead a predictive modeling and simulation team at Thermo Fisher. We are a global team. We deal with the predictive modeling and simulations for drug product development.

That's a great question. I like to tell people when I first started working in this industry over 20-plus years ago, the number of candidates that were poorly soluble were estimated to be 30% or so. Now, depending on the source that people quote, 70% to 90% of molecules are poorly soluble, so it's more than doubled, if not tripled.

It's a challenging environment and the probability of success of getting a poorly soluble NCE to a commercial product is pretty low. It's way less than 10%, probably more like 5%. So by leveraging some of these AI/ML tools that we've developed, we aim and strive to reduce the risk in product development.

We can get to accelerating the early phase especially by leveraging these 

 AI/ML tools for selection of technologies. What's the best technology that might work for any given molecule? And it's all dependent on the type of molecule. Every molecule is unique.

And also in formulation selection, for example, we deal with complex modalities like protacs, which are on the larger end of the spectrum, and the traditional small molecules. And all these molecules violate, typically, Lipinski's rule of five. So they're literally the brick dust, greaseball-type molecules. By leveraging these AI/ML tools we again, as I said, reduce the risk in product development, get to the milestones faster, and shorten timelines, etc.

The big advantage is time and cost savings and resource savings, for the industry and our customers. Time is money, literally speaking, and what we bring to the table is reducing risk in product development, saving time and resources, minimizing API consumption. It's an ultra-material sparing approach by leveraging some of these 

 modeling tools like AI, leveraging AI/ML for supporting product development and formulation development.

Every API, as I mentioned, is unique. So your structural features are very important, obviously the chemical structure, also fundamental properties like the melting point which have bearing on the crystal lattice, [the] energy of the molecule [is] important. Other properties like pKa, LogP, LogD, all of those are important.

Also, things like permeability are important. So typically we try to get data measured in 

 assays like KCO2 or MDCK, and we can leverage that to predict what's happening in the human body. These are some of the features that are important in some of these calculations that we do.

Sanjay Konagurthu of Thermo Fisher Scientific discusses how AI and ML can help solve the dilemma of poor solubility.

Leveraging AI/ML to Reduce Risk in Drug Development: Part One

Mechanistic models and hypothesis-driven strategies generate optimized efficient solutions for drug development, says Catalent’s Nathan Bennette at AAPS PharmSci 360.

Shifting from Trial-and-Error to Rational Design Drives Efficiency in Pharmaceutical Formulation

American Association of Pharmaceutical Scientists

, director, Scientific and Technical Advisory, Catalent Pharma Services, discusses a paradigm shift toward rational design—a method grounded in conceptual and mechanistic modeling. According to Bennette, the rational design approach brings clarity and efficiency to the development of new drug 

Instead of the traditional trial-and-error method, Bennette advocates a structured approach. “The rational design concept is to use models—conceptual models and mechanistic models—to develop more focused hypotheses and then targeted experimentation to more efficiently get at the solution,” he explains. This shift begins with the creation of conceptual models to understand the problem in depth, followed by mechanistic (quantitative or mathematical) models that reveal the fundamental physical and chemical processes at play, he explains.

Semi-empirical models, often realized via computer simulations, Bennette notes, further refine hypothesis testing and solution optimization. Bennette shares a recent example in which his team faced a drug with poor oral absorption. Through 

 experimentation, the team isolated the rate-limiting step—diffusion across the unstirred water layer—and designed formulations specifically to enhance that process. Bennette notes that the 

How are rational design models transforming pharmaceutical problem solving?

Bennette believes this efficiency-first mindset yields tangible industry advantages. “If you can come up with these really focused hypotheses and work on those problem statements, you get there faster. You get there with less expenditure of time and energy and resource, and, ultimately, I think you come up with a better product,” he says.

Bennette adds that, as pharmaceutical companies face mounting pressure to deliver innovations with speed and efficacy, rational design frameworks promise to shape industry standards for problem solving and formulation optimization.

Nathan Bennette, Director, Scientific and Technical Advisory, Catalent Pharma Services

Bennette is a subject matter expert on establishing developability and non-parenteral formulation strategies for poorly bioavailable molecules, including protein degraders and others described as “beyond rule of 5”. He has worked in drug development for over 15 years in a range of technical and leadership roles and has broad experience with formulation of small-molecule therapeutics. Bennette applies his passion for mechanistic understanding of drug delivery to design systems tailored to each molecule and target product profile, utilizing a range of oral technologies such as amorphous solid dispersions, controlled-release platforms, nanocrystalline formulations, multiparticulates, and lipid-based systems. He has experience working across the entire drug development lifecycle from preclinical and early phase development to late phase clinical studies and commercialization.

Hi, Nathan Bennette, I work in our scientific advisory function at catalent pharma solutions, and my role there is to be sort of an in house consultant working with customers to define the problem statement and develop technical strategy for formulations.

So in sort of traditional problem solving and pharma development and formulation and process development in particular, a lot of times, we have historically done a lot of trial and error, where you form a hypothesis, but then you run a large number of experiments trying everything under the sun to develop a solution to the problem. The rational design concept is to use models, conceptual models and mechanistic models to develop more focused hypotheses and then targeted experimentation to more efficiently get at the solution.

The question, it's principally around efficiency. Those trial and error experimentation methods require huge amounts of effort and to arrive at a robust solution eventually. But these mechanistic, driven, hypothesis, driven experiment designs allow us to use the best practices in science to build out our understanding of the system and more efficiently come to a more optimal solution.

I think of it in sort of three layers. First, you have conceptual models, you have to form a picture in your head of what is actually happening in the system that you're studying. Once you've done that, you can start to form a real hypothesis about what might be going wrong with your with your molecule, for example, as you're trying to develop your formulation, all formulation development is a problem solving exercise, and it's important to know what problem you're trying to solve. So this conceptual model allows you to have a framework for forming a hypothesis. Then there's another level that I call mechanistic models. Some people call them quantitative models, mathematical models. These are the basic physics that underlie the physical situation of your drug molecule and the question you're trying to answer maybe it's from an absorption perspective or from a stability perspective. And so once you have your quantitative mechanistic model, you can start to form a more specific hypothesis. How fast does this process happen to? What extent can we expect this process to proceed. And then finally, we have sort of empirical or semi empirical models where that mechanism and concept have been reduced to an actual computer program, and you can run simulations. So these tools are all available for a range of different questions that are relevant to pharma development, like pharmacokinetic absorption or chemical stability or process modeling for different pharmaceutical manufacturing processes. So we use the starting from conceptual to mechanistic to semi empirical models. We use this whole framework to describe the problems that we're trying to solve and develop actual formulations to resolve that.

I have a really good one from recent work that does both. It helped us to get much more quickly to an optimized solution that actually we tested it out in vivo, and it worked really well. So we were developing this compound that had very, very poor oral absorption. And we asked the question first, why? And we started with the conceptual model that every molecule has to get into the GI tract. It has to dissolve it then has to diffuse across the unstirred water layer, up against the epithelium and then partition across the epithelium to get into systemic absorption. So that's our conceptual model. And we asked, Well, why is this molecule having a hard time doing that? What? What is the fundamental rate limiting step? And we learned through a series of in vitro experiments that the rate limiting step was diffusion across that unstirred water layer, the mucus layer. And so with that in mind, we didn't, we didn't go out and try every formulation concept under the sun that would be a trial and error approach. Let's try this. Let's try this. Let's try this. We said, Okay, we know the key problem is diffusion rate. So what formulations can we develop that enhance the diffusion rate? Because that's the key step. So we went in the lab and we developed a few formulations that we proposed would help with that, and we tested them in vitro and demonstrated that they did, in fact, enhance the diffusion rate. And then we took the best of that set into an in vivo study. And what was really cool was that the extent to which. We enhanced diffusion with those formulations, in vitro correlated very well with the in vivo enhancement factor as well. So we saw, we figured out what the right problem was. We developed a conceptual framework for how to resolve that, and we went and tested that specific hypothesis, and much more quickly, much more efficiently, came to a workable solution.

I think I've been hitting a lot on the efficiency, the efficiency of the process of coming with a workable formulation. I think that translates to faster timelines if you're if you're spending enormous amounts of time trying everything in order to find the one thing that works. Then you spend a lot of resource. You spend a lot of time working on stuff that isn't isn't going to be the answer. But if you can come up with these sort of really focused hypotheses and work on those problem statements, you get there faster. You get there with less expenditure of time and energy and resource and ultimately, I think you come up with a better product. It's a it's a more optimized outcome.

Motor neuron connecting to muscle fiber, 3D illustration | Image Credit: © Julian - stock.adobe.com

A positive opinion from the European Medicines Agency’s Committee for Medicinal Products for Human Use (CHMP) has moved a higher-dose regimen of nusinersen closer to approval in Europe, according to a Nov. 17, 2025 announcement by Biogen. The committee’s positive opinion signals a potential shift in treatment approaches for spinal muscular atrophy (SMA), the company noted in a press release (1).

On the back of the committee’s recommendation, a 

 is expected from the European Commission in January 2026. The recommendation reflects a growing interest in dosing strategies that may offer expanded therapeutic benefit for individuals diagnosed with the most common form of the disease, 5q SMA (1).

SMA is a rare, genetic neuromuscular condition that affects people across age groups. With treatment landscapes evolving rapidly, decisions around dosing, long-term outcomes, and real-world effectiveness have become increasingly central for clinicians, researchers, and 

“While we’ve seen great progress over the past decade [Table], there is urgency to do more to address the clear unmet needs of the SMA community,” said Priya Singhal, MD, MPH, executive vice president and head of development, Biogen, in the press release (1). “The CHMP’s positive opinion for the high dose regimen of nusinersen represents a promising advancement in our commitment to support the evolving needs of individuals living with SMA and deliver therapies that can enhance patient outcomes.”

Biogen licensed the global rights to develop, manufacture, and commercialize nusinersen (brand name Spinraza) from Ionis Pharmaceuticals (1).

What study findings are shaping expectations for clinical impact?

The CHMP’s opinion is based on data from a multi-part Phase II/III study (DEVOTE), which explored safety, efficacy, and pharmacokinetics of a higher-dose regimen across treatment-naïve individuals and those who had been receiving the existing 12 mg dose. In treatment-naïve infants in the pivotal Part B cohort, investigators reported statistically significant improvements in motor function relative to a matched sham group drawn from an earlier study. Participants treated with the higher dose also experienced a reduction in the risk of death or permanent ventilation. These findings underscore the potential for dose intensity to influence therapeutic outcomes, an area of active discussion within neuromuscular research (1).

In Part C of the study, which followed individuals who transitioned to the higher-dose schedule after years on the standard regimen, investigators observed motor function improvements during the first 302 days of treatment. Because long-term durability and safety remain important considerations, the extension phase of the study continues to assess outcomes over time (1).

“The positive CHMP opinion for the high dose regimen of nusinersen is an important milestone for the SMA community,” said Eugenio Mercuri, MD, PhD, professor of Pediatric Neurology at the Catholic University, Rome, Italy, in the release (1). “Based on the results from the DEVOTE study and my experience with patients receiving this novel regimen, I am confident that high dose nusinersen has the potential to bring meaningful benefits to people living with SMA.”

How could a new dose option affect global regulatory decisions and care models?

Nusinersen is approved in more than 71 countries at its current dose, and the higher-dose regimen has recently received approval in Japan, according to Biogen. The company also stated in its release that regulatory reviews are underway in multiple regions, including the United States, for which a decision is expected in April 2026 (1). If additional regions move forward, health systems may see shifts in treatment planning, administration logistics, and long-term monitoring practices. Intrathecal administration, which requires specialized experience, will remain a key implementation factor for hospitals and clinics (3,4).

Safety observations from the study were generally aligned with both the known profile of nusinersen and common clinical events in SMA populations. Ongoing surveillance will continue to shape clinician guidance, particularly around procedural considerations and laboratory monitoring (1).

 the evidence, decisions around dose flexibility may influence how clinicians personalize therapy and how drug developers design future trials aimed at optimizing motor outcomes, durability, and safety for diverse SMA populations (5).

High Dose Regimen of Nusinersen Receives Positive CHMP Opinion for the Treatment of Spinal Muscular Atrophy

. Press Release. Nov. 17, 2025.
2. Servais, L.; Day, J. W.; De Vivo, D. C.; et al. Real-World Outcomes in Patients with Spinal Muscular Atrophy Treated with Onasemnogene Abeparvovec Monotherapy: Findings from the RESTORE Registry. 


3. Finkel, R. S.; Ryan, M. M.; Pascual Pascual, S. I.; et al. Scientific Rationale for a Higher Dose of Nusinersen. 


4. Cordts, I.; Lingor, P.; Friedrich, B.; et al. Intrathecal Nusinersen Administration in Adult Spinal Muscular Atrophy Patients with Complex Spinal Anatomy. 


5. Finkel, R. S.; Day, J. W.; Pascual Pascual, S. I.; et al. DEVOTE Study Exploring Higher Dose of Nusinersen in Spinal Muscular Atrophy: Study Design and Part A Results. 

U.S. FDA Approves Biogen’s SPINRAZA™ (nusinersen), The First Treatment for Spinal Muscular Atrophy

. Press Release. Dec 23, 2016.
7. Ionis. 

SPINRAZA® (nusinersen) Approved in the European Union

. Press Release. June 1, 2017.
8. Novartis. 

AveXis receives FDA approval for Zolgensma®, the first and only gene therapy for pediatric patients with spinal muscular atrophy (SMA). Press Release

FDA Approves Oral Treatment for Spinal Muscular Atrophy. Press Release

FDA Approves Genentech’s Evrysdi (risdiplam) For Use in Babies Under Two Months with Spinal Muscular Atrophy (SMA)

Articles

The European Medicines Agency's Committee for Medicinal Products for Human Use has given a positive opinion on a higher-dose regimen of nusinersen for spinal muscular atrophy (SMA), potentially altering treatment strategies. This decision, based on the DEVOTE study, showed significant motor function improvements and reduced mortality risk in treatment-naïve infants. The higher-dose regimen, already approved in Japan, is under review in other regions, including the U.S. If approved, it could impact global treatment planning and administration practices, emphasizing the need for specialized intrathecal administration expertise.

CHMP supports an enhanced nusinersen regimen, according to Biogen, indicating possible SMA treatment shifts and international regulatory decisions.

Implications on SMA Care of High-Dose Nusinersen

At AAPS PharmSci 360 2025, Elly Zhou says digital twins helped forecast the effects of drugs on human drugs via a digital control arm.

Achieving a Personalized Digital Control Arm for Preclinical Drug Evaluation 

In this interview regarding the poster presentation “A Novel Approach to Preclinical Drug Development Using Digital Twins of Ex Vivo Human Lungs” at 

, Elly (Xuanzi) Zhou, MS, PhD-Candidate, University Health Network, Toronto, Canada, discusses a project that seeks to address significant limitations inherent in traditional randomized control trials, namely, she says, the limited availability of human organs for testing and the high inter-subject variability that often complicates result interpretation. Furthermore, Zhou notes that conventional evaluation models like animal models or organoids often fail to fully capture the complexity of the human system.

The central innovation of her research is using digital twins to create a personalized digital control for every organ being treated. The digital twin leverages large-scale multimodal datasets to train machine learning models that accurately forecast future lung and organ function, Zhou notes. Specifically, after a human lung receives a drug treatment, the digital twin generates the hypothetical untreated effects—the outcome 

 the organ was not treated. This, she adds, serves as a robust digital control arm, enabling direct comparison between the observed treatment effects and the digital twin-generated untreated effects within the same lung. By reducing the reliance on traditional cohort-based analysis and the number of human lungs needed for control groups, this approach is designed to reduce study size and accelerate clinical trials, according to Zhou.

The foundation of the data collection is an

 lung perfusion (EVLP) system, an artificial circuit developed by thoracic surgeons at Toronto General Hospital in 2008. EVLP keeps human lungs alive outside the body by perfusing the organ through vasculature and keeping the lungs ventilated at 37 degrees Celsius. This method allowed Zhou and colleagues to test human lungs directly with a condition of interest, such as perfusing a lung with disease on the circuit. The resulting data are considered highly valuable because they are“clean and capture the lungs in isolation,” free from the systemic effects of the rest of the human body, Zhou explains.

The models incorporate multiple layers of data, including lung physiology (air pressure, compliance, gas exchange), biochemistry (pH level, electrolytes), imaging data (X-rays taken during EVLP), proteomic analysis, metabolic biomarkers, and transcriptomic pathways, says Zhou. This rich dataset is used to generate personalized forecasts. As she summarized,"If you are treating a lung in real life,… you give the treatment and then you observe the treatment effects in terms of organ function and behavior, but... digital twins… generate… untreated effects to serve as this personalized digital control for this organ being treated." Furthermore, the EVLP system demonstrated utility in Zhou and team’s study beyond testing, having been used as a therapeutic platform to perform gene editing using CRISPR technology on human lungs.

*Editor’s Note: This transcript is a direct, unedited rendering of the original audio/video content. It may contain errors, informal language, or omissions as spoken in the original recording.



 Hi everyone, my name is Elly. I'm a third year PhD candidate at the University of Toronto in Canada, and my supervisor is Dr. Andrew Sage. And my project is focused on developing digital twins of ex vivo human lungs to potentially accelerate preclinical drug development.

Yeah. So if you think about a randomized control trial where you have the two arm design, you have the control and the treatment. So essentially what happens is that we face two major challenges. The first is the limited availability of human organs for testing, and a second is, we're looking at a significant variability within the inter subject conditions that can cause the result interpretation to be very difficult.

So what digital twins can do is that it can generalize a personalized control for every organ that's been treated. So essentially after you provide a drug to treat the human lung, and then what you observe is essentially the observed treatment effects. And then for the digital twin part is that it's going to generate the untreated effects as if the organ was not treated.

So essentially this will allow us to reduce the reliance on a cohort based analysis where we can do direct comparisons between the untreated and treatment effects within the same lung. So this will reduce the study size because this will reduce the number of human lungs we use as controls. So in that way we'll be able to have a faster clinical trial.

So if we think about the traditional drug evaluation models, for example, animal models or organoids, they typically don't fully or often fully capture this. Complexity within the human body or system. So what digital twins can do is that we actually leverage a large scale multimodal data set to train the machine learning models to accurately forecast lung function, organ function of future time points.

So in this way we'll be able to essentially say what is the untreated effects of this drug in this given lung that's been treated? And in addition on this EVLP system, ex vivo lung perfusion, we’ll actually be able to test human lung with the condition of interest directly.

Like for example, we can perfuse a human lung with disease directly on the circuit and observe its behavior, the organ function.

So for this model, we actually incorporate multiple layers of data. So we have lung physiology data, for example, lung, air pressure, compliance, gas, like oxygenation exchange and also biochemistry data such as pH level, acid base and different electrolytes. We also have imaging data, x-ray images because on this EVLP profusion system, we can actually take x-ray images of the lungs.

And in addition to that, we also have proteomic analysis metabolic biomarkers and also transcriptomic pathways. And they have all been incorporated into the model.

Yeah. And so what digital twin does is its going to serve as this robust and also effective digital control arm in the study. So if you can imagine if you are treating a lung in real life while you give the treatment and then you observe the treatment effects in terms of organ function and behavior but what digital twins do, is that it's going to generate this, I guess, untreated effects to serve as this personalized digital control for this organ being treated. So by doing this comparison between the treatment effects and digital twin-generated, untreated effects, we can actually do a direct comparison to say, what is this drug's impact on this organ been treated.

Yeah, so what EVLP system is Ex vivo lung perfusion, which means we're trying to keep a set of human lungs alive outside the human body. So what we do is that we have this artificial circuit that was originally developed in 2008 by a group of thoracic surgeons at Toronto General Hospital. So what we do is we perfuse the organ through vasculature, keep the lungs ventilated through the trachea, keep everything at 37 degrees Celsius in a physiologically protective environment.

So during the process we'll be able to collect a set of clean multimodal lung function data. That can be used to train the digital twin model. And also this data is very clean in the sense that we're actually just looking at the lungs in isolation without systemic effects of the human body.

And in addition to that, at our hospital, we actually have demonstrated that the EVLP system can be used as a therapeutic platform in the work or in the study led by Dr. Shaf Keshavjee at Toronto General Hospital, we have shown that you can actually do gene editing on a set of human lung using CRISPR technology during EVLP