Thermo Fisher Scientific Inc. (TMO) Earnings Call Transcript & Summary

[Audio Gap] Headline, and I'll be your moderator today. I'd now like to introduce the speakers for today's session. First, Sabine Zollner is Global CMC and Process Validation Lead in the Pharma Services Group of Thermo Fisher Scientific, where her role is to provide cohesive internally and client-aligned CMC commercialization strategies for late-stage drug development programs. Secondly, Christy Eatmon is the Global Subject Matter Expert for Sterile Drug Products at Patheon, which is part of Thermo Fisher Scientific. She supports the global sales and business development teams in providing technical support, designing strategies and supporting new business opportunities for the sterile manufacturing business. And Daniela Decina is Senior Director of Regulatory Affairs and Head of the North America Regulatory Affairs Function at Thermo Fisher Scientific. And with that, I'd like to hand things over, Sabine, Christy and Daniela, over to you.

Great. Thank you, Andrew. Thanks for the introduction and for getting us started today. So let's get started, firstly, just by defining what PPQ is. So the official definition is process performance qualification. We -- in drug product, we kind of interchange that with PV, process validation, but the same thing is true. Our goal is to really execute a study, which shows that we have a robust, repeatable process and that we're running in a state of control basically that the process is validatable. So how do we validate processes, the same as we'd validate anything else. We do it many times, we gather data, and we prove that everything fits into a specific design space, we have control. So executing PPQ is critical for commercial approval. This is how we prove the process is at a state where we can get this to market, right, where regulatory bodies, and other regulatory bodies, FDA and other regulatory bodies around the world can improve and say, yes, we approve this product to go out to the general population. And it demonstrates that we have also long-term stability, right? So we can set a shelf life for these processes and products. You can see here at the bottom of the screen, we have kind of like the development life cycle from preformulation, drug discovery, preclinical up through clinical, early stage and late stage. And then PPQ comes after. So PPQ is really the stage gate into commercial manufacturer. Typically, our PPQ batches are used for commercial launch. There are some exceptions to that, and Daniela will get into that a little bit. But this is really our way to get ready for commercial production and to prove that we understand this product and its processes from start to finish and all the intricacies of the whole process. The time line for PPQ will it can range from 12 to 24 months. And why we make this, we state this big range is because execution may only take some time, but we need to also gain stability. So typically, we'd run 3 batches and maybe those would take just a few months to run. We run them consecutively. But then we have to gather all the data, we have to put them on stability and then file with stability data. So typically, and again, Daniela will bring this up. But typically, we want around 12 months, the more the better, but at least 12 months of data in order to file. So -- and today, we'll be talking about large molecule biologics. So by filing, we are referring to BLA typically. So in 12 to 24 months, again, some of that time line is taken up in execution, but the majority will be in stability. So with that, I'll hand it off to Sabine and she'll get into a bit more details about all the different specific parts of executing PPQ. So Sabine, take it away.

Yes, happy to do so. This diagram depicts the swim lane of required PC and PV studies between late-stage process development and PPQ for BLA filing and chose the flow of complex overlapping and parallel PD activities relative to each other and their linkage. And I will start to explain in the left upper corner. As the prerequisite for PPQ specific PC/PD studies are conducted at bench-scale using a qualified scale-down model. For PC work, this scale down model is typically used to characterize and study all process parameter for impact on product quality and process performance and defines boundaries and limits. This information is formalized in a final process risk assessment, basically a failure mode evaluation assessment, ranking each process parameter with respect to criticality to product quality and process performance. Consideration to equipment and facility requirements and limitations, but also operational control during execution is also applied during this FPRA process. Based on FPRA outcome, the process control strategy is defined which drives execution of the PPQ campaign and requires confirmation at the end of PPQ for successful validation for BLA filing. For PV, as per our standard bench scale study entail process characterization of individual unit operations, using QPD approach, process stability, hold stability, impurity clearance [indiscernible] reuse cleaning and storage, reprocessing and working serving stability, the results are considered a prerecorded for at-scale process validation studies to provide boundaries and confirmation by the at-scale studies and derisk them. That's very important. At the start of the at-scale PV studies, a formal risk assessment and gap assessment is performed to assess and define the requirements for the actual processed validation plan as the guiding document for the individual studies during PPQ. The main scale -- the main at-scale PV studies can be divided into activities prior to PPQ and studies during PPQ, the headers are labeled blue in this slide, the PV activities prior to PPQ typically leverage pre-existing in-house data from process and product diagnostic studies, then alone studies. If required -- if equipment is already fully qualified, they can also be done with program specific. They can also be done program-specific using representative GMP material. PV studies during PPQ are typically conducted as product and process-specific studies unless client data can be used to justify, not to repeat. Leveraging preexisting data and reduction of the actual number of studies during PPQ significantly reduces operational complexity, sampling and testing, manpower need and save cost for both CDMO and client. The PV report summarizes the data from all at-scale studies as deliverables for BLA submission. Additional PV support studies, labeled with the orange and green headers pertain to risk assessments for impurities, stability for working serving a drug substance, analytical method, equipment, cleaning validation as well as raw material validation. Okay. We can go to the next slide. I will now give a few more details on highlights for PPQ strategy. So I'm starting from the left, the labeling is basically adapted to the previous slide to enable direct spotty. PPQ of USP and DSP process that is required for the process validation studies at scale, running at least 3 runs to prove for a producible and consistent process performance product quality. As I already said, for mixing and microbial hold study, we aim to leverage a stand-alone study using surrogate solutions and strategies for bracketing and worst case conditions to the largest possible extent to reduce PPQ complexity using really a risk-based approach. For mixing, we cover process solutions for buffer and process intermediates with available data, mixing studies for media or certain critical mixing steps like BDI or VF, have to be done per study and repeated. For microbial hold, we aim to cover process solutions for process intermediate, media and buffer. Typically, for process intermediate, we do a product-specific stability study to confirm maximum stability for all process intermediates from small-scale studies in representative containers. Similarly, the at-scale protocol for the concurrent validation of resin and membrane reuse it tends to repeat the boundaries defined by small-scale studies for maximum reuse cycle, efficient cleaning and storage conditions. Validation of reprocessing refiltration of all process intermediates is covered with small-scale studies and additional at-scale protocols for concurrent validation for high-risk filtration, such as viral filtration and BDS are opened. Shipping validation takes place during PPQ to cover shipping lanes and conditions for all seasons, BDS homogeneity is considered critical and conducted as product-specific PV study to prove homogeneity. CPV start after PPQ, and we include typically the results in PPQ and defined a plan based on a revised process control strategy after PPQ outcome. Moving over to the orange study. [indiscernible] validates the maximum cell generation number to prove working setting stability, purity and identity from sow to harvest. Risk assessment for elemental impurities and extractables and leachables are done to evaluate the safety risk imposed by raw materials used in the production -- in the manufacturing process. We cover PI stability during repeated [indiscernible] cycle. We measure the LP detection by transmission electron microscopy to prove absence of virtual vile like particles taking into account the maximum clinical dose. And we have minimum requirements for stability for pre-PPQ batches. Moving over to the green box. Facility, utility and equipment, IQ/OQ/PQ to prove suitability for its intended use in the process. And can -- and that the equipment can be properly cleaned will be done and equipment summary report will be written. Raw material qualification will be conducted. And last but not least, very important is analytical method qualification for in-process testing and validate for release testing to prove [ analytical ] comparability for pre-PPQ material, maybe put on stability or after introduction of changes. We move to the next slide. We go into the details of the top risk, how we go on a summary slide for the top risk complicating PPQ such as process complexity, equipment limitations raw material variability, critical process parameter and changes to scale up or technology transfer. If you move to the next slide, I will start with process complexity for biologics. Today, I will mostly talk about case studies, but also summarize more general observations made across programs. So the first topic is multiplexing at multiple cycles per unit operation and [indiscernible] sampling. This really increases operational complexity with more than 1 [ C-train ], sometimes 2 or even 3 and subsequent production bioreactors, precisely staggered going into a staggered downstream processing in stagger processing steps. This complexity is often exacerbated by having to run several cycles per step in a staggered manner, of course, until merged. This imposes really executional complexity and constraints during PPQ and leads to excessive in process and PV sampling. Even without multiplexing and upstream execution of multiple cycles per step in BSPs is already complex. An additional risk, and that's not worthy, occurs if 1 [ C-train sails ] and you should have a validation strategy and risk mitigation plan in place, because typically, the same process control strategy should be applied. Moving over to complexity of individual process validation studies and excessive sampling -- as I said, historically, we have done -- this has required a high number of individual process-specific studies for nonroutine work with increased sampling and testing, high workload and additional PV stuff and witnesses on the floor to prevent errors during PPQ. But now with leveraging our new standard PV strategy for mixing and microbial hold for process intermediates and buffers using this packeting approach, employing worst-case solution, we could significantly reduce workload. And as an example, plying the PV strategy for microbial hold on process solutions and process intermediate, conducted as a stand-alone study led to a reduction of number of studies by 75% and number of samples by 60%, hence, reducing significantly complexity. A case study is -- I'm moving over to the next case study. It's -- Basically, it has been seen that over control of operational execution and narrow control limits and ranges for, in particular, nKPP and pose high risk. A high level for a project, the high-level control was introduced by defining unnecessary -- unnecessarily restrictive ranges for operational process parameters designated nKPP, hence, without any proven impact on product quality and process performance. Over controlled parameters were [indiscernible] speed derive data from automatically controlled sources and individual routine operation had to be recorded multiple times for those nKPPs. This led to unnecessarily high number of deviation outside of the defined ranges. And it's very complicated in execution as it was really wasting time to complete that record, imposing increased operational constraints. So the lessons learned is do not include time limitation requirements for routine work and be careful assigning 2 strict ranges, avoid over control of processing time for routine activities. The last study is also -- the last point covers case study where basically the validated hold times versus we're not matching to the operational execution requirements in terms of timing. So the overall validated hold times with below 12 or 24 hours did not allow for sufficient time for operational execution, particularly for most of the multiple cycle, chrome steps and complex equipment set up. So this is basically describing process complexity for biologics. I hand it back to Christy.

Great. Thank you, Sabine. We'll talk a little bit about the complexities -- process and product complexity for drug products. Now we know drug product steps aren't just nearly complex as drug substance. But there are a few things that can add some complexity to drug product processing. A standard process for drug product is that we would receive the drug substance as a frozen liquid, we would thought we would pool it so mix it together, filter aseptically and fill in an aseptic environment. There are a few changes in that standard process that can add some complexity. So the first is when we scale up. So maybe in clinical or even in late based clinical drug substance may have come to us in the drug product site in bottles like 1- or 2-liter bottles and maybe for scale up to go to PPQ since we need to execute PPQ at the commercial sale, they may switch over to bags, something like a 10 to 12, even larger bag. So there's a bit of complexity there, determining how to thaw, the length of the thaw, the thaw conditions, again, making sure that we keep everything homogeneous. So we might have to purchase or use some sort of [ wave ] mixer as part of thawing or a controlled thaw unit and sometimes that large commercial scale, large [ cryotrovessels ] will be used for several hundred liters of drug substance material. So again, that thaw process would add some complexity and some automation to the general process. And we have to make sure that we understand the length of time to hold times at 5 degrees in that room temperature and then also ensure that as we're thawing, we're not doing anything that could be detrimental to the product. We have to maintain cleanliness as well to make sure that once we switch from bottles to bags, we don't have issues with shipping or leakage and things like that. So that's one change that can happen between late phase and PPQ, which could add some complexity. Another is potentially a change in formulation or dose. And what we mean by that is more concentrated drug substance, maybe in late phase, our customers might kind of hone in on the dose that they need and determine the dose kind of later on. So understanding what changes there might be in the formulation, the concentration of the drug substance or in the dose-to-fill volume or sometimes we also change container change format. So that can be another late-phase PPQ change, which could add some complexity, specifically if we're changing from a vial-to-syringe. Now we're looking at different formats, different product contact parts and things like that. So whenever we scale up and move to from late phase to PPQ, we also have to do a gap assessment and an FMEA, so that we would address any risk in the from late stage to PPQ. FMEA will always be completed as a pre-PPQ activity. Typically, that's pretty simple if we're using standard processes and we're keeping everything the same from what we did in Phase III to PPQ. But whenever we make changes, that becomes a little bit more complex. And it can be hard to identify every single risk as part of the FMEA, because we haven't executed it in the same way, maybe on the same line at the same scale. So we definitely have to understand all the risks to make sure we address all those risks and make sure we have a thorough gap analysis and a thorough FMEA. Once we have all the risks assessed, then we can, of course, mitigate those risks. The other process complexity, not specifically listed here, but potentially it could be if we have to do any forward processing at the drug product site. So if we have to do something like a concentration adjustment, anything like that can add some complexity to a product, we have to make sure that we have the right IPCs in place. So that we know that we can execute again a robust process and hit the concentration at the first go around every time. Of course, as we execute PPQ on the drug product side, we have a lot of sampling that we wouldn't do on straightforward runs in clinical batches. So we have to understand what the sampling plan is, make sure we work with our validation teams to get a good sampling plan and then work with operations for smooth execution. During PPQ, we take many, many samples to prove that we have control throughout the whole process. We have to make sure that we identify those samples, have a plan for testing them, do everything in a timely manner because that's really going to be the data, the raw data that we capture in batch records, but also the data that we have from the laboratories to prove that our process is in control at every step of the way in that. During PPQ we're often challenging what could be the worst-case scenario. So we're going to challenge hold times, we're going to challenge mixing times, mixing speeds, filters on the pressure on the filters and things like that. So we really need to gather samples from that potential worst-case scenario so that we understand even when we're operating in that worst-case mode, we still are within control, and we still have a good high-quality and effective and safe product, right? So those are some of the process and product complexity impacts for PPQ for planning and also for execution. And I'll hand it over back to Sabine for more on drug substance.

Okay. I'm happy to talk about equipment limitations for biologics. And again, I will basically -- 5 case studies. The first is narrow process range is not matching with equipment accuracy. And this is a good example for an approved commercial process still prior to PPQ for scale-up. We experienced extremely narrow pH ranges for several buffers for chromatography step from one of the sending units with ranges of plus/minus 0.05, which was not possible to achieve with our fully automated system and equipment we wanted to use in our brand-new facility using a temperature adjusted calibration and automated measurement and adjustment. In addition, the ascending unit did pH measurement and adjustment manually using tempered samples at 25 degrees for an at-scale use at lower ambient temperature for use, where it's we adjust and use the buffers at the same temperature at 21 degrees. Hence, we were facing the risk that our targeted pH, the target temperature of 21 degrees may deviate substantially from the sending unit using equipment normalized to 25 degrees. And as pH adjustment is temperature dependent, in particular, for Tris containing buffers, which were used for those specific unit operations. This problem was really exacerbated as it applied to chromatography step also approved for vile clearance. Hence, we could not rule out an impact on vile clearance. Also impurity clearance, with -- and the process performance and product quality impact could potentially be impacted. So we did -- as a risk mitigation, we did a very, very thorough risk assessment in the study measuring actual samples, prepared by the sending units for their buffers at our location to bridge and adapt the specification in the PCS, process-control strategy and updated regulatory filing. So not spotting this prior to PPQ could have led to a true intrinsic failure during PPQ. I'm happy with it. Another case study is cooling capability, not matching short forward processing in hold times. Again, short overall hold times repeated pooling and mixing of cycle [indiscernible] or the use of short forward processing times might not match with the cooling capacity of the equipment used depending on the cooling system. So that's a general observation we had from several projects. So it means you need to have validated hold times that really match also the equipment capability to risk mitigate that. You should -- I'm moving over to transient temperature excursion handling. You should have a plan prior to PPQ how you handle transient temperature excursions as they typically occur for transfer of process liquids from a waste -- to let waste system or we see trends in dips in temperature or at cold-room temperature or temperature for heat treatment steps and the learning is really -- it is important to define clear well in advance what excursion should be allowed, how long and how many and clearly define them in the process control strategy. This really greatly reduces potential for deviation during PPQ. Another case study was the late introduction of brand-new fully qualified equipment. So we used a new fully qualified [ UFDS ] system was really introduced very late just prior to PPQ. And despite proper IQ/OQ/PQ, many software and control issues were experienced during PPQ and as the lessons learned in risk mitigation, do not push to implement a system without properly performed wet tests running the actual process itself prior to PPQ to confirm full functionality of the equipment. In a more recent case study, we had unexpected equipment failure despite appropriate regular maintenance. So during PPQ, using a complex perfusion system, the controller fail due to a burned-out circuit board of the control unit. This led to a slight delay in perfusion by a few days. So even though all PCS acceptance criteria for vital cell density and viability are met and additional duration was covered or is covered by the validated maximum generation number through [indiscernible] the representative for this PPQ batch can still be questioned and need to be thoroughly assessed during the risk assessment. So lessons learned have well established connections and workflow for equipment vendors for a fast turnaround time to repair even during a weekend. We managed to get this repaired on a weekend, on Sunday. Delay -- to keep the delay as short as possible. If no spare pieces are for equipment are available for several reasons, for example, used for a competing program cost consciousness for expensive equipment from the client. So I hand it over back to Christy for the PPE part.

Great. Thanks, Sabine. So when it comes to equipment limitations for drug product, what we mean is mostly the filling machine, so filling equipment. As I mentioned before, it's standard sometimes to move from a smaller mid-scale line in late-phase clinical to a larger scale line for PPQ. And the rationale for that is that we must do the PPQ batches on the commercial equipment with everything as it would be for commercial routine production. So typically, when we're moving from one line to another per PPQ, we try to match up the lines capabilities. So that means we would try to match up the dosing mechanism like if we're already dosing for a large molecule with a Peristaltic pump to keep it very gentle. We would also look for a large-scale commercial line that was driven by Peristaltic pumps. We would also try to do size the filters appropriately. So as we scale from small batches or medium-sized batches up to larger batches, we need to make sure that we size the filter appropriately, so that we don't get any filter clogging or even partial filter clogging, which adds more pressure to the filter and then we could have some issues during the fill. So we just need to make sure that the scale that the equipment matches the scale or the batch, right? So that we have the right size line that's most efficient. We have the right filters. We have the right mixing vessels in that we do studies to prove that as we scale up, we can transfer from small to mid to large scale. There's some ways to mitigate that using equipment that is made to scale. So some of our disposable mixing tanks they are made to scale from small to large with sort of the same energy or same RPM in terms of mixing to be able to understand or to test out, okay if we mix for 15 minutes, at this rate, at this scale, we can do the same. We still need to do studies to prove that, right? That's why we execute PPQ and take a bunch of samples from beginning, middle, end or top middle bottom of the batch to prove that we've successfully prepared a homogenous product. Another complexity when it comes to equipment is the viscosity. So I mentioned for monoclonal antibodies and other large molecules, similarly, we use Peristaltic pumps so that we don't increase this year, right? We need to reduce this year stress as much as possible to prevent aggregation. Aggregation is probably our top concern when it comes to large molecules and the handling of drug product. The drug substance team has spent a long time making this drug with substance. We want to make sure that everything we do on the drug product side is very gentle and does not cause any aggregation, any particular formation, anything like that right to the drug substance. So for these large molecules as they get more concentrated, it becomes a bit harder to maintain a product that doesn't aggregate. So we can do a couple of things here like slowing down our lines. We can maybe sometimes use larger needle sizes, play around with larger needle sizes to reduce this year. But it is important to understand the limitations of the equipment. So if we're going for a full tolerance range, we need to make sure that the equipment can match. What we've done previously. And as we get to smaller and smaller fill volumes, of course, that range -- or that tolerance range can get wider, right? For example, if we're filling 10 or 20 milliliters in a larger vial than getting ranges -- fill tolerance ranges of 1% and 2% are pretty easy. However, if we're filling something like an intravitreal product, which is 50 to 100-microgram microliter of dose rather, maybe we're filling -- we're overfilling, right? So maybe we'll fill the limitations of these equipment is usually in the 200 to 300-microliter range. So we're filling a very small amount. Now as that material gets more viscous as well for highly concentrated, it can be hard and harder to deliver an accurate fill weight. So those ranges would get broader, right? So we need to make sure that we're within the tolerance of what the machine can manage, right, what the equipment can manage. The good news for us is many of these filling lines are 100% recheck. So we can actually check the weight of every single mile and we can increase our yields by doing that. So those are just a some of the equipment limitations from a fill finish perspective, right, just making sure that everything matches up and that when we design the PPQs and the CQAs that those match what the equipment can deliver. And in most cases, for biologics, they can. We just have to watch out for very concentrated products as they get more viscous, those gentle filling machines, the peristaltics can typically deliver or manage up to about 30 centipoise, maybe up to 50 in some cases. And some of those very highly concentrated biologics you get into that 20 to 30 centipoise range. So we need to make sure we do a lot of filings studies to make sure that we have our recipes well developed and that the automation is running properly for those viscous products. And we often do that with a surrogate to derisk the PPQ batches. The other thing is a lot of these biologics will move into prefilled syringe in an auto-injector for commercial launch. So we need to make sure that we have all the functional testing done on the auto-injector. That's the understanding, again, the break, loose and glide force, things like that. So that we know that once we get past the filling part when the patient actually goes to use this device that again, we're not causing any aggregation or any high share there. So those are really the equipment limitations or format limitations for highly concentrated biologics. So pass back to Sabine for discussion on raw materials.

Raw material variability for biologics. I am going to touch upon 3 case study. The first one is actually the most interesting one. The role of trace elements in complex synthetic media impact and product quality. We -- during a PPQ after scale up, we observed a markedly higher variability of glycan side chain complexity with potential significant impact on product quality, because we saw a clear shift towards individual specification limits for process intermediates, but also final product. The likely root cause were raw material depletion after scale up, because we had a significantly increased productivity for the scale-up. As an in-depth root cause analysis for raw material it can be conducted in raw material analysis, and this suggests that, of course, media lot impact. After a thorough search and multiple single component analysis were conducted using what we designated good and bad media lots in terms of impact to glycan modification. And we could actually pinpoint trace element as a potential root cause. And as an additional risk mitigation activity, a batch record review of the selected media lots was done together with a media [ vendor ]. And after several rounds of investigation, we discovered a very consistent, but really insignificant difference was just a factor of 2 for trace element in the [ PBB ] range, but this was the likely root cause. Digging deeper, it was -- the cause for it was a [ vendor ] which changing from a specific media raw material component with trace elements as known impurities to this material to a material with higher impurity -- higher purity so less impurities of the trace element. And finally, the change back to the old vendor and spiking of the most suspicious trace element at-scale production run be able the final proof of concept and fully restore the desired glycan profile. The spiking experiment had to be done at production scale as the fully qualified scale down model was not sensitive to this phenomenon despite several attempts to optimize towards mimicking the scale observation. So this was quite a journey for the CDMO together decline and only really achievable through a relationship based on trust and partnership between both, because you can't really risk mitigate for that. Another case study, we experienced constraints of customer-specific raw materials with complex companion requirements. So in case the vendor has delivery problems or changes discontinue certain raw material [ vendor ] itself and can't supply -- can't guarantee supply of customer-specific specification materials with complex [ companion ] requirement, it's going to be a problem, and we observed that for PSA, where the vendor had to do very labor-intensive screening for the specific customer requirement. So risk mitigation is really -- if you can't find -- you need to -- the risk is not to find a second supplier -- the risk is really to find a second supplier that can provide compliant replacement under type time line. So risk mitigation is search for a second supplier in particular for customer-specific specifications, if the raw material itself is also designated a critical raw material in due time. And as a last case study, the use of raw materials containing animal-derived component should be avoided as it's bearing the inherent risk of advantageous agents, such as viruses and bacteria with undue risk for material and equipment contamination. So do not allow processes using animal-derived components and scrutinized for adequate replacement prior to GMP or PPQ. If you move to the next slide. Moving to critical process parameter as perceived risk for biologics. I will sight 3 case study. So limited PC work leading to inadvertently characterized and defined CPP impacting process performance and product quality. So very often, and we have seen this, the direct technology transfer of fully developed safe-stage programs that were externally fully developed and characterized. So without any PC work done internally to our internal standard really bears the risk of limited PC data and limited characterization. So the perceived risk was insufficient knowledge on process capabilities for impurity clearance and in particular, for HCP, which was designated the critical quality attributes for final COA throughout the process and then, of course, the mismatch with release certification, that or earlier, which can cause batch failure. This was observed for a program using accelerated approval track as a quick to clinic for the COVID indication. And as the lessons learned in risk mitigation really scrutinize available external PC data and squeeze in additional targeted PV studies or PC studies, looking at specific CQA even under tight [indiscernible]. Another case study, we saw that peak media components were indirectly impacting CPP. So during PPQ, a gradual, but marked change for production bioreactor pH was observed upon media feed edition and consumption. The bioreactor pH itself was designated a critical process parameter. This phenomenon was not obvious from pilot batches or again observed in the scale down model work. Therefore, PC studies during the -- using the scale-down model had defined the [indiscernible] strategy as a non-key process parameter, so without any impact on process performance and critical-process parameters. And prior to PPQ, the feed strategy was adjusted within the defined normal operating and proven acceptable range. So what we saw during PPQ the potential cause was the secondary effect of the feed strategy change on the pH profile due to probably exhaustion. Exhaustion of a feed component, which was only observable at final scale. And the observed impact on product quality were changes in the impurity profile, product integrity and [ backing ] profile. Therefore, the feed strategy was changed back to the pre PPQ approach, lessons learned in request mitigation, do not change the strategy in the first PPQ batch even for an nKPP always confirm in the pre PPQ batch at the same scale that this change is acceptable. Pertaining to the last process over control with excessive number of key and critical parameter really to avoid process over control due to an excessive number of key and critical process parameter defined during the FPRA process. This imposes difficulties with regard to reproducibility and consistent operational execution during PPQ and routine manufacturing causing, again, unnecessary high number of deviation. So the definition of critical and key parameter should be solely driven by PC outcome and based on parameter severity only and not be intermingled with occurrence and detectability that you typically assess during FPRA leading to this additional KPP and CPP putting undue constraints for operational execution. And with this, I finalize and hand it back to Christy.

Great. Thank you, Sabine. I'll talk a bit about those critical process parameter challenges for drug product. And I think understanding and defining the CPPs are definitely part of the biggest risk for PPQs, at least in my experience. So products can come to us either through clinical, so we can have a product or clinical or something can come at late stage or as a product that's already on the market and come to us. And in any case to get to commercial production, we need to run PPQ batches. So there are times that there's a lack of historic data. So sometimes good data wasn't gathered from early phases and then we get to late phase or PPQ. And we don't have a lot of information to go on. We don't know how to establish the CQAs, what they might be or even set the ranges for CPP. So good collection of data even from early phase is helpful to establish what those CQAs critical quality attributes are and set CPP range is so critical process parameters. Now I will say there's probably not as many critical process parameters for drug product as there is for drug substance, but we still have to define those as part of PPQ. We have to include them as part of our validation package and make sure that we do have control of those CPPs and understand the CQAs as we execute PPQ. So that can be one of the biggest challenges that I've seen personally is to figure out what are these critical parameters, right? And how do we test them or strain them during PPQ to prove that we have a robust process. The other is process hold times, because many times, we just run our clinical batches through a standard process, right? So we're tracking the hold time, so we're not really challenging those. And there are different hold times, right, from end of thaw to beginning of compounding, from compounding until sterile filtration, from beginning of sterile filtration to end of fill to just name a few, and then total time out of refrigeration for products that their storage temperature is 5 degree C, right? So -- and in some cases, for biologics, early phase, maybe Phase I or Phase II batches might be frozen. So we might be lacking 5-degree data or 25-degree data. One way that we mitigate that is that we can do a small laboratory batch and just mimic what we would do as part of PPQ mimic those hold times to make sure that we have sufficient hold time to execute our challenges as part of PPQ. The good news is that for most monoclonal antibodies, they are pretty stable at 5 degrees C and at room temperature. So we don't have to worry too much about processing times. They don't face the same challenges as some small molecules do with like oxidation and things like that. So and stability and solution, right? They're already in solutions. So process hold time is how to set them, where to cut off one hold time and start the next can be a challenge. The other thing is, of course, time lines, right? We have to link up with a drug substance time line to understand when drug substance is going to be available. We need to understand if we're pulling lots of drug substance, right? In some cases, drug substance may make many lots that we pulled together one large PPQ batch. So you need to understand what the time line is -- and we -- Daniela will get into 3 consecutive batches and some of our customers to choose to do 3 consecutive batches. One way to mitigate rejected batches or repeated batches is to maybe make a first batch or make a feasibility batch or pre-PPQ batch and then understand what could go wrong during that batch and then sort of build the next batches with that understanding or create a little time between the first PPQ batch and the next 2. So if anything needs to be changed or tweaked then you can make those changes before executing the second batch, right? Batch rejections again is a risk, but one that I see very, very seldomly in drug products. Not that we don't have any deviation. Certainly, we might something that needs to be discussed or goes outside what the plan was. But typically, we can have an investigation and close those deviations without rejecting the batch. But part of that the way we mitigate batch rejection is to have a good back record, to have a good FMEA, address all the risks. And then on these PPQ batches, they really do take a lot of, I don't want to say, baby sitting, there might not be the right word, but they really do take a lot of organization, a lot of communication between various different groups because we'll have operations, we'll have a tech transfer team, a validation team and quality teams all working together to make sure these batches are successful. So there needs to be a lot of oversight from the technical team and from project management to make sure that everything is on track, that we don't deviate from the plan. So when we talk about changes in scale-up or technology transfer, again, these can come from external or internal, but it's important to, like I said, meeting all the stage gates ahead of PPQ. So within our network, we have sort of a bullet point or a checklist of everything we need to do to meet PPQ, right? Are these documents ready? Has this been done? Do we have product contact park compatibility, for example, do we have filter validation complete? Have we established the PPQs, CPPs and things like that? I mentioned it before, about identifying all the risks through FMEAs or gap assessments, right, even if we have an FMEA for a line, we have to still look at the product-specific requirements. We need to definitely manage those validation protocols. So within our system, the validation package I've seen is usually a huge binder filled with many, many different studies, the results in the summary. So we need to make sure that we manage all those documents, make sure that they're all completed in a timely manner so that we have the full PPQ package available when it's needed for writing up submissions. Daniela will talk a little bit about bracketing or matrix approach. So I may just skip that for now. But of course, we have to understand the design, right? Are we making 3 batches at the commercial scale? Are we making 2 high, 1 low or are we doing 1 low, 1 medium, 1 high? So we need to design the approach and have agreement from a validation from our customers and our internal quality. And then the other thing is that if we're bringing a product from a second source, which I mentioned, so maybe it's already out there in the commercial market, but we're bringing it to our facility as a second source. We have to understand what do we need to do at our facility that might be different from the existing, right? So it's good to have good transparent communication, even from what 1 CDMO has experienced to another so that we have a successful launch. And then, of course, managing different SKUs. Some products might only have 1 SKU, many of them don't, right? So we have to make sure that when we design the PPQs that we can be the most efficient if there's multiple SKUs, understand which is really the worst-case scenario, because nobody -- if you have 3 SKUs, certainly doing non validation batches, maybe the right way to go about things. But if there is a matrix approach that can be set, I think that's most efficient for everyone. So those are just some things to consider during [ TTs ], during prepping for PPQs from a drug product perspective. So now we'll pass it off to Dr. Daniela, who is going to talk a bit more about executing PPQs and preparing for submission from a regulatory perspective. So Daniela, please take it away.

Thank you, Christy. So once the -- all of the PPQ work and the validation work has been completed, the information has to be presented to a regulator and this is typically done in Module 3 or the quality or CMC section of the marketing application. And this provides the description of the executed process performance qualification studies and that includes a narrative and tabular summaries. And it must include data from runs at the intended commercial scale. This is for biologics, these need to be put into the Module 3 at the time of the marketing application. So the goal, of course, as said before, is to demonstrate manufacturing reproducibility and consistency of product identity, strength, quality and purity. The regulators are also going to review the executed PPQ evidence during a pre-licensing or preapproval inspection. So that's another place where the data will be looked at in more detail. Other validation studies will supplement the PPQ and look at the process consistency and be really a part of that consistency assessment and form the totality of the validation package. Some of these other validation studies can be at full scale, but regulators can accept scale-down models, where justified and that justification should be included in the dossier. Some examples in drug substance would be for scale-down models, resident membrane cleaning and reuse and viral clearance. In drug product, the aseptic fill sterility assurance needs to be in the package that would be at scale. And in drug substance and drug product mixing studies, hold-time studies and shipping will also be a part of the supplemental validations. Steps where reprocessing is intended to be permitted as a normal part of the process have to be validated, and they need to be submitted and approved with the dossier. And this is a quote from a recent example that we had from our client. In this, the FDA responded at a -- from a prefiling meeting and they said that they agreed with the potential reason for reprocessing. But they reminded the applicant that validation data are needed to demonstrate that those operations that are introduced in the reprocessing step do not impact the product quality. One tip that we would really have is to really include regulatory affairs in the core team for input into the PPQ strategy at the time of planning, so that the regulatory nuances can be incorporated into the planning of PPQ from the start. So when it comes to the number of PPQ runs and the regulatory guidance for this, most major regulators generally don't define the number of PPQ batches. And they also don't exactly explain what -- how to arrive at that number. Regulators want to know if the company has an assessment tool to generate a PPQ number. And the number of runs really should be justified in the PPQ protocol itself and discussed in the submission dossier. The common industry practice, however, really still is a minimum of 3 PPQ runs. But from a regulatory point of view, we would say that if you're considering less than 3, then really ensure that this meets the needs of your global filing plan. Because if this is satisfactory for the first region, it might not be satisfactory for subsequent regions, and you'll have to rethink the totality of the package subsequent to your first filing. The total runs that you would need to have may also reflect a bracketing and matrixing strategy, which we'll discuss now in the next slide. So when it comes to bracketing and matrixing, bracketing typically means that the extremes of a design factor or an attribute or studied such that the intermediate conditions between the extremes are really encompassed in the results. Some examples of this would be drug product strength, where you might have, although medium and high and studying the 2 extremes would encompass the middle or container sizes or fill volumes, those are all examples. So given the scenario here where a depyrogenation tunnel may be validated for the smallest and largest vial size. And this information might really be presented to a regulator in a master file. And then the clients, the applicants dossier would justify why their products container is bracketed by the studies already done on the depyrogenation tunnel. Matrixing in is a way to investigate the effect of interchanging 2 or more similar conditions while streamlining the study design. So an example could be interchangeable production suites or using equipment families. So in this scenario, I've given a drug substance facility perhaps has 2 interchangeable purification suites and call them Suite A and B. The PPQ strategy could consider reducing the number of runs. So instead of running 3 runs in Suite A and 3 runs in Suite B to a strategy such as 3 runs in A and 1 additional in B or a 2 in 2 strategy. So really using a risk assessment to justify this matrixing strategy and providing that in the submission would be a way to streamline the number of runs that you actually need to perform. And as a regulatory tip, if you're considering using as an example, more than 1 purification suite, and you have a matrixing strategy in mind, but you haven't completed all of the work in the initial -- for the initial marketing application. You can consider a comparability or post-market change protocol to go into that marketing application to propose the strategy and acceptance criteria for adding to matrix. The benefits of this are potential agreement from the regulators on a streamlined study and fewer runs for your post-market change and also potentially to produce -- to propose a reduced filing category once the data is available. So downgrading the post-market change category by providing the protocol upfront. It's really helpful in your dossier to highlight product and process knowledge that would support the risk and impact assessments. In terms of lot consecutiveness, this is a topic that comes up quite often. PPQ runs should be consecutive, but consecutiveness isn't always defining guidance. PPQ batches should be predominated and there should be internal procedures to define what situation would be acceptable if you changed, which batches you're going to consider to be your PPQ lots. Cherry picking or picking the best 3 runs is certainly not acceptable. It's hard to find an example of consecutiveness defined, but we do have a Health Canada example that we've given here. And in this, they explained that consecutively doesn't mean the batches need to be immediately produced one after another. But it does mean that every lot made in support of the qualification have to be a designated qualification batch from the outset, so prenominated and executed as such. If there's a production of a lot on the same equipment, where the manufacturing issues that are occur, for example, deviations if they are identified, these need to be reported and assessed in the validation. So there could be reasons not to include them in PPQ. In the dossier, we really also urge to you to be clear on the lot numbering methodology and how this relates to consecutiveness, because sometimes if lot numbers are assigned by an inventory system, they might not visually appear to be consecutive. So there are some reasons why you would exclude a batch from a PPQ sequence. As we've said, deviations can happen in the PPQ campaign, and some of these are extrinsic to the process. For example, if you had a power outage or some sort of a mechanical failure and some are intrinsic to the process. If the unit operation is not truly capable of a consistent control and the outputs can't meet the acceptance criteria. So those would be intrinsic. Regulators want to see that the deviations are evaluated thoroughly and objectively and to assess the impact to the qualification of that process. So for example, can a batch be excluded without restarting the PPQ count. The answer is potentially if deviations, for example, are extrinsic, and they're justified with a risk assessment. Robust impact assessment needs to be laid out for the reviewer. So in that risk assessment, can you draw upon your product knowledge and your characterization, what do you already know through development studies about your product and its characterization? The process knowledge and what you know of your critical quality attributes, for example, what was the basis for selecting control parameters. These things can be drawn upon from your development work to justify why a deviation is or is not negatively impacting the demonstration of control. In a gap assessment, what you don't know -- understanding what you don't know and what can be proposed to mitigate this, you would consider these factors and maybe plug in some additional work to close the gaps. Some examples might be some further characterization or electing to put batches on stability that were not originally destined to be on stability studies. And finally, we always encourage regulator engagement. Discussing the PPQ strategy with the regulator can really reduce risk in executing the plan and reduce risk in reviewing of the application dossier because the regulator already has a view into your strategy. There are some mechanisms such as ad hoc, topic-specific meetings during development, perhaps late development. Some examples are FDA Type C meetings or EMA scientific advice and then there are the dossier pre-meeting -- prefiling meetings themselves. However, if your prefiling meeting is scheduled after all of your PPQ campaign is completed and your data is gathered, that might be a little bit late in discussing the strategy. So just be aware of timing to get good feedback from the regulator in time to implement that into your plans. And as we've said before, consider the global filing plan for the product. Does that PPQ strategy meet the regulatory expectations in all of the jurisdictions that you want to go into? With that proposed bracketing or matrixing strategy be acceptable in each of those regions? Or might you have to consider further studies to have filing readiness? And if you would need to do that as an aggregate in front of your first application, would that impact the timing of that? Or would you do that extra work after your first application? And how would it impact the timing of the registrations that follow? Reducing regulatory risk is always good by seeking the regulator engagement. I'll pass it back to Christy.

Thanks, Daniela, and thanks, everyone, for joining us today again. So some key takeaways here. for PPQ execution are, of course, demonstrating that we have a robust, repeatable process, right? That's the #1 goal when we talk about PPQ. Again, the second is to include regulatory affairs. We do have Daniela leads up our regulatory team. So we do have staff here to run things by and to make sure that we're on target with what the agencies would expect. But just to understand and plan from the beginning, of course, not only look at purely one aspect of the PPQ over another, but the whole package, right, the whole process from end to end and also ensure we have good analytical data needed for PPQ. So I mentioned this because we didn't bring it up too much today on the webinar, but sometimes analytical can get kind of left off as like almost a necessary [indiscernible], and we have to do analytical, but those analytical results are really what's going to that we have a good process, because we're doing so much testing. So we will be doing another webinar in the future, specifically on analytical methods and hope you can join that one. But we don't want to think of analytical sort of as an add-on, right? We wanted to be a part of the process and one that's going to help us prove that we have a good, robust, repeatable, validatable process. And finally, we want to emphasize the partnership with CDMOs and sponsors, right? So we need to make sure that we're very collaborative that we have transparent communication between all teams, right, between our internal teams, but also between our customers. And one way to do that is make sure that we have project managers that can oversee all the activities, make sure that we stay on task and we stay on time. Also, another thing that we can do to mitigate risk is to integrate drug substance and drug product. It's not always possible. But when possible, if we link those 2 together, we understand each other's time lines, and we can reduce any risk for scheduling or things like that. Also, again, we can leverage regulatory support. So I see a lot of customers that sort of go about things, maybe based on what they've done in the past especially customers that have a lot of experience with commercialization. But we do know that the regulatory agencies they do grow, change, evolve over time. So it's good to have a team that are experts in that area that can advise us on sort of what they've seen with regulatory bodies, what the newest info is out there and how we need to satisfy everything the regulatory agencies are looking for. And then, of course, make sure that we plan right, plan early, even in kind of Phase II, Phase III, start thinking about PPQ, make sure that we're gathering good data so that we have a successful execution of PPQ and also understand where the risks are due to those FMEAs, understand where the risks are. I know that during COVID, we certainly challenged time lines how to accelerate even how we do PPQ, right? So the industry is changing on more of a risk-based approach. So it can help us in a lot of ways, right, it can help accelerate things, but we have to understand what those risks right, if we're taking a risk-based approach. So thanks for joining again. Those are some of the key takeaways today. Again, look in the future for our next webinar on analytical. And with that, I'll pass it back to Andrew to wrap us up.

Well, I'm afraid that we've run out of time there. I'd like to thank our speakers, Christy Eatmon, Daniela Decina and Sabine Zollner for that great presentation and our sponsor Thermo Fisher Scientific for making this event possible. On behalf of Thermo Fisher Scientific and Citeline, have a productive remainder of the day, and thank you for watching.