PODCAST: Where Programs Fail: Bridging Assay Development and Manufacturing for Lateral Flow

Trevor Brown: Hello, I'm Trevor Brown, the chief revenue officer at DCN Diagnostics, and this is Expert Insights podcast. Today I'm joined by our chief operating officer, Pat Vaughan, who runs operations across DCN's assay development, engineering, and manufacturing service businesses. Pat recently wrote a piece called Bridging Assay Development and Manufacturing for Lateral Flow Success, and it actually addresses one of the biggest pitfalls of rapid assay development programs. Pat, this is my first time hosting Expert Insights, and I'm excited for us to sit down and talk about this topic, which is near and dear to my heart, and I certainly know yours as well.

Pat Vaughan: Exciting stuff. We deal with it a lot at DCN with our clients and prospective clients. So welcome to the podcast, Trevor.

Trevor: Excellent. Thank you. Yeah, the handoff between development and manufacturing is something I've seen over and over again in my career in diagnostics, generally speaking, where the most friction occurs because both parties are working earnestly and trying to make this work. And I think getting some insights from your side of how this is going to translate from a lateral flow perspective would be very helpful for our audience. You make a claim early in the article that I think catches people a bit off guard. You say that most lateral flow programs that fail don't fail in the development phase. If they do, it's part of the process. Where that failure hurts is when that high-performing assay working at the bench starts to lose that performance when it comes to scaling up the manufacturing. So what does that failure actually look like in practice?

Pat: The physical failure can look like many things. Many scientists and manufacturers out there will recognize — it could be suddenly you get nonspecific binding that appears, and the R&D people are saying, well, we never saw that, but the manufacturing people are saying, yeah, well, we see that. Or suddenly you start to see a signal drift across the whole production run — the early strips are good, late strips are bad, or vice versa, right? So what used to work in R&D all the time, suddenly now you're seeing differences. It could be something like the material is breaking on the reel-to-reel, on the scaled-up automated equipment, and the scientists say, well, we never saw that in the lab. So it can be a lot of physical things, but a lot of the time, what's the pain point? The failure can be intermittent — it could be an irreproducible result on the manufacturing line. So the scientists are called back in, but they can't reproduce it, and they're kind of scratching their head as to what this is, because you can't just summon it on demand, basically. So there's a lot of probing around. Obviously the R&D scientists are going to start chasing the chemistry, right — that's just their first point of call. It could be something very physical on the manufacturing line as well. And that's where the conflicts and the issues occur, basically. I like to describe it — that's where the fistfights happen between R&D and manufacturing.

Trevor: So in the article you describe a paradox — a lot of these assays that are performing beautifully at the bench, with great performance characteristics, can be quite challenging to actually manufacture, even with a well-designed manufacturing line. And that can introduce anomalies that can be hard to troubleshoot, and then you get this back-and-forth — he said, she said — between the manufacturing and assay development groups. How does something that works that well on the bench end up being the problem?

Pat: Great question. I think you look at that from both directions, both from R&D and from manufacturing. So great R&D performance does not equal a manufacturing product, okay. And then look at it the other way — high-throughput manufacturing does not equal a robust assay performance, right. They're two very linked events, but they're two separate questions as well. So think about it from the point of view of the bench scientist, the R&D scientist — that's a very different, maybe even a controlled environment. It's forgiving because you have an expert scientist working on this assay. Usually it's small pieces of material lying flat on your development equipment. There's no real mechanical stress on it. They can take their time — speed is slow, there's no pressure of production there. And the assay works under these conditions. But the assay works — it's not necessarily the product at that stage. So what the R&D person is looking at is really a prototype or a concept. It's not necessarily a product at that stage. Manufacturing is when somebody comes along and says, okay, I need it to be robust, reproducible. They're looking at it on a much larger scale — it could take all day to do something because there's so much of it to be done. And then obviously, manufacturing — think about it — they need to make this over many months or years. So they're going to get so many different lots of nitrocellulose, lots of pads, lots of antibodies and stuff that may all be just slightly different, all within the vendor specification, but all subtle differences that will cause issues, potentially. Manufacturing is now introducing variation, okay, and this can go on for many years. That's by nature — you should expect that. But unfortunately, that's not what happens. So suddenly you take something that's working really well, put it onto the manufacturing line — it's now going to run faster because you want to reduce the labor associated with it, so you need to get it through material, maybe going through drying towers and not ovens, basically. So the original assay is doing what it was asked to do, but it wasn't designed for what's almost like a different set of physics, right? So that's where the issues arise and where the two sides don't meet head-on and have a nice handshake.

Trevor: So you're careful to say this is rarely any one team's fault — it really comes from treating development and manufacturing as sequential steps rather than one integrated activity. Why do you think manufacturing or development teams still make this mistake?

Pat: I suppose most organizations are set up that way, where you have an R&D team and a separate manufacturing department or manufacturing team. They could be in separate buildings — sometimes they could be in separate companies, if both aspects are outsourced. So that kind of structure really promotes sequential thinking rather than combined thinking. In the CDMO business, where your development partner and your manufacturing partner are different companies, you may have communication silos. There are different incentives for both parties as well — they're not necessarily playing off the same financial hymn sheet. So think about it — the milestones for each — a development milestone rewards lab performance: how sensitive your assay is, how specific it is, the limit of detection. Nobody in R&D is going to get a pat on the back or credit for, well, my assay tolerates variable line speed, variation in manufacturing, so it performs consistently across multiple shifts. R&D scientists don't think about that, but that's the reality for manufacturing. And then, when teams come to pick a contract manufacturer, the development has already been done — by the time manufacturing gets involved, the design is locked. So that obviously can be a real issue. So really you need to consider design for manufacturing, and it's really misunderstood. People think of design for manufacturing like tech transfer, right — like you check it off your checklist when you get to that portion. And really it's not — it's not tech transfer. It should actually be part of the design of the product. It's as important as selecting the right antibody, okay — as to what your manufacturing parameters are. And they're important inputs for the R&D scientist, okay, not just for the manufacturing personnel.

Trevor: Do you see this happening more with early-stage companies — assays that have maybe been spun out of academic institutions, where there's not that depth of experience in bringing a product to market necessarily at the start?

Pat: That's a prime example of where you're going to see this sort of thing, because again, those people are being credited — oh, you've developed an assay. Well, you've developed a proof of concept, or maybe — let's call it an assay, right — but it's not a product. An assay and a product can be two completely different things. A product is where you take the assay and can manufacture it. An assay doesn't mean that you can manufacture it. And that's specifically — yes, I mean for a startup company, they're all focused on getting the assay to work. Manufacturing it well — that's somebody else's issue down the line, and we'll cross that bridge when we come to it. Well, unfortunately, when you get to that bridge, it could be too late — or maybe not too late, but it's going to be an expensive walk across that bridge.

Trevor: You say a lot that assay design decisions in the development phase really embed a manufacturing consequence, and I think that bears true in what you've just said. You have one example where R&D teams underestimate this — in the article you reference conjugate behavior, for example, during long dispensing runs, and a choice that in development might not factor in processing at manufacturing scale.

Pat: Yeah, absolutely. So let's take — leaving aside whether you can make the conjugate okay in a small volume or a large volume — but moving on from there, as to how you use it: conjugates are typically nanoparticles, so they're suspended solids, okay. You shake up the tube, you can see everything swirling around, and everything is great. But when the bench scientist is making a small little batch, okay, they have a little tube of it, they're dispensing it, they swirl it, maybe they're holding it in their hand or it's just on the desk in front of them — everything is kind of controlled, right? So now on a production run — a production run could be running all day, and you have multiple rolls of material that need to be sprayed with the conjugate. The conjugate is sitting on your dispenser, okay, it's a solid — it doesn't always have to, but it could actually start to settle. So now, if you imagine it's settling, the bottom of that tube is going to be more concentrated. So what you find is early on, the material you spray early in the day is getting one concentration of those particles, and the stuff later in the day is getting a lot more concentrated solution — way more particles. So when you go to transform that into strips for testing, it's going to be totally different. You may get big, whopping results and big strong positives later in the day versus earlier in the day. So ultimately you could start generating false positives, basically, or nonspecific results. So it's very, very difficult. That's just one example — the conjugate. But the other thing follows for all materials, whether it's membranes, antibodies, and so on — you get lot-to-lot variation. It's the same membrane maybe, but a different lot from the manufacturer. You need to make sure you're including that in your development. You can't leave your manufacturing colleagues to figure that one out later.

Trevor: This is a great example. Maybe you could start by giving a little bit of an explanation of what lateral flow manufacturing actually looks like, and then walk through the detail — the concept that the first strips you run may not perform nearly the same as the last strips on the line. Maybe set the stage of what happens in an actual lateral flow manufacturing scenario.

Pat: Okay, so I'm walking into the lab, I've got a full agenda for today. Manufacturing is going to be over several days — you're making all your buffer solutions, you're making your striping solutions, you're doing your conjugations, and you're getting everything ready for dispensing. And really, dispensing is the key, unique part of manufacturing. So in the lab, the scientist may have striped a few feet or a few meters of material — in manufacturing, you're setting up to potentially dispense on rolls of material, maybe a 100-meter roll, maybe the manufacturing run is two or three or nine rolls of material. So you're going to start spraying or striping, but remember — all through that time your reagents are sitting out live, potentially at room temperature, and then you're going to stripe one roll, undo everything, put on the next roll of material — it's a continuous process throughout the entire day, okay. So that's why it's going to be very different — the scientist doing a few quick experiments just looked at a small portion of material and maybe did the striping and spraying in an hour. The manufacturing technician is looking at this as an all-day event, or maybe even over multiple days. Okay, so if I go back to my example of the conjugate sitting there all day and settling out, maybe I need to consider — okay, to avoid it settling, I'm going to start mixing it or recirculating it in some way. Great idea — now I've solved the technical problem of the difference between early morning and late afternoon, because concentration is identical. Well, what you need to consider then is that's a late-in-the-game decision to make, right, because — what's that vortexing or recirculation going to do? Is that going to destroy the conjugate in some ways — a lot of shearing forces? Is it going to introduce a lot of air bubbles into it? You're going to start forming — you and I have probably spent many years in the lab, shaking a little tube and suddenly you get that little froth or foam on top. Well, we were all taught that's not good, because that includes denatured protein. So now suddenly you've lost most of your antibody, okay. So that seemed like a great idea to start mixing this stuff — now you're ruining your conjugate. So the moral of the story is, let the R&D scientist — with a manufacturing brain on their head — figure that one out during development. So it's a no-brainer once it gets to manufacturing, because everybody knew from the start that this solution will sit on that machine all day long and needs to be, you know, a hundred percent performance level, 9 a.m. in the morning, 4 or 5 p.m. in the afternoon, and it needs to act the same. So those issues should be found and solved in development, and not in manufacturing.

Trevor: You touched on materials, too, and the difference between the bench and manufacturing, which is reel-to-reel — where hundreds of meters of media are placed in a reel format, like a film reel, and strung through a machine to be deposited and striped. So what role do materials play in all of this, in terms of variability at the scale of manufacturing? You mentioned in the article that nitrocellulose can vary between lots and within a lot — beginning and end of the lot — and that membrane and pad interactions can only show up under continuous processing, not bench-type processing. What are ways that assays can end up optimized for conditions that don't really exist on the manufacturing line?

Pat: Yeah, for sure. And let me say — before I start talking about all the variability in these materials — the materials we use today, compared to what we used 20, 30 years ago, are incredible, okay. In the early days of lateral flow, we all used materials that really weren't designed for lateral flow — probably designed for something not even within the life sciences. So you kind of figured out, oh, this might work here. Today they're very refined, and the manufacturers do an incredible job trying to get as much consistency. But these materials are made in large volumes, okay — so like nitrocellulose, we kind of colloquially say it covers an entire football pitch, and then you start cutting it up into rolls after that. So how do you ensure that uniformity across the entire football pitch, when you're only getting a little portion of it? So from roll to roll and from batch to batch, you get subtle differences. They're all within the specification the vendor specifies — that's fine, that's not a problem. But those subtle differences — if your assay is very touchy or very sensitive to some small variances, you're going to find — what we're saying is the development scientist should be finding, understanding that, and building those tolerances into the product design. You can't wait until your manufacturing colleagues are figuring that out, because, six months into manufacturing, suddenly they get a new lot of material and it acts completely different. It's within the specification the vendor says, but now it's impacting your assay. So that's part of the transformation from an assay to a product — part of that is ensuring that every time you go to make it, you can make it, even though you have a different lot of antibody, a different lot of membrane, and so forth. And the processing of that could be different too. So think about it — when you're striping just a small piece of membrane and then laminating it onto your card, everything is flat on the bench in front of you. When you're going reel-to-reel for high-throughput automation, everything is on a roll — it's curved. Now everything has a curvature — is that going to impact your assay? And also, is it going to impact the tension on the materials as they're laminated? Is it going to stretch some of the material, or compress some of the material? All of those could be very influential in impacting the performance of your assay. So there's a lot to think about. But the moral of the story is: find and fix these things in development, not in manufacturing.

Trevor: That makes sense, and I think we're transitioning nicely into an area that's macroscopic — thinking about the physics of doing anything at scale. Lateral flow — this applies to almost any diagnostic. I've certainly had a lot of experience, some of it negative, where assays have just fallen apart at the manufacturing scale — it just wasn't designed upfront to handle it, with very dire consequences. When I talk to people who aren't that familiar with how diagnostics are developed, I use the analogy of baking cupcakes — you might be able to bake a dozen cupcakes and everybody loves them, right, but if you're asked to make a thousand cupcakes using the same recipe, just scaled up by multiplying the ingredients, you're absolutely not going to get the same taste or texture — it's just not going to work the same. And reel-to-reel is a different environment, not just a faster version of the bench. So for someone who's only worked on benchtop or batch platforms, what kind of differences come into play as you move to higher scale?

Pat: You've touched on it already — really on the reel-to-reel automation, because you need that for high throughput, to drive your unit cost down — you need to scale to that level and keep up with your market demand as well. So on the reel-to-reel, the first main difference is your material is moving. At the bench level, where the scientist did it, everything is fully supported — it's lying flat. On the reel-to-reel, now that material is being dragged from the outlet reel to the uptake reel, and it's moving all the time. So everything is variable there, and there's tension involved. So wet materials can now suddenly behave differently. You picked up a roll of dry material, looks fine — you run that from roll to roll and it seems okay. But now you could be dipping it or spraying it — it gets wet. What's that doing to, let's say, the tensile strength of it? It could snap, right. So the manufacturer is looking at that and saying, what's going on? That's where the scientist should have figured that out already. So that material was great when it was lying on the bench, but now, when it's on a roll and it's getting wet, it's lost its tensile strength. So that's a very important, very specific example, but it's an important thing to think about during development. And then, as it's going through the reel-to-reel, you have to dry it — now it's going to dry in a drying tower at a certain temperature, because it's moving, it's only going to be in that drying tower for a few minutes. Whereas the bench scientist, when they striped or sprayed it, just stuck it into the oven and walked away, and said, well, it dried. Now you're looking at the difference between two drying processes — an oven, where you can have almost unlimited time, and a drying tower, which has very limited time. Those two things are very, very different, and there could be pros and cons of either one, but you need to design that into the product, because — are you over-drying something, are you under-drying something? Something as simple as using the oven — how do you know it was at that temperature for the 30 minutes or the hour when you turned your back and walked away? Your colleagues — were they pulling their materials in and out of the door, or was the oven open, right? So how do you know that was a constant temperature for that amount of time? So all of those things need to be taken into account. And that's part of what development should have been doing — putting that assay on the reel-to-reel and figuring out all those nuances, and designing the system so that the reel-to-reel and the drying tower process is working, basically. And then there's a whole thing about what's touching the material — all of those things vary as well: the speed, the vibration — all of those things can cause things to move. One line can run into another. You need to take all of that into account.

Trevor: That's great. I think many of our listeners — they're really competent, capable developers, capable of developing accurate, specific assays at the lab scale. Many have actually come to our Basic Lateral Flow Training course — our next one runs in September, so I'm putting a plug in for anyone who may be interested in that. For a team that manufactures in-house and thinks it has all of this covered — where does the development-to-manufacturing gap still bite them, and how should they think about this design-for-manufacturing concept?

Pat: Again, following on from what I was just saying about moving from bench to reel-to-reel, there are a lot of different — maybe some of them not so subtle — differences. So if I'm spraying something at the bench level, it's lying on a flat platform, I'm spraying, basically the material is sitting in that solution. When I'm spraying something on the reel-to-reel, I could spray it so hard that it actually goes right through the material. So that could be a fact of life — I may need to turn down the PSI on those air jets, or whatever, but I may also need to figure out — oh, I need to spray more material onto the reel-to-reel material, so that there's enough left to make sure the assay performance is correct at the end of the production run. So there are a lot of nuances there that you have to — that's part of the design, okay. Just because you were able to spray it on an XYZ platform, for instance, doesn't mean you use the exact same parameters — and the same goes for dipping, or whatever else you're doing. Really, you need to be conscious of those differences and take them into account.

Trevor: When we think about manufacturing and designing for manufacture — from the bench to the manufacturing — you have this hard line between an assay that can work and an assay that can be made. You call it a hero batch, or golden batch — you see these conditions rarely survive automation in many cases. You may be proud of the performance until it gets to that automation. What's the type of question a team should be asking instead of "does it work" on the development side? What are the other questions they should be asking?

Pat: I guess that's the classical conundrum a lot of people encounter. So the hero batch — you know, we call it the golden batch, that's the word I used to use in my past as well, but I know it's also referred to as magic hands.

Trevor: Yes, exactly. Or green thumbs.

Pat: Yes, exactly, or green thumbs. So what that golden batch or hero batch proves is that, yes, the chemistry is possible, but for me that's a prototype, or maybe even just a concept, okay. It doesn't prove that the assay can be manufactured — therefore it's not a product yet, okay. So the golden batch is a prototype — it's always the early output. You should have an early prototype to say, okay, I think I've selected the right antibodies and so forth, I think I can hit roughly where it needs to be in sensitivity and specificity, I'm good to go on to the next step. You're not ready for manufacturing, okay — so you basically design and build it to be repeatable. You're not designing it just to impress somebody, okay, by saying, I have this sensitivity — well, the sensitivity of a prototype or a concept isn't going to make you any money on the marketplace. It'll never even get to it. So it's more — you should be asking yourself: can somebody on the manufacturing floor make this over multiple shifts, during the day into the night, maybe, with a whole load of different batches coming through — incoming — because the vendor has finished one lot and is going to send you another lot. Is it going to work then? And you have to remember how the assay was designed — if it was designed where all the specifications were ten out of ten, you had all the dials turned up high, you've left the manufacturing guys nothing to turn, right, because if they turn it down, they're just going to lose performance. You have to design it so it's kind of in the middle, where you can crank it up or down to incorporate or tolerate all those variances, basically. It's kind of like — you know, golfers say, you drive for show and you putt for dough, right? It's the same thing with a product — yes, sure, you need all that performance, like sensitivity and specificity and LOD, but you need to be able to make it at scale. That's what the product is — the prototype or concept is not the product.

Trevor: Interesting. I think you mentioned that if an assay is only working with someone with golden hands, or an expert — many lateral flow tests are designed to be used by the regular general population. If it's only working on the bench with someone who does this day in and day out, how do you deliver that message to people who are really confident in the bench data, but it hasn't been robustly tested, hasn't gone through manufacturing or guard-banding? How do you deliver this message without telling them that what they're doing is completely wrong?

Pat: Yeah, without hurting feelings or getting somebody bent out of shape, right. So the R&D scientist's bench data — it's real, right, it's good, it shows the chemistry is working. But again, you're at a certain stage in your development — you're not finished, you're at a prototype level, you have an assay, okay. So it says the assay works — that doesn't mean the process to produce that assay at high throughput is robust enough that you actually have a product. There are two different things. The bench scientist has proven something really important, but the second part of their job is: let's prove that it's manufacturable — which is an additional but imperative objective of development. That's not the objective of manufacturing — they should be following instructions, and the correct product should pop out the other end, right — that's the job of the developer. So you need to determine — yes, sure, the chemistry works, but does the process work as well, to make sure you can produce the product? Because think about it — this assay or prototype will now have to be made and work with huge differences in conditions. Will it still pass and come out the other end? So you really have to think about it and take it through the whole process and make sure it works, yes, at small scale and then at high scale. The goal here is not to invalidate what the scientist did — it's to extend that proof into proving manufacturability. Both are necessary to ship that product.

Trevor: Let's shift a little bit to some of the impacts of these decisions within development and manufacturing, and what happens when things go wrong, right. There's costs involved here — yield loss on the development line could be days, a few weeks, small amounts of the project, but that could be the entire margin of the product once built at scale, and you have to rework an entire batch, right, for the lateral flow line. So talk a bit about why first-pass success is so important — it's much more unforgiving.

Pat: So there are the kind of messages you get from your CFO, right — you think everything is rosy, you've got great numbers of sensitivity and specificity, and suddenly that email says, well, why is it costing us so much to make this? I'm sure that's daunting for a lot of people. So you really need to do your math early in the process — you need to model that production yield. Are you getting a hundred percent yield? You'll never have that, right. Are you getting 95, 90, 85 percent? All may be viable, but all may not be, as well. So 95 percent — okay, that's viable — but drop into 90 percent, it's not economically viable. So you really have to determine that loss — it could be totally insignificant during development, right. If I produce 50 strips and 45 work fine, that's not a big deal. But if I'm making 100,000 strips and 5,000 don't work, that's a big deal, right — so there's a very big difference. You need to look at that early. There are lots of different reasons for that as well — using the higher-throughput equipment, there's a lot of dead volume, maybe very expensive reagents — and this is not just for lateral flow, in the pumps you have to load up expensive conjugate or antibody. It's the same for making an ELISA — let's say you have an Oyster Bay or something like that — all those pumps and lines contain dead volume that will never see the product because they're just there. You need to load so much more and it'll all go in the trash. In R&D, you were able to figure all that out, and when you're doing everything manually, that's not an issue — but that could be a large economic or financial impact on your product. And things like — when you're running a roll of material from one roll to the next, if you have to set it up before you start spraying, you're losing the first part of the roll. So you need to consider using leader sequences and stuff as well, so you're actually spraying a hundred percent of the material and not just wasting the beginning and end of the roll of material. So all of those things impact the economics of it, and those need to be considered during development, not when it gets to manufacturing.

Trevor: That brings up a great point, because we speak with a lot of customers who have assays or are contemplating assay development opportunities. Not every assay developer manufactures — so if you're not actually concerned about the manufacturing phase, how are you going to validate or design for manufacture? Our mantra here at DCN Dx is design for manufacture from the very beginning, and because we do both development and manufacturing, we're able to manage and optimize those conditions. What does that look like in practice from a company like DCN Dx, compared to how maybe other teams might run a program where they're development-first and then seeking an outsourced manufacturer?

Pat: Here at DCN, obviously the assay development scientists and the manufacturing engineers are all in the same room, all day, every day, basically. They're together from day one for a program — it's not a hand-off, it's, I suppose you'd call it, a co-design, right. The manufacturing engineer is asking, well, what is the reel-to-reel going to do to this, while the assay scientist is still choosing the chemistry, or the materials — they're saying, oh, this pad works great, but the manufacturing technician is saying, yeah, that's great, but when that gets wet, it's not going to stay on the rollers, basically, it's going to break. So it's hand-in-hand development — manufacturing isn't an afterthought, it's an early input. And the other way we do it here at DCN as well is that all our development scientists are highly trained in manufacturing — they're developing the product thinking as a manufacturer, right, because they've all worked in manufacturing as well — we interchange people a lot, they're hands-on on all aspects. But because we want them to learn and understand all aspects of product development, and product development isn't just the chemistry of selecting the antibodies — it's making sure we can package it into, like, tens of thousands of devices into kits, and every time we go to manufacture something, it works, we can ship it. So they're working hand-in-hand from day one. So it's not — one of your inputs isn't "oh, I need an LOD or sensitivity of X" — it should be "what's the manufacturing performance, what do I need to do to release this to the market" — that's an input, and now you work back from there, and that's how I start to design my product. So it's not an afterthought — it's not a nice-to-have, saying, oh, well, the results I have are great for a scientific publication, or these are going to look great in our FDA submission — that's not going to make it a manufactured product. So those design inputs, manufacturing inputs, are so important as part of that. So I would say process development is an integral part of assay development — you can't distinguish them, they're interlocked, intertwined, and you should be looking at both — unless your business is just, I want to see if a prototype works, that's fine, but that's not product development. Product development really is your assay and your process working cohesively together. So design for manufacturing — it's kind of a misnomer, because people think of it, as I said earlier, as tech transfer. No — design for manufacturing happens from day one, and those manufacturing constraints are inputted into your design inputs from day one.

Trevor: What happens then — if someone has an assay that they've developed and want us to manufacture it, where do you see the biggest problems, or where have you seen some of the biggest issues with a fully baked assay that wants us to scale it up for them? How often does that just go off without a hitch?

Pat: It doesn't go off without a hitch — very often there are always issues. I mean, sure, we get some people who even have a concept and they're saying, can you manufacture this, and we say, well, you missed a step, right. But let's say they have designed it — that happens a lot. Again, as we've been talking here, just because the assay strips or the prototypes are hitting sensitivity or limit of detection, it doesn't mean it's manufacturable. So for all the reasons we've been laying out today, that's what comes up. That conversation can be very awkward, because the customer comes in saying, we're ready for manufacturing because our development scientists have said it's ready — it looks great, it's got great CVs, it's got great sensitivity. But could that scientist go back and make a nice small batch of it and have it work exactly the same again? Of course — now they need to consider if they have to use a different lot of membrane or antibody in the meantime, but let's say they use the same one, that'll pop out. But that's not manufacturing. So it's a really important question, and we're the ones ending up having to answer it. And unfortunately, as you know, Trevor, it's a tough pill to swallow, right, because it usually means — oh, suddenly you had a financial projection of where your product was going to be hitting the market, and suddenly now you have to say, well, now I have to finance almost a redevelopment. And really it is redevelopment — it's not doing something for the sake of doing it, it's because it hasn't been designed for manufacturing. So it can be a costly mistake sometimes, assuming that the next step is just manufacturing.

Trevor: Yeah. So you close the article with a five-step framework on how this approach should be taken. Which one have you seen repeatedly move the needle the most, for folks as they consider their development programs?

Pat: Involve manufacturing from day one — it's part of assay design, or make sure your development is being run by scientists who are fluent in manufacturing. Okay — define the manufacturing performance, that's an important input before going off and trying to optimize an assay, right. Think about the scaled process — you scale it during optimization and development, not afterwards. You stress it, you do your challenge studies, you stress it to almost try and break it, so the scientists can have it fixed by the time it gets to manufacturing. And then you're optimizing for the total system, right — it's not just the chemistry, it's not just running on a reel-to-reel, it's the entire system, and you're not looking at any one metric like sensitivity — it means, is there a viable product that meets all the specifications popping out the other end? So to answer your question, overarching, it's design for manufacturing — but maybe what's a critical step is getting into that scaled-up process and equipment during development, because that is part of assay development. Early automation means you'll have the truth about your assay, right — does it work or not, when I go to manufacture it? And that's the least expensive time to deal with those issues, right, whether it's part of development. You're not finding out the failure mode when you have a production line stalled and there's no product coming out of it.

Trevor: We've locked things down.

Pat: That's — yes, otherwise— TREVOR Exactly, yeah. PAT Exactly, yeah.

Trevor: I really appreciate getting the insights from the ground on what it takes to consider the scale-up process from development. If there's one thing a listener would take away from today's discussion, what would you want it to be?

Pat: Maybe I'm sounding like a broken record, but it's design for manufacturing — either get your manufacturing people involved from day one, or use development people who are fluent in manufacturing, because remember, they're both coming at it from different directions, and when they're not working together — if they're siloed, their performance is assessed very differently, right. The scientists are being asked, well, how sensitive can you make the assay; the manufacturer is being asked, oh, can you produce this at the lowest possible cost, most reproducible, and high performance — two very different things. And sometimes that actually comes from day one, from when you're going out raising funds from your investor. The investor hears, oh, it's sensitive, it can pick something up — sometimes they're not interested in, well, it'll work with all the variations of manufacturing — that's not going to get any interest at a board meeting, right, it's more the economics of it: I'm giving you money to develop a product, I need to hear that it can detect whatever antigen you're looking at, rather than, I can make 50,000 of these, because in their mind, that's a way down the line as well. But in reality, it's not, right. So I would say, for everybody — the startup, the investor, the manufacturing people, the R&D people — it's a combined approach from day one, basically.

Trevor: I think the two things investors are going to care about, having some experience here, are how much is it going to cost, and when can you start selling it. And if you're not designing for manufacturing, it sounds like it could cost a lot, and the timeline could be a lot longer than if you started right from the get-go.

Pat: The later the time, the more expensive it gets, basically, right.

Trevor: Well, this has been great, Pat. I want to thank you so much for walking us through what is maybe not so well understood in the development process, I think. Assay development is approachable — we have a lot of expert and great scientists — but there's a particular set of processes that need to be undertaken to develop something for manufacturing, which is the goal of most of our customers. They come to us with a goal in mind, and that's the commercial success of the assay they're building, and that commercial success is usually tens of thousands of units built at a time to get to market. For listeners who want the full argument on which this podcast was based, feel free to go back and see Pat's article, "Bridging Assay Development and Manufacturing for Lateral Flow Success," which is on DCNDx.com. Pat, I want to thank you again. And until next time.

Pat: Thank you, Trevor, it's been a pleasure. And, you know, we can make it work, right — it's never too late, it may be a little expensive when it's late, but if anybody has kind of gone down the road and said, oh no, we didn't follow what he's just saying there — well, give us a call, and we can figure it out for you.

Trevor: Sounds great. All right. Thanks. Until next time.

Pat: Alrighty. Thank you, Trevor. Bye-bye.