PODCAST: Specimen Strategy Is Development Strategy: Why “We’ll Source It Later” Fails

Mitzi Rettinger: What's the fastest way to add months to an IVD program without changing a single assay parameter? Specimens. Not the fact that you need them, but how you define them, collect them, and package the data around them. Today we're talking about specimen strategy. Why? It's a part of development strategy and what the experienced teams decide early so they're not rebuilding their evidence plan halfway through. Welcome back to Expert Insights from DCN Diagnostics. I'm Mitzi Rettinger, Chief Revenue Officer at DCN. And today I'm joined by my colleague Jim Boushell, Senior Vice President of Biospecimens. Jim has spent decades on both sides building biorepositories and supporting diagnostic developers, and now he's leading DCN's prospective biospecimen collections offering. Jim, thanks for joining me.

Jim Boushell: Thanks for having me.

Mitzi: For listeners who haven't met you yet, what's the through line in your career and how did you end up focusing on biospecimens for diagnostics?

Jim: Well, it's funny, I never set out to become the biospecimen guy, but my entire career I've spent sitting between patients and products and all of a sudden I've become that guy. I started out on the patient side, working for Abbott Laboratories. I was selling some of their near patient testing, and so I got a good idea of what the patient experience is just from sitting so close to that. Then I moved over to the development side and started working with a startup where I was responsible for setting up collections and logistics to move specimens from the patient's clinic over into the research labs that are used in developing products. You know, for me, the question that always arises as we do next generation technologies — you know, this is the result of being in this industry for 30 years — it always comes back to: can you find the right specimens at the right time, with the right level of data that the regulators are really going to care about and that you can use in your programs? As far as the through line for me, I'd say it's just that — it's solving that same problem over and over again. You know, making sure that development teams aren't just thinking about specimens after the fact, that they're actually building a strategy in order to conduct their studies, specimen planning and building for programs. It's not a last minute procurement task. It's something that has to be planned for. So, you know, here I am now at DCN. I get to do this all over again. I love it. And not to sound corny, but it's in my blood. This isn't — you know, it's not really — I don't look at it as what I'm doing. I'm looking at it as who I am.

Mitzi: That's really nice. You know, you've worked in biorepositories and specimen operations for a long time. I mean, you're, you know, decades, as you said, over 30 years. What would you say has changed in the last 5 to 10 years in terms of what IVD teams need from biospecimens?

Jim: Well, obviously a lot's changed in the last 30 years. It's just light years away from where it was way back when. You know, number one, obviously the science has become more sophisticated. It's more sensitive. We've got technologies like molecular diagnostics, next gen sequencing, multiplex platforms. These technologies are incredibly sensitive. And with that sensitivity is a double-edged sword. Those same assays that are super sensitive — well, they can detect things that years ago you wouldn't think they could dream of. And they're also picking up a lot of the variables and the garbage that you didn't control for. So you have to be careful with that. Things like a different tube type, a slight delay in processing. You did an extra freeze cycle with the samples — back in the day, let's say 15, 20 years ago with ELISAs and LFA assays, that didn't really matter much. But in today's technologies, those little variables can wreck the data and essentially take your whole study and nullify it. Probably the next thing I would say that's changed in that time period is the regulatory rigor. You know, reviewers today are asking really tough questions about the intended use population for your assay, about how your specimens are being collected, about whether or not your study population actually represents the clinical reality of where you're going to be using your test. You know, the days are gone where you just grab samples out of a freezer, cobble together some data, and call it a validation. Those days are gone. Probably the final area that has changed now versus in the past is the competitive landscape. There's a lot of companies out there chasing a product and bringing it to market. And what that has done is take timelines and really compress them. So the tolerance for getting specimens wrong and actually having to recollect, and then trying to explain to the board or other project teams why you need more money and more time to go ahead and do your collection over — there's no tolerance for that today. In the past, they might have been more understanding. So considering the advancements we talked about in technology, the increased regulatory focus, and now finally the competition, specimen strategy that wasn't there 20 years ago has gone from being an afterthought to being one of the most important, consequential decisions in someone's development program. The teams that figure this out early on, those are the teams that move forward fast. The teams that don't, they pay for it later. And that's every time.

Mitzi: That's really insightful, Jim. You know, I like how you kind of grouped that as far as technology, regulation, and competitor landscape. One thing when you were talking about sensitivity and the things that have changed — you said, okay, now it's more sensitive, there's things that you see that you didn't see before, which can wreck your whole study. Does that mean that that whole idea of prospective collections, inclusion and exclusion and data are way more important today and can be way more difficult than in the past?

Jim: Yeah. Oh, absolutely. Absolutely. And really Mitzi, it's that they're more intentional. And you've got a plan for it.

Mitzi: Well, quick definitions. So we're all on the same page. When you say prospective collection, what do you mean? And what do you not mean?

Jim: In simple terms, a prospective collection means we're going out and collecting specimens specifically for your program. We're controlling the who, the what and the how — from the moment we enroll the patient into the collection to the time it actually is used in testing — and we're making sure that it meets whatever your intended use is. So as I like to phrase it, I call it: every prospective collection is purpose built. What it's not is pulling specimens from a freezer in a biobank. That's what we call or refer to as remnant specimens. And frankly, there's a time and place for banked specimens if you're doing feasibility work, early method development, comparative studies, things like that. I'm not anti-biobank by any stretch, but what you have to realize is that biobank specimens come with baggage, and it's baggage you can't always see. You're essentially inheriting someone else's collection protocol, someone else's handling protocol. The data that's coming with the samples was based on someone else's data, not necessarily what you need to set up your study. At the end of the day, prospective collections give you the control you need. You define the inclusion and exclusion criteria. You define the matrix. You explain in detail what the handling conditions need to be, and most importantly, you get the data that you need upfront. Once you get all that together in a nice package, it's typically audit ready from the very first day. It's not something that you're cobbling together after the fact. And so prospective collections definitely have a place in the development of diagnostics.

Mitzi: That's great. I appreciate that definition. When you were talking earlier, I guess, concluding what's changed and you kind of landed on strategy — that that's way more important now than it had been in the past. When you say specimen strategy, what are you including in that phrase, or are you explicitly not including something?

Jim: That's a great question. Specimen strategy essentially involves every decision that determines whether or not the specimens are going to be able to support the claims that you need to make. So that starts at understanding what the intended use is of your assay, the regulatory pathway you're going to take, and then working backwards from there. What specimen strategy isn't is procurement. Procurement — that's "I need 500 samples from diabetic patients." That's a shopping list. Specimen strategy is: given my intended use, my claims, my regulatory pathway, here's the plan that, in my study, will generate the evidence that I need. And those are two very fundamentally different conversations. The teams that treat them as the same conversation — those are the ones that end up in trouble.

Mitzi: So why would you say that specimen strategy belongs in the development strategy conversation and not, you know, last minute thinking? And I'm sure it's tied back to what you just said.

Jim: Yeah. I mean, simple. If you treat it as a sourcing task, then you've already made decisions — maybe without ever realizing that you made them — and now your specimens have to support what those decisions were. So your claims are set, your intended use is defined, your pathways have already been chosen for you, and now you've got to go back and say, how am I going to put this puzzle together with specimens so that it fits all of that? And you know what I say to those people? Good luck, because it's really a tough task to try to conquer.

Mitzi: You've said teams can lose months without noticing. What are the early signs that a program is headed into that trap?

Jim: The one phrase that makes me pull my hair out is when someone says we're going to have the specimen conversation later. So, you know, if you're in a development meeting and someone says, "we'll figure out the specimens once we lock the assay" — huge red flag. Throw the flag. If you're locking your assay without understanding what specimens you need for validation, it assimilates you. You're building a house and then deciding where to put the foundation after you put the walls up. The second signal is when teams talk purely about specimens in terms of volume. "We need 300 positives and 500 negatives." Okay. But from what population? How are they supposed to be collected? What types of information do you need to have for them? If people aren't asking those questions, you're heading down a difficult path. And then the final sign — and it's probably the sneaky one — is that people are using banked specimens for everything: for feasibility, for development, even for validation. And no one's ever asking how the specimen is going to hold up for the pivotal study. So you get comfortable with what's easy to access. Then one day you realize the banked specimens they're using aren't really what's going to be able to support your regulatory claims, and that you have to go ahead and do a prospective collection. But what you just did is you wasted six months of time, and it's unnecessary. So the sooner you address these things, the less painful this process is.

Mitzi: You know, it's fascinating, Jim, as you were explaining that — I can recall two situations where we're supporting some redevelopment on our development side because they were in clinical and although their analytical validation was spot on, their clinical was not coming out the same. And I bet this is probably very similar to what you're describing. They were probably using different, you know, samples or something that's not authentic to a freshly collected specimen.

Jim: Yeah. People don't intend to set out doing the wrong thing, especially anybody involved in science. It just happens. And sometimes it's a matter of awareness.

Mitzi: So this might bring up some of the same answers. But if you had to pick one decision teams delay too long, what is it?

Jim: Hmm. I'd say defining the intended use with enough detail that we can go ahead and build a collection protocol for them. Typically, development teams spend months inside the lab refining the assay, making sure the chemistry works great, making sure the platform is operating as expected, and the entire time no one's really asking, you know, who's this test really for? It's never been defined. It just stays fuzzy — not at the highest level, because everyone knows we're building a cancer test or we're building a panel for respiratory assays. I mean, at the level of detail that a reviewer is going to care about. The reviewer is going to ask: what's the clinical setting, what's the patient demographic, what are those inclusion and exclusion criteria that'll make your data more credible? Until you nail those things down, you can't really design a protocol. And until you have your protocol, every specimen decision you're making without it is just a guess. And guesses cost money and real time downstream.

Mitzi: So you're talking about those costs and, you know, I'm mentioning some of the failures. We're kind of saying, okay, we know that the plans fail, nobody wants that. So let's talk about ways specimen plans fail in real life. What are the top three failure modes you see most often?

Jim: There are three big ones. You know, number one is population mismatch. So you're collecting specimens from, let's say, a blood donor center. But the test you're developing is actually going to be used in a community clinic. So you know, things like disease prevalence, patient demographics, who the operators are — none of that's going to match what the real world clinical setting is. And so when you go and submit this to a reviewer, they're going to see right through that. And then you'll have to explain why the data doesn't match your intended use. Again, avoidable. I'd say the number two issue is pre-analytical chaos — and I will mess that word up a hundred times over here. But in simple terms, it's handling the specimens. Are you using the right collection tubes? Are the processing times being documented? Are the storage conditions what they're supposed to be, or do you really not know? One thing you'll find in today's assays, especially molecular and these really super sensitive ones, is that knowing those variables is important because they create what's called noise, and they directly affect the results that you get from your testing. I've seen programs where assays have performed wonderfully in-house in the labs, and then when they went out to the market — and you mentioned it, Mitzi, the gap between the analytical and clinical validation — they fall apart. And the root cause is specimen handling. It's not the science. Three — the final one — let's just lay it pretty simple. It's missing data. You've got to make sure you get all that data. If your plan is to go backwards and try to cobble together data to fit what your study needs to be, you're setting yourself up for a very, very frustrating amount of time spent. At the end, you don't end up actually getting what you need. And so in the end, all those variables that we talk about — they're preventable. Every single one of them. But only if you plan for these things before you collect, not after you collect.

Mitzi: That makes a lot of sense. Are there any particular modalities or indications where you see this the most often, or would you say these are all things that are just universal?

Jim: Yeah, I'd say they're universal. Everybody falls into the traps. Every process can fall into the traps. But you know, some are a little bit more obvious than others. So like in today's science — and I know we've talked about this a number of times between ourselves — it's liquid biopsies. That's a huge one. One of the most important parts of a liquid biopsy collection is the handling of the specimen. The difference between a cfDNA sample that was processed within two hours and one that sat for six hours is significant. It really is. It's the difference between having valid data and having garbage. And the companies, the people that asked you to do the collection, they know this — but they don't necessarily know that they need to tell you this in advance. They assume you know it. So it's important that you do. And then if we look at a less sensitive technology — some of the POC and the CLIA waiver programs — their version of this problem is different. It's all about who's the user. You know, you need to collect specimens from untrained or minimally trained operators in the actual settings where the test is going to be used. You can't just grab lab-collected specimens and pump them into a POC or CLIA waiver tester. The reviewer is going to know it and they're going to recognize it immediately. So you need to know real world what your usability data is going to be, and then require a very specific kind of prospective collection to match that.

Mitzi: What are the minimum inputs you need from a developer to design a collection that's mapped to these claims and an evidence plan?

Jim: So again, as we mentioned earlier, I'd start with intended use — and intended use being not the marketing version but actually the regulatory version. So: who's the test for, what's it going to detect, in what clinical setting, what specimen type are we dealing with. All of that information forms the foundation that everything else is built upon. The next thing that's important is the claims. What are you going to put on the label? What are the performance specs that you need to prove? Is it sensitivity? Is it specificity? Is it limit of detection? Is it reproducibility? Is it all of them? Or maybe it's just one of them. You've got to figure that out early on because each one of those claims has a very different specimen implication. The next one I would concern myself with is matrix and the handling requirements. We talked about it with liquid biopsy, but in other assays and technologies — so are we looking for serum? Are we looking for plasma, whole blood? Is it saliva? Is it passively collected, or are you actually prospectively collecting it? What types of tube types? Anticoagulants? What types of plasma? What are the processing and storage instructions? All of those things are part of that pre-analytical control — the term we've mentioned many times already — and you have to account for that from the very beginning. Fourth would be your target population. So not just the demographic profile you're looking for, but the clinical profile — age ranges, diseases, comorbidities. Something very important these days is geographic diversity. All those things need to be built into your protocol and must match your intended use. But you also need to think about prevalence expectations, because that's driving the math for positivity enrichment for your program and your study. Probably the last one I'd say is the comparator method. So what are you running your test against? Are you collecting specimens that have already been tested by a comparator? What do you need to arrange for a central laboratory or a reference laboratory that's running that comparator test so that you're able to get the testing done as part of that collection event? If you give me those five things, we can design a collection that maps directly to your evidence plan. Without them, we're guessing. And again, guessing is time and money.

Mitzi: What's your perspective on perfect specimens versus representative specimens? And when does that trade-off matter most?

Jim: It's an important distinction, so I'm glad you brought it up. Perfect specimens are just that — they're collected under pristine conditions in a controlled environment, handled exactly to specification, from a well-defined patient population. So what we all dream about. They're great for development work. They help you understand what your assay can and can't do under ideal conditions. Representative specimens — those are the ones that are out in the real world. So what your test is actually going to encounter in the clinical setting. And that means variability — variability in patient populations, going after the right target in the collection settings. Again, we mentioned a blood donor center versus a clinic. Operator skill — is it a trained user or an untrained user? And then finally, handling conditions. That's the clinical reality that your test is up against. You asked about the trade-off. Well, that matters most when you move from development into validation. If all your validation specimens are perfect, the reviewer is going to challenge you. They're going to ask: does the performance hold up when the test is used by the intended operator, in the intended setting, with the intended population? And if you can't answer those questions with confidence, you've got a gap and they're going to know it. The approach we always try to recommend is: use controlled specimens for your analytical validation when you need precision, but use the representative real world specimens when it's time for clinical performance.

Mitzi: When should an IVD team use banked specimens, and when should you consider a prospective collection?

Jim: Simple framework. Banked specimens — use them during the early stages. So feasibility, method development, proof of concept. Anytime you're trying to understand whether your assay works, bank specimens are just faster to access and they're usually cheaper upfront. And for the early questions that you have, those trade-offs are acceptable. Now, switching over to prospective collection — you do that when the stakes increase. So when you're doing a pivotal validation study, when you need to support a regulatory claim or some sort of reimbursement argument, or when your target population is so specific that you can't find a good match in the biobanks — that's when you're doing a prospective collection. If your assay has a very particular collection and handling requirement that a specimen coming out of a biobank can't guarantee, that's when you do a prospective collection — especially if you're developing POC or CLIA waiver tests, because that real world operator-in-setting data that we talked about, you don't typically get that out of a banked specimen, or you can't confirm whether or not it was actually part of the original collection. So at the end of the day, we try to recommend a hybrid approach for developers. If you're doing feasibility and utilizing remnants, that's fine — you do that for the first three to six months. When you go to switch over to your validation and your pivotal clinical study, that's when you've got to bring in prospective collections. And by doing that, it allows you to stagger the costs and manage the risk.

Mitzi: So prospective collections can be done poorly, too, I suppose. So what distinguishes a collection that's actually designed for IVD evidence?

Jim: Mm. Interesting question. I would say all prospective collections are not created equal. So just because you're collecting a specimen prospectively, it doesn't mean you're actually collecting the evidence that you need to be collecting. So a collection designed for IVD evidence starts at the protocol, not at the logistics of how you're getting to and from the specimens. Your protocol is mapped to your claims, which is mapped to your intended use and your regulatory pathway. So your protocol specifies exactly what specimens you need, who you need to be getting them from, how you need to collect them, and what level of collection data you need. All of that is then tested against whatever that comparator method is that we've talked about. In a nutshell, every decision in the protocol exists because it supports a specific piece of your evidence plan. The second thing that's really important, Mitzi, is site qualification. Are the sites representative of your intended use study? Are the operators at the site appropriately trained — or appropriately untrained, if you're doing a CLIA waiver study? Do they have the infrastructure to handle your specimen requirements? Specimen handling is not a light task. It's a heavy lift, and you have to make sure they have the infrastructure to support it. We've seen many a study fail because the site couldn't maintain processing timelines for a specific protocol. And that's a site qualification problem. That's not a problem with the science. The final thing I would think about is data capture. You need to build it in from the start. Don't try to figure it out later. Every specimen needs full chain of custody — handling documentation, clinical annotations, comparative results — everything captured at the point of collection. If your plan is to go chase it after the fact, you're already compromising your evidence package.

Mitzi: You know, it's crazy, Jim — as we've been going through all of this, it's like I keep having these little aha moments, like a conversation I've had with a company in development where the one you just mentioned, where you said it was the processing time, that was a challenge that somebody was having. This was for a clinical study in our clinical services area where they were saying that was a big challenge for them — that some of the sites were not following that part of the protocol, and it was really making a huge difference. So you're definitely enlightening me on and putting some of those pieces together for me. So I appreciate that, Jim. JIM Sure.

Mitzi: What types of programs does DCN Diagnostics support right now, and what specimen types and special handling scenarios are you routinely supporting?

Jim: Sure. So really we have a bit of a broad mix going on right now. We have classic infectious disease. We have a very strong oncology focus. We have respiratory panels that we're building, LGI panels, doing a lot of work with cardiology, neurology and Alzheimer's. We know that those are big areas of focus and concern for the world, and diagnostics is really driving some of that innovation. Women's health. Sexually transmitted diseases. And as I mentioned earlier, we're doing an increasing number of liquid biopsy programs. So those are the indications. Now on the actual sample side — or the specimen side, as we call it — that's requiring collecting whole blood, plasma, cells, any component of blood, swabs, urine, stool. Sometimes it's paired matrices, sometimes it's a standalone collection. Most of the collections that we do have very special handling requirements. So those might be utilizing Streck tubes, processing samples with tighter handling windows, very specific temperature-sensitive cold chain logistics — and really everything to match what the assay's tolerance is, so that when it goes into a developer's laboratory or into a clinical study, it meets their intended use.

Mitzi: When does integrating specimen collection with clinical operations and development work reduce that risk, versus when a standalone collection is just fine?

Jim: So I'd say whenever a decision made in one area directly affects the outcomes in another. And in diagnostics, that's pretty much most of the time. Think about it. You know, we have development teams that are making decisions about the sample matrix, how stable the analyte is and how it needs to be processed. And then in another room, we have the clinical operations team designing the study that we need to run for this program. And then we have the regulatory team who's defining what our claims are going to be and our submission strategy. And suddenly your specimen collection has to align to all of them, right? And when those functions are siloed across different vendors, handoffs are where things start to break. So, you know, at DCN, we all operate under the same infrastructure as the CRO and CDMO functions. We fall under the same quality system. We use the same or similar site networks. We have the same regulatory frameworks. So when the development team identifies a new pre-analytical requirement, it doesn't get lost in an email chain or between three different vendors. It gets built right into the collection protocol by the people who are sitting in the same organization. And you know, that's different. A standalone collection is fine when your program is pretty straightforward, you know exactly what you need, the requirements are locked, and you just need someone to execute. That's fine. But the more complex your program becomes, the more variables you're bringing into play, the more value there is in actually having integration. And honestly, most programs are a lot more complex than people think they are at the onset.

Mitzi: From your perspective, what handoffs tend to break when collections, assay development, and clinical execution are split across too many groups?

Jim: I would probably say the first thing that breaks is assumptions. You know, development thinks one thing was collected. Operations did something slightly different. And by the time clinical sees the data, nobody's really sure what's true. But, you know, the common thread beneath all of these is nobody intentionally dropped the ball. It just happens. Those handoffs tend to break because each of these groups are working on their own statement of work. They have their own timeline, and really they have their own definitions of what "complete" means. So if you bring this all under one roof, under one quality system, under one project management framework, one set of assumptions — these types of handoffs don't disappear, but they become conversations instead of surprises.

Mitzi: For the developers listening, what should they expect to see in a high-integrity data package?

Jim: In short, they should see the entire story of every specimen from the time it was enrolled through its final disposition, whatever that may be. So from the developer's perspective, they should have full chain of custody — documented, not just "here's the specimen collection date." They need to know who collected it. They need to know what collection device was used. What time was it collected? How was it processed? What temperature was it stored at? When was it shipped? How was it received? Every touchpoint should be documented, traceable, and timestamped. You should also have complete clinical data for your protocol — so demographics, diagnosis, relevant medical history, medications, whatever your study requires — and where possible, make sure you capture it at the point of collection. Don't make a plan to reconstruct it later from medical records. We already talked about that earlier. It's an exercise in futility sometimes. Next, they should really be focused on making sure they have the comparative method results and that they're actually linked to the specimen. Not a separate spreadsheet, not a different database. They need it linked, reconciled, and QC'd. Once they have all that collection and handling data — all that pre-analytical stuff that we talked about earlier: the tube types, the fill volumes, the processing times, the centrifugation speeds, storage conditions — for today's assays, these aren't nice to have. It's all essential for interpreting your results and defending them in a submission. The way that I look at it, all of this should be in an electronic format. It's not required, but it really should be, so that you can just hand it off to biostatistics or your regulatory team, and they have a nice clean data export to work from — not PDFs or handwritten forms. If your specimen provider can't deliver that, you're going to spend a lot of time and money cleaning up data before you can use it, and it's unnecessary.

Mitzi: So when you've said privacy controls and quality systems appropriate to the program, what does that mean operationally for specimen collection and data handling?

Jim: So operationally, it means your specimen collection and data handling operations operate under a defined quality system with documented procedures. At DCN, everything runs under our ISO 13485 quality system. It's the same system that governs our CDMO and our CRO operations. As far as privacy is concerned — well, that means everything should be collected under IRB oversight with proper informed consent. The data is coded, meaning we separate the direct identifiers from the patient specimen and clinical data, and then we maintain that link between the two securely with access controls. And then as far as the quality system goes — well, that means SOPs for every step of the process: site qualification procedures, training documentation, deviation and CAPA capture processes, and deviation management. When something goes wrong — because something goes wrong, and things go wrong — it matters. It matters how you catch it, how you document it, and how you correct it.

Mitzi: What's the most common false sense of security you see around specimens?

Jim: The belief that if you have specimens, then you have evidence. You know, a team will say, "I've got 500 positive specimens in the freezer. We're good to go." And on the surface, that sounds fine. But when you dig in, you can't determine where the specimens came from. You don't really know how they've been characterized or what the collection and handling conditions were. And if you have somebody giving you vague answers — like "the biobank said they were characterized," or "we think the handling was fine," or "we have most of the demographic data" — those odd phrases don't hold up to regulators. They just don't. The fix is simple, but it requires discipline. Don't confuse having specimens with having the right specimens. They're not the same thing.

Mitzi: Jim, I've really enjoyed this conversation, and I feel like it's been insightful and I've certainly learned a few things. So thank you so much for being here.

Jim: Thanks, Mitzi. This is the stuff I love to talk about, so I really appreciate the opportunity to be here today.

Mitzi: Thank you. If you want a quick check on whether your specimen plan and data requirements line up with your intended use and claims, start with the Biospecimens section on DCNDx.com. You'll see what DCN's prospective collections cover and what the deliverables look like. Feel free to reach out with any questions. Thanks for listening to another episode of Expert Insights. Don't forget to subscribe and we'll see you next time.