In the company’s 10th year of business, 908 Devices’ Chief Executive Officer Kevin Knopp, Ph.D., reflects on the company’s history of innovation and discusses how they are leveraging novel microfluidics techniques and miniaturized mass spectrometry to serve a growing user base that spans from biopharma to homeland security, in conversation with Pharma’s Almanac Editor in Chief David Alvaro, Ph.D.
David Alvaro (DA): Can you give me a concise overview of 908 Devices and the history of the company from its inception to today?
Kevin Knopp (KK): As of February 2022, it’s been 10 years since we started the company with a mission to change how and where chemical and biochemical analysis are being done by undertaking a platform change to a gold standard technique: mass spectrometry. We see it as a disruptive platform based on a technology we call “higher-pressure mass spec.” We’re taking the gold standard, laboratory-grade mass spectrometry technology out of the centralized laboratory environment where it normally lives and bringing it to what we call the “point of need.”
In terms of the evolving technology, it’s analogous to what we see in the computer industry, where there has been a progression from large, centralized supercomputers to desktops, laptops, and even mobile devices. We see parallels for chemical and biochemical detection, and we view our products as analogous to simple and convenient handheld and desktop devices, in contrast to large, supercomputer-like devices. As a result, our products are much more accessible and serve a much broader range of customers. We joke about this, but it’s true: our customers span “the FBI to the FDA” — from forensics applications to biopharmaceutical applications.
DA: Did you start with a particular use case in mind and expand from there, or were you considering all the different possibilities for the technology from the beginning?
KK: We always thought about the many possibilities. We could see mass spectrometry and chromatography-based technologies that large instrumentation companies like Agilent and Waters were making to serve a range of customers, both in the life sciences and in applied markets, and we could draw parallels. We saw that we could launch solutions to serve clients directly who would normally send their samples to these traditional instruments in centralized labs.
We started from a technology perspective, working on high-pressure mass spectrometry itself. We launched a handheld, lunchbox-size mass spectrometer, MX908, which was an entirely new category of mass spec devices at that time. Then, we started working on the applications — it fits naturally into areas like counterfeit drug detection and forensic toxic hazard detection.
Still, most of the traditional mass spectrometers on the market have some sort of chromatography on the front end, which is vital for analyzing complex samples. So, as we desired to broaden our application set and enable innovation in the life sciences, particularly biopharma and bioprocessing, we started developing a chromatography alternative that is based in genomics — a microfluidic device similar to what is used in genetics labs, where it’s a capillary electrophoresis (CE) device. What we produced is a CE microfluidic device that separates on the basis of size and charge to perform the initial separation that would then be fed into the mass spec.
In short, our development process began with the backend mass spec, which allowed us to achieve measurements with key forensic applications, and then we developed the technology further with microfluidics, which has allowed us to process liquid phase samples, opening the door for applications in biopharma and bioprocessing.
DA: Can you expand more on the fundamental foundational technology: what is unique about high-pressure mass spec, and why it is now possible to create these devices that are smaller than what anyone had ever built before?
KK: We partnered 10 years ago with our science founder, Professor J. Michael Ramsey at the University of North Carolina at Chapel Hill, who had previously been part of a research group at the Oak Ridge National Laboratory working on making a smaller mass spec using miniaturization technology and microfluidics. The common approach at the time was to simply try to shrink the existing technology, but a major issue was that mass spec requires a powerful vacuum pump that needs to reach outer space–levels of vacuum pressure, so reducing the size of the pump was a losing battle. Instead, Mike theorized a solution that does not require such vacuum and started working on realizing an ion trap design that could enable mass spec to work at higher pressure.
High-pressure mass spec still requires a vacuum, but it operates much closer to atmospheric pressure, which allows us to use a pump that can fit in the palm of your hand. Mike was able to accomplish this by using a microscale ion trap. As you miniaturize the dimensions of the trap and increase the drive frequency, theory shows you can achieve comparable resolution even at higher pressures — and in a much smaller form factor.
DA: Within the pharma space, where can mass spec be performed with 908’s technology that it couldn’t before for practical reasons?
KK: Each of our products has enabled customers, across pharma and other industries, to use mass spec as a newly accessible solution for longstanding or ongoing problems. Today, a lot of the application of our MX908 handheld device is in the detection of counterfeit pharmaceuticals and related substances. There is currently a massive overdose problem among young adults who purchase pills of what they believe are conventional drugs like Adderall and Xanax, but in which the active ingredient has been replaced with fentanyl. The Journal of the American Medical Association conducted a study that showed that the number of adolescent deaths from drug overdose doubled between 2010 and 2021, even though teen drug use is at a historic low, and they noted that fentanyl was involved in 77% of those deaths. We have a wide range of customers for our handheld device, most often law enforcement agencies, customs organizations, or other health and safety organizations trying to stop the tide of counterfeit drugs and fentanyl from entering our communities.
With our desktop mass spec REBEL, we serve the top 20 large pharma and biopharma companies who use the power of mass spec to measure key process parameters, critical quality attributes, and ultimately product quality attributes. We’re providing this technology directly to customers who could never have purchased a mass spectrometer and would normally send samples to a laboratory and face slow, cumbersome turnarounds.
REBEL is designed to sit next to a bioreactor in a process development lab. Today, REBEL is able to measure 32 analytes: amino acids, metabolites, vitamins, and so on. Biologists are able to get that data as often as needed throughout their run of a bioreactor, which could be a 12- to 14-day process. This frequency of data collection is unheard of. Researchers would normally take samples infrequently, freeze said samples, and send some or all out for analysis after the entire run is complete. However, those data are in the rearview mirror; they are not actionable. The rapid and frequent feedback from both our desktop and handheld products allow biologists to adjust and tailor their processes. At 908 Devices, we are excited to create access to high-quality chemical and biochemical data in an actionable way that enables users to adjust their workflow.
DA: How does bringing this formerly offline technology to at line help to advance the industry’s goals around the Bioprocessing 4.0 vision?
KK: Taking the power of mass spec out of the central lab and bringing it ‘at-line’ next to the bioreactor is a major step. Now, researchers can take a sample from the bioreactor and place it in our device for measurement, which is a manual process. There is a lot of interest to push forward toward “Bioprocessing 4.0,” and an ‘on-line’ connection of analyzers and mass spec would be a powerful technique to enable this. We see our future will be to take our technologies and connect them in an on-line, integrated fashion to reduce sampling error and reduce labor.
Miniaturizing and increasing accessibility to this technology supports the Bioprocessing 4.0 revolution by closing the loop in research and development. It quickly generates a data set that can be fed into predictive models and be used as feedback to help adjust process attributes in a live, real-time fashion. By closing the loop, your entire process development time can shrink. Rather than many combinatorial, parallel design of experiments (DOEs), you can do a single timed course and adjust feeds and process in real time to get the best outcome of critical quality attributes, such as titer.
If you were to send out samples for analysis at a core lab, a typical turnaround time is two to three weeks, but there is an additional logistics burden that adds a few weeks of delay. In contrast, by having the REBEL cell culture analyzer right next to the bioreactor, you can have a meaningful result in 5–10 minutes, and then you can run it again as often as you like.
DA: Do the benefits for researchers go beyond simply making processes quicker and more efficient?
KK: Beyond the benefits of speed, the technology also enables manufacturers to use the data in ways that were not logistically practical before, and we’re working with key opinion leaders to drive its use in new areas. We work with groups like Johns Hopkins University, which wants to make in silico predictive models of their bioreactor but needs inputs into those models. The comprehensive REBEL analyte panel enables them to tailor predictive models and ultimately to more accurately predict outcomes.
We’re also collaborating with groups at Emory University that are working to optimize CAR-T cell therapies. They’re looking at these novel microbead libraries for manufacturing, and REBEL is allowing them to measure the amino acids and process attributes to look at the activation of T cells with their microbeads attached. Beyond boosting efficiency, REBEL is enabling a new type of quick monitoring to inform their feeding schedules and to control processes in a way that hasn’t been done before. And the list goes on: we’re working with groups at MIT, CPI, Boehringer Ingelheim, and even Amgen and GSK in areas of critical quality attribute monitoring in a much more rapid, simple way.
DA: Can you expand on the microfluidics technology and what it can enable?
KK: We have a technology that’s a microfluidic chip about the size of a business card and designed with an approximately 22-centimeter-long channel that weaves back and forth across the chip. It’s designed to desalt, concentrate, and separate on the basis of size and charge over a multi-minute period that can range depending on analyte size but is generally 2–8 minutes. Then, it electrosprays a mist into a mass spectrometer for analysis.
Since the vast majority of mass spec machines on the market have some variation of chromatography that helps separate the sample and spoon feed it to the mass spectrometer, allowing the mass spectrometer to decipher the results for complex samples more easily, we designed our microfluidic technology to accomplish this critical step.
Two of our products use this microfluidics technology. Our ZipChip device, which is an open-access platform that plugs directly into a mass spectrometer, allows us to work with researchers in central labs on current assays, like critical quality attributes. We then took our microfluidics technology and connected it to our high-pressure mass spec in a very purpose-built way to create our REBEL product.
Our technology also opens the possibility of working with innovators on new areas, new applications, and new use cases. We have folks in the academic space that embark on new projects using our technology who are looking beyond critical quality attributes in biopharma and are researching biomarkers for disease states in clinical samples. Other folks are using REBEL more in the context of synthetic biology, even looking at new solutions in the realm of pesticides.
We also have some very interesting next-generation products and technologies coming up in proteomics, where we are using these microfluidic chips to separate many, many proteins at low concentrations much more quickly compared with chromatography techniques.
DA: Are there any limitations or particular cases where it would make more sense to use conventional mass spec rather than REBEL?
KK: For sure. It’s analogous to a large supercomputer versus a laptop or a tablet. The large devices have more “horsepower” and performance settings, but there’s so much that can be done with a laptop or a tablet. We see REBEL as a purpose-built device, like an appliance or a medical device. We look at things like true positive rates and false positive rates, and we also pay attention to things like receiver operating characteristic (ROC) curves that quantify the system because we’ve designed it to do a particular job.
Our REBEL and handheld devices are purpose-built and meant to do a particular job: measuring 32 analytes next to a bioreactor over and over again to inform that bioreactor or testing for trace levels of counterfeit drugs. Our ZipChip device is ideal when customers need an open architecture: they want to discover and develop new application use cases. The tradeoff is that REBEL and ZipChip are much more affordable and suitable to users who are experts in their own fields but don’t want to do method development or to be experts in chromatography or in many cases even mass spectrometry.
DA: How important to you is the user interface, and for what kind of user are you ultimately designing the system?
KK: I appreciate that question because I think it touches on something that really needs to be well understood. We have discussed our hardware capabilities that shrink down the mass spectrometer and simplify the sample prep and separation for microfluidics. But if we had the same interface as a conventional mass spec with all the nobs and the squiggly lines at the output, then we wouldn’t really have enabled a broader reach. In order for our technology to actually enable customers, it is critical that we couple all the machine learning, algorithms, and automation that we provide with a good graphical user interface.
If you take the device out of the central lab and give it a smaller screen but still need an expert operator with a Ph.D. to run it, you’re only halfway there — and if you’re only halfway there, you really haven’t done it.
We are quite literally making devices that can be used by a fireman, a law enforcement professional, a mailroom operator, or someone in a health and safety organization, while also designing our tools to enable biologists to get quantified information for their bioreactor. Our devices must help each of these customers do their jobs faster and with more informed decision-making capability. When our team talks with these customers, we’re not talking about the technology; we’re talking about the analytes — the results and what these results enable them to do.
DA: Can you discuss the role that machine learning plays?
KK: At 908 Devices, we have three pillars of technology. We discussed the high-pressure mass spec technology and the fluid-handling side — including microfluidic sampling and separation — and the third is the algorithmic, machine learning, and related user experience side, which is critical to enabling the control over processes that we discussed earlier. For example, let’s say that a user is measuring a counterfeit pharmaceutical using our handheld device. When our handheld is showing the user a result, it not only indicates if it detected fentanyl, it further specifies if it found fentanyl versus carfentanil, remifentanil, or other substances that could be in the sample. And when it gives a precise identification for the drug, it literally puts a name up on the screen for the user to see. It’s not showing any squiggly lines or any other type of “science-y” data. It’s showing the user a clear identity in a reliable, robust way. To make this possible, we need all the machine learning underneath that drives the device’s ability to interpret results.
Similarly, with our REBEL device next to a bioreactor, we’re giving biologists quantified information for all of those 32 analytes. There are no calibration curves, and there is no need for manually looking at migration times and the mass spec data and then trying to do calibration curves. All that’s been taken out, so users just get a report that tells them: here are your 32 analytes, and here’s your quantitative level. We leverage all of our machine learning automation to deal with the environments to give those robust results.
DA: Do you have a clear path forward in terms of what the next generation would look like or additional functionality and connectivity that you want to develop?
KK: We have a very strong belief that smaller, faster, and simpler analytics can take us very far across all of our end markets. If we zoom in to where we are serving the life sciences sector, including biopharma and bioprocessing, we’re very excited to use those analytics and that expertise to move from at-line to on-line to offer more analytes in an on-line configuration: being able to stream a broad set that’s very informative to how users will run and optimize their processes. Today, we see clearly that analytics are a cornerstone of Bioprocessing 4.0, and we are working hard to make our analytics simpler, more connected, and more robust for measuring key process attributes all the way to critical quality attributes.
DA: Shifting gears a little, how have things evolved for 908 Devices and your team since you took the company public a couple years ago?
KK: We were very fortunate in December 2020 to become a public company, and I often say that it’s been like an injection of caffeine, because it brings dollars; it allows us to scale. We’ve grown our team to around 200 people, which has really allowed us to double down on our R&D efforts and build out both our channels and applications and our sales and commercial teams to better serve our customers. We’re working hard, and I think that we’re successfully leading innovation and maintaining our entrepreneurial spirit even as we scale.
We were recently notified that we’ve ranked number 48 [out of 100] on Fast Company’s fourth annual Best Workplaces for Innovators list; so that’s very exciting news for our company. We resonate with that, and we work hard every day. Even earlier today, we were talking about the impact that these technologies and our customers have in these marketplaces. The prospect of stopping lethal counterfeit drugs and the prospect of helping customers get therapies to market faster are exciting to us.
DA: Looking to the future, what comes next for 908 Devices?
KK: One area in which we anticipate seeing developments is the field of advanced therapies from the process development labs that use REBEL. As the pipeline of advanced therapies that these researchers are trying to develop grows and becomes more complex, these folks need have quick insights to do DoE much more rapidly to bring their therapeutics onto the market. The number of investigative new drug (IND) filings has really been increasing many-fold over the span of just a few years. During Advanced Therapy Week earlier this year, there was a conference talk about how the number of new cell and gene therapies that reach the market is expected to be 10–20 per year, and that they’re expecting to have approvals through 2025.
So, there is a lot of momentum in those areas, and we think that having REBEL’s insights is game-changing. It’s also a little bit of a Wild West right now between the directions that autologous cell therapies versus allogeneic therapies will go and how both could benefit from online and integrated sensors during manufacture. We fundamentally believe that analytics in these areas need to be simplified, because the development of these therapies requires the generation of broad sets of analytics that will be streaming off of these processes. We think our customers will do great things to enable some of these new classes of therapies.