David Li - Meliora Therapeutics - Part 3

Developing a Blood-Based Lung Cancer Test | The Rifle-Shot Nature of Biotech | Software Vs. Biotech Product Market Fit | The Super Power all Startups Need

Find us on your favorite platform:
Apple PodcastsSpotifyYoutube

Show Notes

As a podcast listener, you can redeem exclusive discounts with a growing list of biotech vendors and get $500 off your first equipment lease by using promo code “TBSP” on https://www.excedr.com/rewards.

Part 3 of 4. 

My guest for this week’s episode is David Li, CEO and co-founder of Meliora Therapeutics. Meliora's goal is to develop life-saving cancer therapies using cutting-edge science and machine learning. The company derives a comprehensive picture of how drugs interact with cancer biology by combining biofunctional readouts from numerous modalities using advanced machine learning and other computational techniques. 

Join us this week and hear about:

  • Working at Everest Detection and developing a blood-based lung cancer test
  • Excelling in a non-iterative work environment
  • How software and biotech companies think about product market fit differently
  • Why money is a super power for biotech startups today
  • And much more!

Please enjoy my conversation with David Li.

Articles & Resources

Everest Detection https://everestdetection.com/ 

Meliora Therapeutics https://www.melioratherapeutics.com/

Marketing & Sales Strategies for Startups https://www.excedr.com/blog/marketing-sales-strategies-for-biotechs

Lab Equipment for Biochemistry Research https://www.excedr.com/blog/lab-equipment-list-for-biochemistry-research

Strategies for Better Cash Flow Management https://www.excedr.com/blog/cash-flow-management-strategies

How do Core Labs Support Life Science Research? https://www.excedr.com/blog/core-labs

People Mentioned

Episode Guest

David Li headshot
See all episodes with 
David Li

David Li is the CEO and co-founder of Meliora Therapeutics, a biotech working to develop life-saving cancer therapies using cutting-edge science and machine learning. Meliora derives a comprehensive picture of how drugs interact with cancer biology by combining biofunctional readouts from numerous modalities using advanced machine learning and other computational techniques. Before founding Meliora, David served as Chief Business Officer at Everest Detection, an early detection liquid biopsy startup, and was head of commercial operations at Benchling, a life sciences software platform company which has achieved a $6 billion valuation and has attracted over $500 million in funding from firms such as Altimeter Capital, Tiger Global, Lone Pine Capital, Benchmark, Sequoia, Excel, Lux, and numerous others.

David began his career in Goldman Sachs Healthcare Investment Banking Group and KKR's Private Equity Group, where he advised and invested in transactions worth over $10 billion in enterprise value. David's multifaceted experiences in the financial and life science sectors give him a wide range of experiences that founders can learn from.

Episode Transcript

A hand holding a question mark


Intro - 00:00:01:


Welcome to The Biotech Startups Podcast by Excedr. Join us as we speak with first-time founders, serial entrepreneurs, and experienced investors about the challenges and triumphs of running a biotech startup from pre-seed to IPO with your host, Jon Chee. In our last episode, we spoke with David Li about pivoting from finance roles at Goldman Sachs and KKR to early stage entrepreneurship. His time at Benchling as Head of Commercial Operations and Everest Detection as the Chief Business Officer and the challenges of moving from private equity to startups. We also talked about aligning personal values with career choices, building solid teams and the role of luck in success. In part three, we talked with David about his experience at Everest Detection, focusing on early stage lung cancer detection and developing a non-invasive blood-based test. We'll also discuss the differences between life sciences and software innovation, product marketing fit, de-risking core hypotheses, and distinguishing between one-way and two-way decisions. Additionally, we'll cover fundraising, capital allocation, and the importance of knowing when to pivot or cut projects.



Jon - 00:01:26:


So now, you know, after Benchling, you're over at Everest Detection. Can you talk a little bit about now being like number two company building? Can you talk about that experience?



David - 00:01:35:


Yeah, it was a very early background was Brian's former company, Stemcentrx, was a small cell lung cancer focused company. Lung cancer, broadly speaking, is one of the areas where you're usually asymptomatic until the late stage. But the survival curve drops off dramatically as you really progress through the different stages. And so your overall five-year survival at stage three, even still today, is around 20-25%. If you can catch it stage one, stage two, you're hovering above the 80-85% survival rate. Because mostly what you can do is just scoop it out and then you can treat there and you're pretty much good to go. As long as the tumor hasn't really metastasized. And so the challenge really is to try to find it early. In our country, in the US, you don't usually do any CT-scan or any other type of other screening technology unless there is a symptom, there's a reason to do so. There's a bunch of arguments for why or why not to do this. Some of them are health economics, is it cost effective? Are there too many false positives, etc. The second important factor is doing a CT-scan does expose you to radiation. And at that age, there's cost benefit trade off of whether or not you should be doing that at the population wide level. And so the motivation for us thinking about, is there a possibility we could do a non-invasive blood-based test that would be for lung cancer specific early stage cancer detection test. And at the time, Grail and Thrive and a number of others hadn't really hit the market yet. They definitely started. So people were thinking about it, but they were also working on something different. They were working on a multi-cancer test versus doing a single cancer test. And so clinically, there'd be just a very different use case. And so we didn't know whether it would be actually technically possible to develop a blood-based lung cancer early detection test. There's a lot of physiological reasons for this. You need to have the actual cancer signal seeping into the blood at a significant rate such that you can pick up whatever target or signal that you're looking for. You need to be able to pick it up. And so there's a lot of reasons why this is a more difficult challenge than, let's say, colorectal and a couple of others. But this indication then, because Brian had already previously been working in the lung cancer space, we were able to get going in terms of access to some clinical samples, early stage lung cancer patients, and also why Brian had strong conviction that this was a real problem, that the way to really move the survival curve and to save a lot of lives is to detect early. And so I started thinking about that problem alongside Brian, and we talked to a number of different folks. Technical and academics, folks to think about diagnostics and whether or not this was possible, whether or not the markers, biomarkers we were looking for would really show up in blood. And so we ended up starting doing an internal R&D and developing a blood-based assay. And this was very different. Yeah. To your question on the differences between life sciences, technology and innovation versus true software, I learned how to, one, work with scientists rather than software engineers. They're related breeds, but distinctly different. Then I also learned a lot of what I would call early stage organization building at a very different business cadence than your usual make and sell and iterate with the customer. Here, you're saying, here's the spec of what we want to get to ultimately in terms of a product. What is an engineering way to de-risk the possibility of reaching that type of prototype in the shortest amount of time possible? There's no iterative customer feedback. There's no, you put out a product and you make it better. And then you start taking more share. This is like a rifle shot. You got to figure out exactly what you want. And you better make sure that you have the resources then to de-risk actually getting there. So it's a very different business cadence and something that I had to learn. And I think I'm still learning, in fact, on just kind of how to get this right. Because there's the balance of, on the one hand, when there's constant customer feedback, there's constant customer usage of the product, there are clearly ways you can improve it. Then there's a very quick feedback cycle that you're saying, hey, every day, let's make an improvement. Every day, let's make a change that's going to ultimately feed into some commercial success. Here, I think it's much easier or more natural to take into, like, because there's extremely high consequences for the decisions that you make, let's take a lot of time to really be sure about the decisions that you make. And I think we had to really learn, I've had to learn kind of the balance of the two. What are decisions that you can take time, you shouldn't take time? What are decisions that you would benefit by moving with more urgency and moving faster? I think there's certainly a lot that can be learned from both sides, actually, vice versa for each other.



Jon - 00:06:34:


That is a really interesting concept. And I'm curious, what is your decision-making framework for, we need to measure many times before we cut versus like, one, two, cut. What is like your framework for thinking about that during, and now that you're in it?



David - 00:06:51:


Yeah. You know, now I would say, I think about things in terms of, given the budget I have, how many parallel experiments can I run that the spectrum of results from those experiments run in parallel, gives me the most information to make the most informed decision as the next kind of tranche of decisions. So the balance of the urgency versus the thinking, the way I think about it now is, we should already know what the decision tree is the minute we get the data, the minute we get the outcomes. We're just, we're waiting. Yes, we're waiting, but we're actually really observing to think about what is the full spectrum that can come out of this and then game plan for each of that spectrum. And so, which, try to minimize as much as possible. Do things serially, right? You get data, you think about it. You kind of really need to do a lot of research and thinking and outreach. And then you're like, okay, this is what we're going to do. And then you get another piece of data and you're like, okay, so now this might change our path. And then even if you do manage to shrink down that decision time, once you change one thing, you may have to go back and change the other thing. And so now, as much as we can, we do things in parallel. There is obviously the challenge where you do things in parallel, your burn rate does go up. And so you have to think about how do you do this judiciously. But at the same time, maybe you can leverage some existing infrastructure, some overlapping experiments. So those are types of things that I think you can learn a lot from the way that tech does it in terms of iterating quickly to something that works very effectively in parallel. I think that the old broader decision principle is anytime you reach a node in the decision tree where it is a very irreversible decision, then you think about it and you make sure you're making the right decision. But I think for those types of decisions, you should have been thinking about it for a long time, all the way from the very beginning when you're picking a target and saying, okay, this is the profile, you're setting a clear no-go, right? It's still tricky. It's definitely still tricky. I've found from my time at Everest to now my time running Meliora, you do get a sense on whether or not this data is real and whether or not this is worth investing further in. Because the danger with science is you get kind of a middling outcome. You could always do more work. You could run another experiment. You could run another condition. You could run another molecule. You could try to improve them all. There's so many things you could have done. There's kind of no end decision point. And that is the most critical decision. I think a leader you need to have a really strong discipline in making. And it's hard when you don't have a very good benchmark of no one knows whether this is going to work or not. So how do you really take that right balance of not being too aggressive and pivoting away, but at the same time, really making sure you have a sense of urgency and making forward progress?



Jon - 00:09:33:


Absolutely. But with the life sciences, generally, you start off with product market fit. You already know. Curing cancer has product market fit. You know that from the get-go. Whereas a piece of software, you're like, I don't know. We can launch this in the market and no one's going to like it at all. So there's different risk profiles. Obviously, on the biotech side, you have the product market fit, but then you face exactly all these R&D hurdles and regulatory hurdles. And then on the software side, less regulation, more of just like, do people like it? And can we iterate fast enough to the point where people really, really like it? So that's always a bifurcation in my head that I see. But I love you talk about this iterative process that is now coming into Biology.



David - 00:10:16:


Yeah, I think so. I totally agree that there is an answer for whether or not this thing's going to have product market fit for the lifelines. You should know. But what I've actually found oftentimes is, especially in the earliest stages, more time, more iterative thinking can go into what is the actual product profile that is going to have that product market fit. When we talk about cancer, okay, there's a safety profile, there's a tox profile, there's a dosing profile, there's a usability, obviously there's efficacy. What is that bar clinically? And the challenge I found in biotech is you need so many people around the table. You need a cancer biology targets person, you need a pharmacology drug discovery, you need a chemist, you need a CMC, you need downstream regulatory. And that's before you even get into the clinic. And then once you're in the clinic, you have early clinical development. You're talking about early dose escalation, finding safety signal. Then you can get really creative if you have a regulatory path that gets you accelerated approval or breakthrough designation, et cetera, et cetera. Those strategies are very different. And they all then lead into ultimately, is this a product that is going to move the needle clinically in the treatment paradigm for a patient? And oh, by the way, all the analysis you did years ago for when your product started at the beginning of that funnel is no longer relevant. You need to be able to project into what will happen when you arrive the finish line, if and when you arrive at the finish line. So I think that people definitely say there's going to be product market fit, but it's actually pretty challenging. Life science to really prognosticate puck is moving somewhere about where that product market fit is going to be. And what does it really look like? And how can you kind of shoot it out of the sky when you know it's going to take us five, six, seven plus your journey to get there? It's almost like in the beginning, it almost sounds it's easier to get product market fit in the life sciences because you kind of know it and it's harder in the software. But now as I've stayed longer in therapeutic side, I've seen that actually it's not as straightforward at the very least of ascertaining whether or not you have product market fit on the life sciences side. Whereas in the software side and tech side, you know, right away in the business, if you're experienced up, you can kind of just know right away whether or not this company has hit PMF product market fit or not. So there's nuances on both sides. It's definitely a complicated issue for both types of businesses to tackle.



Jon - 00:12:40:


Absolutely. And very, very broad strokes here, because I think what you're alluding to on the biotech side is like, it's the product market fit and the intensity of it, right? If you're not going to get reimbursement and physicians aren't going to prescribe it, even though we know it's a, you know, again, broad strokes, cure for cancer, this type of cancer, then it's kind of like dead in the water and your bark is not even able to interface with it. So all this kind of strategic planning upfront, you at least have a North star. But like, again, like you said, the course changes. And like, do you stick the landing? Like you have the product market fit. Can you stick the landing with all these multivariate? But yeah, I think something on the software side about that iterative process too, you're absolutely right. You can quickly kind of tell maybe, in like the go-go days during Zerp, you can kind of disguise the product market fit by just pushing cash into the business. And then, you know, you can kid yourself that product market fit is there because you're like, I'm selling a dollar for 50 cents.



David - 00:13:34:


Right. Exactly. Who wouldn't take that?



Jon - 00:13:37:


Who wouldn't take that? Who wouldn't take that? We got product market fit. So there's kind of that weird element to it as well. Different industries for sure with their own nuances, but always something I noodle on for anyone who's embarking on this journey. And at Everest, you're talking about learning sequentially versus learning in parallel. At Everest, were you guys able to do the parallel learnings? Or is that something that you took away like, oh, we're doing this sequentially right now. And then you take that now to Meliora?



David - 00:14:02:


That makes it both for sure. I think towards the end we definitely did a bunch of different kits testing and DNA prep and library prep, et cetera, testing that we would do in parallel and things like that. So I started learning. But I think what I mean by now, what I mean by parallel learning is what can you do to test the core hypothesis versus just like in parallel optimization of the problem? Let's say there's a core hypothesis that sits at the center of whether or not your drug or your product or your diagnostic is going to be successful. The fundamental bet that you're making is this class of targets is represented enough in blood. And you're going to find using this type of either deep sequencing or whole genome sequencing, or if it's targeted sequencing, that your method to getting that signature is going to unearth the signal. And it's going to be scalable, production ready. There's enough signal there that you could make it production ready and scalable. There's fundamental risks that you're talking about here in terms of product design that you should start testing in parallel. You're trying to de-risk that as early as possible so that you can, one, not spend as much cash, but more so, two, not spend as much optimizing stuff that doesn't really matter. If you're just swapping out chairs on Titanic, you got to-



Jon - 00:15:16:


Yeah, yeah. That's a great metaphor. I'm just like visually in my head, just exactly what you need to ascertain early. You need to ascertain that early. To go back to the one thought about these decision trees and the finality of some of your decisions, I think the Bezos saying of like, is this a one-way door or a two-way door decision here? And just testing that every single time, like, what are we looking at? Because if this is a one-way door, we need to be so certain before we put time, money, people power behind this. The way the framework of de-risking this is critically important versus just like Craig.



David - 00:15:51:


Yeah, yeah, absolutely. I think also the third, potentially fourth axis is just your money, right? If your money situation is, this is really a superpower for folks in the biotech space. It's a superpower for any entrepreneur. But if you, especially I think in biotech, you're not getting any money in door. Generally speaking, on a consistent basis, maybe you do have a partnership that can change. There's certain companies out there that really fundamentally changes their financial trajectory. But generally speaking, you're looking at a big amount of investment up front to ultimately get to a product. And I would say that if you're able to raise enough such that your experiments then can be done in parallel easily, then that's actually a very significant advantage. I didn't really appreciate that as much as when I first came in, you're from a software background, software background in the beginning, you shouldn't raise as much money. You're taught, take as little dilution as possible. If your engine's really working, you're spending a dollar to acquire a customer that will pay you back three dollars over time, then this is going to be a fire that grows bigger and bigger and bigger. In biotech, that's not necessarily true. In biotech, it's a mix of time versus money versus risk. It is really that you're really trading off the three variables here all the time. That's like the third leg here is if you have enough money and you can go raise that money by being a stellar fundraiser, storyteller and can really convince folks. I think that's a really big part of probably success overall for the platform, for the company that I really needed to learn how important that was and exactly how to allocate and titrate our capital resources. The way to accelerate to value inflection points. Because as I mentioned, it's always going to be a mix of those three. You're always taking on some sort of risk or more time or you need more money.



Jon - 00:17:40:


And all of which are finite resources. The way you elucidate it is like, you can just ride out this one thing forever, but it's critically important to know when to say it's a no go because of that nature of where you can just keep running more experiments, more assay, just keep digging, digging, digging, digging. But then if you don't cut it off, that's going to be the black swan event where it's just like time is out.



David - 00:18:02:


You didn't reach the next value inflection point, right? And so it's really critical. The other thing I would say about biotech and the financing side is also in relation to the parallel decision-making. What I found is actually, once you're at a certain scale, everything in preclinical is generally speaking, a fraction, a fraction of a fraction of what you'd be spending in the clinic. So if you can, it really pays to be figuring that out, where even now, where I'm sure some of the decisions I think are one-sided doors are not necessarily, yeah, because we're still pretty early. But at some point, that's not the case. And we just need to know the difference between the two.



Intro/Outro - 00:18:40:


That's all for this episode of The Biotech Startups Podcast. We hope you enjoyed our conversation with David Li. Tune into part three of our conversation to learn more about his journey. If you enjoyed this episode, please subscribe, leave us a review and share it with your friends. Thanks for listening. And we look forward to having you join us again on The Biotech Startups podcast for part four of David's story. The Biotech Startups Podcast is produced by Excedr. Don't want to miss an episode? Search for The Biotech Startups Podcast wherever you get your podcasts and click subscribe. Excedr provides research labs with equipment leases on founder-friendly terms to support paths to exceptional outcomes. To learn more, visit our website, www.excedr.com. On behalf of the team here at Excedr, thanks for listening. The Biotech Startups podcast provides general insights into the life science sector through the experiences of its guests. The use of information on this podcast or materials linked from the podcast is at the user's own risk. The views expressed by the participants are their own and are not the views of Excedr or sponsors. No reference to any product, service or company in the podcast is an endorsement by Excedr or its guests.