We recently interviewed Uplaksh Kumar, COO at Foresite Labs and Venture Partner at Foresite Capital, a multi-stage healthcare and life sciences investment firm.

‍

Key Takeaways:

• There is a lot of excitement about the potential of AI and ML in drug discovery and development, particularly in areas like molecular design, understanding disease mechanisms, and identifying drug targets.
• There is optimism about the future of gene therapies and diagnostics, especially the use of combination therapies and multi-cancer detection tests, and the belief that these technologies will continue to improve in specificity, sensitivity, and positive predictive value over time.
• Despite a difficult funding environment for life science tools and diagnostics recently, investment is slowly picking back up and hopefully continues through the new year.

‍

Thank you so much for your time, Uplaksh! To start, please briefly describe yourself and your career path.

I started in the sciences, was a PhD by training, and I realized as I was taking that journey that it's cool and interesting to do discovery work. But if you have to bring science and technology to people to make it more available, you have to build successful products and make it easy for people to access them. My journey has been about taking the foundational science and background that I have and bridging the gap, or crossing the chasm, between science and product: how do you make products available and how do you scale up products in really exciting spaces at key intersections like biology and software engineering including AI, ML and platform development. The best developments are taking place at those intersections, and I find myself very comfortable in those uncomfortable intersections; some of the hardest and the best work that I've done has emerged at those intersections.

Thanks for that intro! You've done molecular biology, some genomics-based diagnostics, like Qiagen, Verily, and Grail, and life science tech like Lonza, and Rooster Bio. How did you move from one to another, and was it motivated by this AI or ML intersection, or anything else?

Starting at Digene, which was acquired by Qiagen many years ago, I started getting a feel for how products are built, how science is transformed into products, and how products can help people. I would say one of my proudest accomplishments was knowing that 10 million women were tested with a product that I helped create and bring to the marketplace for early detection of cervical cancer, which is the HPV product. So early in my career, it was all about exposure and building tools and capabilities that I did not possess. I had great mentors along the way who I learned a lot from. I gathered this vast breadth and depth of experience and expertise ranging from very early-stage research to product development and commercialization of products, even things like supply chain and operations, which is very important to get right in any industry. That's the beginning of my career. Once I acquired these tools, I wanted to apply these tools in areas that were novel and different. At Lonza, it was all about building out cell therapy operations when cell therapy was very nascent. It was all done at the bench scale, in the lab, or academic institutions. The first cell therapy facility for commercial manufacturing was built at Lonza. Then it expanded over time. Rooster was also about starting a company, which I thought was going to be very challenging, but very satisfying using an engineering solution for a biological problem. We all know we need cells, but to grow cells at scale, at the highest of qualities, and be able to use them was and is still challenging, quite frankly. It's not easy or inexpensive, even to this very day. I was the founding operations head at Verily, the first Google company focused on life sciences. It was an opportunity I couldn't refuse to build out the infrastructure for a strong software engineering team with the best engineers in the world and apply it to healthcare. More recently, looking at Grail, the biggest problem to solve in the diagnostic space has been multi-cancer early detection. What we have traditionally in place is cancer by cancer detection. Five cancers are covered, but the rest are not. There's an unmet need, but there's also a need for significant investment and different types of expertise that need to come together, such as lab, software, hardware, and clinical to build out this product. It was a very satisfying journey because in five years we went from an idea to a product, which is currently Galleri in the market. Most recently, we launched Xaira therapeutics, which was an incubation effort at Foresite that now has emerged as this company that was announced earlier this year. I'm the interim operating head of that company and I've been working on this idea for about 12 months, and now it's launched with a significant amount of funding. I think the encapsulation here is building tools, understanding, learning, discovering early in my career, and then transitioning into taking these tools, where some unique and key intersections develop not just incremental products, but exponential products. I've enjoyed that, and that's what I will continue to do.

I'd be curious to hear a little bit more about that as well as the most exciting areas you see for AI and ML. Are drug discovery and target discovery the primary applications you see?

It's being used in different areas within the drug discovery and development continuum, ranging from molecular design at the front end where one can iteratively design millions and billions of molecules with the right characteristics using AI and ML, which is not humanly possible otherwise. Secondly, there's so much data that the cell systems are generating. How are they mechanistically connected? Are they connected? That's a big data problem, where we're using artificial intelligence and machine learning models and tools. You still must rely on traditional drug discovery and drug development efforts, but for me, the key is that it's not just using the same tool and applying it across the board and that one tool can solve every problem. It's knowing where some of the inefficiencies are, where can AI and ML be applied, and where do we still have to continue to rely on traditional processes and methodologies to advance this to the clinic. You have to be humbled by biology and not just assume that if you have good models, you can predict everything.

That's very interesting. Sometimes with these large startups, it seems the tremendous resources at the outset can also be a detriment in certain ways to deciding a focused development plan or area to impact. What do you think about that as you're getting Xaira officially launched?

Starting with the obvious, the bar should be high. There are two fundamental problems to solve. One, you have identified targets, but you don't know how to drug those targets. That's one area where this can be used. The second area is where you understand the disease, but you don't understand the mechanism (e.g., the phenotype and the genotype). We're using these two areas to focus some of our efforts as we make investments in this field.

I like that breakdown. Perhaps we can pivot now to your journey across cell and gene therapy. You mentioned when you were at Lonza, cell therapy was still highly novel. Obviously the field has made tremendous strides since then but hurdles remain.  How do you think the field will evolve over the next five to 10 years? Are there any specific indications or technologies you think will be most promising? 

Some of it is driven by not knowing the mechanism at the basic level. Most of these are not used as first-line or frontline therapies. This is used once. So, you're already starting in a very high-risk population. I don't hope or wish for any failure, but that's always inevitable in a situation or environment like this. The modalities that are becoming interesting are bispecifics. You're seeing a lot of targeted siRNA. You're seeing these different molecules where each part of the molecule serves a different purpose. What we're trying to do today is figure out the right combination that addresses either the toxicity issue, the delivery issue, the efficacy issues, etc. We know in isolation, these molecules or these modalities work well, but how do we bring them together to address the major issues, like toxicity, which sometimes lead to death? There's a lot more of that happening than I ever thought I would see in this recent environment. A lot of these “combinations” are being created and tested out, and I'm sure many of them will solve the issues that we've encountered with traditional types of molecules or types of modalities.

Do you foresee these becoming first-line therapies?  

I’m not sure they will be the first line in 5 years. But I think as there's more data that's available, and there's more safety associated with some of these modalities that are better understood, eventually, I think we'll get there. What we want to prevent is the ‘therapeutic odyssey’. It's taking too long. You don't go into second-line and third-line treatments until it's sometimes too late. So, could we bring that forward? But you need more data. You need more safety to bring it into earlier lines of treatment and that should happen over time.

‍

That makes sense, especially with some of the recent FDA warnings regarding cell & gene therapies. Let’s turn to the diagnostic side of treatment now. Guardant had their Advisory Committee in May for SHIELD and Freenome has released clinical data also. Given that both Freenome and Delfi have single-cancer panels, but are looking to expand into multi-cancer panels, how do you expect the market to play out between single and multi-cancer testing, and what do you think is going to be the determinant of success?

Because Foresite Capital is an investor in Delfi, my comments are more general and not specific to what I think about the company in that sense. I think sometimes people are looking at these problems from the wrong lens. It's very easy to be skeptical of these new technologies and say that there's just marginal improvement relative to current standards. If you were the patient and you were the recipient of that “marginal” improvement, you would have a very different answer to that question. I think it's early and these technologies are still evolving. Cell-free DNA for early cancer detection, or even single cancer detection has not been around for all that long, relatively speaking. It's a bit unfair to be skeptical of data. I think instead, we need to embrace and recognize that this is early data that's only going to get better over time, the models are going to get better, and the ability to identify and develop the probes is going to get better. Over time, you will see an improved specificity and sensitivity, which is important for these tests to get wider adoption. Also, it's important to realize, that if you look at the current standard of care, the positive predictive value is not really that great. So, if you're comparing it to something that's not great to begin with, and then you say it's marginally improved, there's a long way to go. A big open space for improvement lies ahead of us, which I think these technologies will be able to address over time.

That’s a very valid point to recognize as I think there is a lot of excitement around this area but it’s still quite nascent. Moving on to your investor side, it’s been a difficult funding environment for life science tools and diagnostics recently. What were the lessons learned from these slowdowns and how do you see it changing over the remainder of 2024 and beyond?

Yeah, I think during Covid, the market got very frothy. There was a lot of investment in anything and everything biotech, Covid obviously being one of the major drivers. I think what has changed, and I believe it's going to improve over time, is the life cycle within biotech is very long to get products from discovery to market. As investors, we wanted to preserve some dry powder to support our existing companies, rather than make new investments in new companies. That way, we can make sure that our existing companies can survive the wave and get to a key value inflection point. So, it wasn't like there was no money available, it was a conservative approach to make sure that we could sustain investments. I think that's going to change, and has changed, and will change through the end of the year when there will be more investment in new ideas that are emerging, perhaps at a much higher clip than we saw happening in maybe the last 12 months.

‍

That resonates with what we’ve seen in the market, where the total amount of investment was similar to previous years but most of it was re-invested into existing portfolio companies. You’ve seen both sides of this given your unique position as both an investor and an entrepreneur, straddling both Foresite Capital and Foresite Labs. Does this enable you any unique freedom in terms of what you invest in and how? 

It's the ability to do early-stage, high-risk investments. And even the fail-fast, succeed-quick investments, where we can get to a proof of concept which we know would derisk further investment. It gives us the ability to take on big challenges, recognize that these are high-risk investments, but also recognize what those key value inflection milestones need to be in an early experiment to justify additional capital. It's a different appetite and a different cycle time of effort and risk-taking that's required.

That sounds very exciting! To wrap up, I’d love to learn about what inspired you to join Foresite, and what's been the most challenging and rewarding part of your journey. 

I get to see and meet some very amazing scientists, whom I believe are the best and the brightest. That's what excites me the most, being exposed to people who are the best in the field and who are pushing the boundaries of science constantly. I'm learning, quite frankly. Even though I've spent 25 years in the industry, I get to learn from these individuals. In addition, seeing the landscape of products and the landscape of ideas that are being transformed into products, and then being able to use my operational background to figure out what it would take. I don't think there's a bad idea out there. I think ideas fail because of execution and underestimation of the risks involved in getting them from idea to project or product. That's my strength. So that's how I complement the rest of my team at Foresite.

Sounds like the hunger always to be learning has served you well. Thanks so much for your time with us today!