How can artificial intelligence help our scientists find the hidden patterns in cancer's complex biology? In this episode, host Danielle Mandikian is joined by Anwesha Dey, Distinguished Scientist and Executive Director of AI - Oncology and Cancer Biology, to talk about the new reality of the oncology lab. They explore how teaming up with artificial intelligence allows researchers to challenge long-held assumptions, generate bold hypotheses and design novel molecules at speed. Join us to hear how working alongside AI is accelerating the path to new cancer therapies.

If you would prefer to read a transcript of this episode, please click here.

SUBSCRIBE BELOW TO CATCH EACH EPISODE

If you want to learn more about the groundbreaking science happening in our labs, click here. To learn more about the jobs in our research and early development group, click here.

Transcript of Two Scientists Walk Into A Bar: “AI in Oncology” with Anwesha Dey

Maria: I’m Maria Wilson.

Danielle: And I’m Danielle Mandikian.

Maria: And we are scientists. We. Love. Science.

Danielle: Yeah, we do. So, when we aren’t doing it, the next best thing is to talk about science! And what’s really awesome is that we’re surrounded by some of the most brilliant minds in research!

Maria: We are going to step away from the labs today to talk to other scientists about the cool stuff they are thinking about, working on and imagining . . .

Danielle: . . . as well as how some of these discoveries just might lead to new medicines. So, grab your favorite drink, get ready to unlock your science brain and join us for Two Scientists Walk into a Bar…

Maria: The show for scientists, science geeks, and the people who love them!

Danielle: What difficult problem in oncology do you think AI is going to actually help solve?

Employee responses:

That's a good question. Maybe faster molecule discovery? Faster go-to-market?

Probably like specialization to like the patients and whatnot.

I think right now, AI is pretty good about helping with brainstorming, like, new ideas that you may not have thought about, as well as it helps with the huge amount of data that we have today to kind of make sense of it, right?

It would probably be something like finding new targets.

Early detection, perhaps? But I'm only guessing.

Well, that's a very good question. I think there's really a lot of really good ideas and thoughts that we can think of. For example, you can ask AI to do a literature search and go around the clinical trials and design, like even the different trials in terms of how I can drive my clinical trial forward. And then with the specific drug that I designed for a specific population of patients, like how can we expand that to other populations with the interest of their genetic background and then all those, like, big data that's embedded. I'm excited about it! [Laughs]

Danielle: Hi everyone! Welcome to Two Scientists Walk Into a Bar. I'm your host, Danielle Mandikian. Over the last seven seasons, oncology episodes have definitely been some of the most popular. And it's not just the prevalence of the indication but the real breakthroughs that have kept our hopes and curiosity piqued. So we are so excited to invite Anwesha Dey to the bar to help us learn more about AI and how this is revolutionizing the field of oncology. Specifically, she utilizes AI for novel target discovery. Anwesha, thank you so much for joining us.

Anwesha: Thank you for inviting me. Pleasure to be here.

Danielle: So to kick us off, I'm going to ask you to take us back in time. What was your journey into research pre-AI?

Anwesha: I would say my first exposure to research was as an undergraduate student, and I absolutely fell in love with research. My PhD was in structural biology and NMR spectroscopy, and as a postdoc, I made the switch into cancer research, mouse models of cancers, and then started my lab in discovery oncology where I got to work on signaling pathways and nodes that are important and deregulated in cancers. And then, eventually, really fascinated by the drug discovery process and how to translate those ideas in the lab, the basic research ideas and results, into making medicines for patients.

Danielle: When we look at the oncology field, what do you think are some of the biggest breakthroughs for oncology therapies?

Anwesha: If you ask me to name one, I would have to say KRAS – like being able to target KRAS, what we all once thought was undruggable – right? – as an oncoprotein. Knowing the structure, finding that pocket, doing that chemistry and finding that molecule. That really opened up the field, but the thing beyond that is this was so fundamentally important because that one study, which started from KRAS G12C, opened up the field broadly for KRAS into the other mutants from KRAS G12C to G12D, to now so many others. But this also brings up a new, another aspect there on what might be some of these emerging drug discovery technologies that made these possible. Right now, it seems like broadly for the field as a whole, there's just so much excitement about molecular glues because they could be the next – or they are in some ways the next set of tools and technologies as we understand more about the biology, it's a great way to then, you know, bring together fields of drug discovery and basic biology to really uncover the rules of the game.

Danielle: Yeah.

Anwesha: To make the next set of targets, the undruggables, druggable.

Wellington: Hi Danielle.

Danielle: Hey Wellington.

Karen: Hey Danielle.

Danielle: Hey Karen!

Wellington: I know this has come up in other podcasts, but can you tell us what KRAS is again? And why is it so hard to target?

Danielle: Yeah, KRAS is just an acronym for this really important oncoprotein. It's a signaling protein that's really critical in a cascade for development of cancer, and a lot of people were trying to target it, but it was really actually hard to get something that bound to it well, just because people didn't even know the structure of it. So that was one of the big breakthroughs in the field, is having these crystal structures, it gave people some kind of ledge to build off of for binding. And whenever she was talking about the different variations of like D, C, whatever – these are just different mutations that have been found on that protein.

Karen: Danielle, my first question is about molecular glues. What are they?

Danielle: Whoa, whoa, whoa – Karen, get out of my brain. I was just about to ask her that question!

Danielle: And for our listeners, can you explain what molecular glues are?

Anwesha: So you could think of molecular glues very simply as a glue or a small entity that glues your target of interest to perhaps the cellular degradation machinery or another molecule that takes it to the cellular trash to get rid of it. The field is emerging in so many ways that it doesn't always have to even degrade. It could sequester your target, right? So it's – think of it as a way to glue two aspects that you want to bring together to go after.

Danielle: This is horrible, but that's actually how I kill spiders.

Anwesha: [Laughs]

Danielle: I have a huge fear of spiders. Yeah. [Laughs] Okay, so you named KRAS as being like one of your favorite breakthroughs. What's something that you're – that maybe hasn't gotten there yet, but you really think is going to be super promising – that you're really anticipating as being like a big showstopper at some point?

Anwesha: So if you think of KRAS – right? – as the most prevalent oncogene – on the flip side of oncogenes are tumor suppressors.

Danielle: Hmm.

Anwesha: Now there, that's an open field in some ways. What's our guardian of the genome?

Danielle: Right.

Anwesha: p53, right? So it feels like that is the next – in some ways, wouldn't that be an exciting next frontier? There's a lot of work going on broadly across the research community. We've been working on this for so many years. Scientists have been – like, I guess, if you count up the number of papers on p53, it's a staggering, shocking number. We still don't have a way to make that druggable.

Danielle: Yeah.

Anwesha: That is something on the biology front, I think, and not just biology because that's a clear example, the tumor suppressors are lost, right? And it's – the fact that we haven't – don't have one for all the p53 mutations or loss of p53 is not for a lack of trying. So this is an example where, together with the biology, it's going to be a really strong collaboration with drug discovery, right? The modalities.

Danielle: Yeah.

Anwesha: And so I feel like the tumor suppressors is one where we can truly bring together biology, drug discovery, new modalities. And that is one that I really hope is the next KRAS frontier of sorts would be tumor suppressors. And I said I'd give you another one, which is also a super exciting one – and that's in the context also with AI – and I would say AI protein design, minibinders. And really thinking of, how can we leverage that? And there is so – the field is so rapidly advancing in the context of minibinders, AI protein design, lots of labs, lots of companies. The next step is really making that into real medicines, right?

Danielle: Yeah.

Anwesha: And so a really potent binder is not the same as a medicine.

Wellington: Danielle, definition alert. What is p53? It sounds like an airplane.

Danielle: [Laughs] p53 is one of the most classic proteins that people have been interested in in oncology. It's really an essential protein in understanding, within the cell, when there's a problem and knowing to kind of – when to self-destruct. And so when there is a problem with p53, cells don't realize that things are messed up and they just keep proliferating.

Karen: Anwesha was talking about AI protein design and minibinders at the end there. Can you explain a little bit more about what those are?

Danielle: From the way I understand it, they're using AI to figure out, hey, if I have target X, I need this protein structure in order to bind it. And in the context of minibinders, like, instead of having to design a really large biologic, like an antibody, you can design a much smaller protein that contains that binding domain, and that's what they're referring to as a mini binder.

Karen: And do you think that whole field is like AI protein design?

Danielle: Well, I mean, AI protein design is really just saying like: if I want to have a structure that is doing something like binding to a target, what would that protein have to be to have that structure? Because one of the things that's really tricky with proteins – and I think we had Rich Bonneau kind of talk about that a little bit in a previous episode – but when we think about protein binding, proteins are not just like a linear chain of peptides, they have these 3D structures within themselves and with other proteins that form like a larger complex. And so if you need to bind to something, it's not as simple as saying, hey, I gotta have these amino acid residues that have these properties that are gonna bind. You have to figure out what that whole 3D structure would be in order to have those amino acids in the right, you know, 3D space to actually bind to that target. And so that's like a whole thing.

Danielle: Recently, in the literature, there's been a huge uptick in how clinicians are utilizing AI for diagnosis. Are we seeing that same kind of momentum in early research?

Anwesha: Absolutely. If you think about that, the early research, the biggest momentum is in that drug discovery, the molecule-making space. And the next one that is also equally exciting – but a lot more to do – is in the target discovery space. How are we leveraging AI for finding new targets from all of these large datasets? We have heard about the virtual cell and the foundation models – you've had podcasts on this before. So yes, absolutely. Early research is super exciting!

Danielle: So I've heard you suggest that we should really be thinking about AI as a member of our team. Can you go a little bit deeper and help us understand how that is gonna push oncology forward?

Anwesha: So as I think of our “new team member” as AI broadly, it's really many members under that same umbrella “one member.” So if we think of that umbrella “one member” as our AI agent, the first one, you have a tool or you think of an AI agent, which is really good for literature. If you are getting into a new topic – right? – and you want to know all about it, now certainly you cannot read 30,000 papers –

Danielle: [Laughs]

Anwesha: – within a reasonable time.

Danielle: Yeah.

Anwesha: But there is a great agent – you have a great team member who will give you a fabulous synthesis. You ask the right questions. You put in the right prompts. And that almost needs a master class in and of itself. If you're really good at prompt engineering, really know how to ask the right questions, you have an agent for literature synthesis, right? Then, there is one you are brainstorming. You need new hypotheses, right? You have one team member doing that. You are trying to understand modality. You have one agent thinking of that. You did some experiments, you have a bunch of data. You have one agent who is doing your data analysis. Based on this analysis, what should be your next experiment? And you have ideas brainstorming on that, right? Then, you have another aspect of, oh well I did these experiments, I got this data, it got me thinking about I should have some more new ideas, some more new hypotheses. What would those be? Think of all of these coming together – and you are doing this with your teams as well.

Danielle: Yeah.

Anwesha: And everybody has all of this. Like think of the opportunity there to truly augment and accelerate you. And then, you are done, you go home at night, right? But before you go, while you are sleeping, before that you had uploaded some of the datasets, put in some basic question – and you have an AI agent who, while you were sleeping, looked at your – the datasets, went through a bunch of those datasets, came up with some hypotheses, tested some hypotheses and did some analysis, ran some more additional analysis based on the results and then morning came up with these were the hypotheses. True, maybe not true. Tested them based on your data, said this is true, likely true, some of them were wrong. And next morning, you come in, you have all of this, right? But then, it's not like these are all perfect. This is not all solved. And this is not a replacement of your scientific judgment. What I find is really exciting is that there is truly an opportunity to do this, do this at scale, for all of us to do that, right? And think of it as a way to scale and augment what we are doing. Now, is this all perfect? And is this that we don't have anything to do? No, not at all –

Danielle: [Laughs]

Anwesha: – because there are some absolute critical watchouts. And then, if we discuss some of those, on the literature search, the thing that is emerging often across, if you evaluate a lot of these, is that many of them – or some of them maybe – are very, very good at deep, narrow, specialized tasks. So I have an agent that does really good literature synthesis. But that doesn't mean that makes it great for hypothesis generation. That's a different one, right? So you have to know which of your agents is really good at which tasks.

Danielle: Like what's the scope for each one?

Anwesha: What's the scope for each?

Danielle: Yeah.

Anwesha: Within those scopes, the thing that becomes absolutely critical and that sort of is – reminds me of, you know how when we were growing up and you go to do math, or now we do that with our kids –

Danielle: Yeah.

Anwesha: – show your work. I'm not just interested in your math answer, I also want to see your process of thinking.

Danielle: Yeah.

Anwesha: It's like trust but verify. It's not that you come in and you see an output, a polished report – it’s not a reliable scientific report.

Danielle: Totally. Hundred percent.

Anwesha: Right? So these are the things to keep in mind. Just because you had an excellent report next morning or an excellent analysis done next morning, doesn't mean that it's correct. That's where we all come in. That's our scientific judgment. Some agents, they show their work, including the code, right? That's where at the end of the day, things that are auditable, reliable, trustworthy and at the end of the day, show their work – those are the ones over time you get more confident in using and they're all getting better.

Karen: Danielle, so much talk about agents that I feel like I need to get one. [Laughs] But in all honesty though, there are other tools that we're using in our AI toolbox – right? – for oncology research. And what are they? Can you remind us?

Danielle: Yeah, we've totally touched on this across different episodes, and it's actually helped me learn a lot from my vantage point, right? And so, there are large language models that are helping us kind of translate the language of proteins. There's foundational models that I can't even remember the specific examples, because those are just so massive in terms of answering really complex questions. But we're putting all these different kind of biology information into these foundational models so that we can ask more complicated questions about how proteins are actually interacting with each other and what could be the consequences of this. And so, whether we're looking at a disease indication like, I don't know, like you know, some kind of like immunology response to something or we're looking in the oncology space, all of it is really centered around understanding complex systems and what the biology interface can actually be used to intervene with.

Karen: Are you a secret agent?

Danielle: I would never tell.

Danielle: Could you give us some examples of how you're actually directly applying this within the oncology space?

Anwesha: So I guess one way to think about it is this is such a rapidly evolving field. As things are progressing, agents are evolving. Our understanding of how to use them is evolving. How to ask the right questions, the right prompts, and how to match the right questions in oncology with the right agent is evolving. I often think of where we are really now is evaluating them – evaluating the agents to match the right agent to the right question. So for example, there are certain agents, which are really good at literature synthesis. I still think of them in that productivity gain bucket that we were talking about. When I'm using agents there for a new area in oncology, a new target in oncology or should we be going after this target? What is known in the literature about it? What are the emerging mechanisms of resistance? What are some of the other therapeutic opportunities in this indication? So there's many questions there. When I evaluate there, it's which is the agent that gave me the best output? It's concise, it answered my questions, it stayed on track. Sometimes, we also know, it's my thought partner when I'm brainstorming. Another example where I think is a good opportunity for us often even on teams – and this is true for not just oncology but even broadly science and research, right? – we are asking a question. I already know a bunch of things that didn't work as part of the experiments that we did in the lab, right? And so then as part of the prompt, I'm like, exclude this ligase, exclude this. It could be –

Danielle: I see.

Anwesha: – any of that. Agents that stay on track and don't get completely derailed, so that's super useful, but that's also part of our evaluating that right narrow space. You know, the other one that we are also evaluating, which I feel is super exciting, is that idea of new hypothesis generation.

Danielle: Tell me all about it. [Laughs]

Anwesha: So there – I think that's something that is really in the, in my mind, I feel is quite in that early space, super exciting. The possibilities are endless, but a real watch out because then, are we really going and doing a bunch of experiments downstream to evaluate these? Because that's the bucket of that creativity gain which I find super fascinating. That's my brainstorm partner. That's where I cannot analyze all our RNAseq data. I can't go through all the supplementary figures, tables, all the supplementary tables of all the papers that are there on my favorite target in oncology.

Danielle: I think most of us, we’re using AI more and more – like for myself I've used it for, you know, harder math problems, figuring out ways to do data analysis that I haven't done before. But I haven't used it for hypothesis generation. Can you give us an example of one of these agents for hypothesis generation?

Anwesha: So as part of the prompt, what has been fun, there's some agents where you can also go back and forth, right? I put in my question. Oh I'm thinking of this gene X. Is there something known about this before? How do I target it? What might be the resistance mechanism? What is the downstream consequence of this? Or, as part of the mechanism of action, I suspect there might be an E3 ligase involved. Which one might that be? However that is, whatever your question might be. As a response, you might get another set of “are you thinking of this?” And then you have an opportunity to correct your prompt or you might clarify. So the ones which are your best thought partner are the ones that you can go back and forth in. And what am I really looking for there? Often, in this specific context, I'm looking for those unexpected connections. For all of us right there – for things for our projects, we read a ton of papers, we’ve thought about this a lot, we've done a ton of experiments – but what are those unexpected connections that I didn't make, that I hadn't thought about, right? This is an opportunity to do that.

Danielle: I see. So it might be like, I'm interested in targeting protein X, and you go back and forth and it says, "Oh, hey, watch out. There's been something about toxicity or some downstream effect."

Anwesha: You can have it way more concrete than –

Danielle: Yeah.

Anwesha: That's an example, and you can also have it way more concrete than that. You can say can you design some aspects of this, right? And you can go – or come up with specific experiments within the realm of this is the specific modality I'm thinking about, what is an experimental workflow? And what would that look like? And maybe aspects of this may not be new, but it's something that you have not yet thought about. Or you upload, you know, a bunch of your large data that you collected over time and ask certain questions, and you might get an unexpected connection of, oh, in this dataset, it connected multiple of those and uncovered perhaps either a downstream target or an upstream target or a node that you hadn't thought about, right? These are the possibilities.

Danielle: Whenever you're thinking about AI in oncology, what do you see as like the quick to address now versus like in the next five years?

Anwesha: So one of the things that's really exciting even now and here to help accelerate the discovery process is, for the targets that we have, leveraging AI for molecule making. Or leveraging AI to execute on those. The next phase is from all of the developments currently across our screens, foundation models, virtual cells is understanding causal biology to then leveraging AI to find new targets. That is something that we're all headed towards – something that will be really exciting as the next phase to see evidence and eventually, further down, all of this being an integral part of all our workflows, as autonomous.

Danielle: That's really exciting. Do you see this being applied more for very specific oncology indications or more across the field broadly?

Anwesha: Overall, where I hope we're all headed is that this is broadly applicable across all of oncology indications, right? Different people in the field, different companies are focused on different indications, some in certain areas, some others. At the end of the day, our patients are waiting. There is a sense of urgency. It's about collaborating with everybody, and I hope where we are headed is a cure eventually. Right now, we're in the process of treating – but imagine the day where we have all worked together to work towards a cure and eventually towards prevention of cancers, and AI having been an integral part of that, where we all work together as a team to make that happen.

Danielle: That would be so exciting to see that really kind of explode within the oncology vaccine space. Anwesha, I really just want to thank you for joining us. This has been so fun. I always love getting a chance to interact with you and learn from you. Thank you for being here.

Anwesha: It's been an absolute pleasure. Thank you for having me.

Wellington: Danielle, nice episode. Ah, we've done a few of these now, correct me if I'm wrong, but I feel like that I can start to see the picture of a new type of – I can to start to see science being done differently. And so, this relates to the speed of how you are doing science, the fact that you may be able to do more experiments in parallel, and also this new thing that you just talked about of hypothesis generation. What do you think about that?

Danielle: Yeah, everything's changing and I think it's happening in real time. I feel like things are changing so much more rapidly now than what I've experienced in the past. And why that is is that, you know, we're able to kind of, like, dig through the literature faster and more efficiently. The way we're kind of integrating the biology and chemistry is so much more massive and informative than the way it's been in the past. And then we start thinking about even, like, new experiments – even in my own lab – we start to thinking about what do we want to do next? We have to also think about how is this gonna be automated and done at scale? How do we think about this from, you know, like, instrumentation and data curation? So now, we're kind of changing how we approach things end-to-end, and it's really wild – but in the best possible way.

Karen: So is this unique to the oncology space or are these AI tools being leveraged across all therapeutic areas?

Danielle: We're looking at this from every possible angle. Like, I work in oncology, neuro and ophtha, and in every one of those disease indications, we're looking at how to utilize AI from end-to-end, just like what we've discussed in oncology. Because we're integrating all of these foundational models with such massive datasets and understanding the interconnectivity of all these signaling pathways in ways that we haven't before, all the structural chemistry for the protein interactions in a way that we hadn't done before. So there's so many things that we're going to understand and discover about the basic biology that we just are going to illuminate different things about biochemistry that we just haven't been able to approach before. What I really think is gonna happen is that we're all gonna go through these massive changes and while it might be kind of, like, awkward as we're getting our footing, I feel like we're just starting off running with it, and what we're all excited about is how this is gonna translate to patients getting medications faster, us figuring out how to approach unmet needs that we haven't been able to approach before. This is an exciting time.

And that's our show! Thanks so much for listening. If you haven't already, rate our podcast wherever you listen. It'll help new people find us. And make sure to subscribe! If you have any questions about the show, you can contact us at podcast@gene.com. And now for me, it's back to stalking cells!

The name Two Scientists Walk Into A Bar is under license and used with permission from the Fleet Science Center

Hearing a doctor say, “you have breast cancer,” can be terrifying. The months that follow – filled with scans, treatments, and often surgery – can feel physically and emotionally exhausting.

But there is good news: most people diagnosed with early-stage breast cancer will go into remission (meaning there is no cancer detectable in the body) after their initial treatment.^2,3

According to the American Cancer Society, stage I-III hormone receptor-positive (HR+), human epidermal growth factor receptor 2-negative (HER2-) breast cancer treatment usually begins with surgery, followed by radiation therapy. Depending on the cancer’s stage, whether it has spread and certain risk factors, treatment may also include chemotherapy, endocrine therapy (sometimes called hormone therapy) and/or targeted therapies.^4,5

Want to learn more about types of cancer treatment?

Breast Cancer Treatment Basics

What Comes After Initial Breast Cancer Treatment?

After initial treatment, many people go into remission.³ Even then, doctors often recommend ongoing treatment known as adjuvant therapy to help prevent the cancer from returning. Adjuvant therapies are given after primary treatments to lower the chance of cancer coming back, because there still may be some cancer cells remaining in the body that can't be seen in scans.⁶

It can feel overwhelming to think about what’s next, especially when that means more treatments and doctor visits.

“The initial treatment phase is often intense and can be overwhelming,” said Rita Lusen, VP, Digital Media Banner Program, Breastcancer.org and breast cancer survivor. "Patients may feel decision fatigue by the time they are determining their adjuvant or ‘maintenance’ treatment, so it's important that they seek the support they need to determine the best path forward for this next stage, which may last several years or up to as long as 10 years."

Why Do I Need Endocrine Therapy After Surgery?

Patients often ask: If the surgery worked, why do I still need medication? Your doctor may recommend treatment, such as endocrine therapy, after surgery to lower the chance of the cancer coming back (also called recurrence).⁶

Most breast cancers (70%) are classified as HR+ and HER2-negative.⁸ This means the cancer cells have certain receptors that the hormones estrogen or progesterone (or both) bind to, fueling the cancer’s growth. After surgery for HR+, HER2- breast cancer, endocrine therapy may be given to help reduce the risk of the cancer coming back.^9,10

How Do Endocrine Therapies Work?

Endocrine therapies for estrogen receptor-positive (ER+), HER2- breast cancer work in two main ways:¹⁰

1. Blocking estrogen’s effects
   Selective estrogen receptor modulators block estrogen from helping the cancer cells grow.
   Another type of estrogen-blocking treatment is called a selective estrogen receptor degrader (SERD). SERDs are designed to block estrogen and break down the estrogen receptor. They are already used in advanced or metastatic ER+, HER2- breast cancer and are being studied in early-stage disease.
2. Lowering estrogen levels
   Treatments to lower estrogen levels to such low levels that it slows the cancer’s growth include aromatase inhibitors (AIs).

Glossary

Selective Estrogen Receptor Modulators (SERMs)
SERMs are medications that block estrogen from attaching to estrogen receptors on breast cancer cells, helping to prevent the cancer from growing.¹⁰

Selective Estrogen Receptor Degraders (SERDs)
SERDs are medications that block estrogen from attaching to estrogen receptors. They also break down the estrogen receptors, preventing the estrogen receptors from signaling the cancer cell to grow.¹⁰

Aromatase Inhibitors (AIs)
Aromatase inhibitors are medications that stop most estrogen production in the body, particularly after menopause.¹⁰

Ovarian Suppression
Ovarian suppression is a treatment approach that reduces estrogen production by stopping the ovaries from releasing hormones in pre-menopausal women. This can be done with medication or surgery and is typically given with an endocrine therapy.¹⁰

Why Does Staying on Endocrine Therapy Matter?

These therapies are important to help prevent the cancer cells from growing again.¹⁰ Across all patients with early-stage HR+, HER2- breast cancer, about 85-90% will not have their cancer come back within five years after completing surgery and endocrine therapy, depending on tumor size, lymph node involvement, and other biological features.¹¹

However, the full benefit of endocrine therapy depends on staying on treatment as prescribed.⁷

Ricki Fairley, co-founder and CEO of TOUCH, The Black Breast Cancer Alliance (@touchbbca), adds: "It’s important that we support patients in understanding the value of taking endocrine therapy medicines, how to manage side effects, and their care options if side effects are impacting their ability to stay on treatment."

What are the Pros and Cons of Traditional Endocrine Therapies?

According to Breastcancer.org, about 25% of women who are prescribed endocrine therapy to lower the risk of their cancer coming back after surgery either don’t start taking the medicine or stop taking it early, in many cases because of side effects.¹² Potential side effects include hot flashes, vaginal dryness, menstrual cycle changes, headaches, nausea, feeling tired, loss of appetite, and pain in the muscles, bones and joints.¹⁰ Additionally, some patients stop taking their medicine because their cancer continues to grow despite treatment. This happens because some HR+ breast cancer cells adapt over time, and these changes can make some standard endocrine therapies less effective.¹³

Disparities in care can also impact treatments’ effectiveness. Ricki Fairley adds: "Recent data shows that the breast cancer disparity is the absolute widest in HR+, HER2- breast cancer, where Black women have a higher risk of breast cancer death compared to white women despite available treatments. Given this reality, it's more important than ever that we work to develop endocrine therapies that include clinical results on Black women."

How Can You Balance Treatment With Everyday Life?

Preventing the cancer from coming back is critical, but as treatment is often long-term, it is important to select an option that helps you to feel well enough to live your daily life. If side effects are interfering with your ability to stay on treatment, it’s important to talk with your healthcare team.

HR+ breast cancer research continues to evolve. Ongoing research is focused on better understanding changes in cancer cells and developing new endocrine therapies. The goal is to help address some of the limitations of traditional endocrine therapies, such as treatment resistance and challenges with taking treatment as prescribed.

At Genentech, we are committed to advancing science and supporting people with HR+ breast cancer, so they can live full and healthy lives.

Third-party organizations and websites are not owned, controlled, or endorsed by Genentech.

Our Advancing Inclusive Research® Site Alliance is proving what happens when clinical research is built for the people it serves.

For decades, clinical research has had a critical blind spot, the patients enrolled in clinical studies often do not reflect all people who may one day need the medicines developed to treat these diseases. As of 2023, only 6% of FDA trials reflect the racial and ethnic makeup of the U.S. population.¹ This matters because people’s health is shaped by many factors including access to care, treatment experiences and the realities of daily life and our research must account for those differences.

Nearly 10 years ago in 2017, Genentech decided to change that. We formed the External Council (EC) for Advancing Inclusive Research, a multidisciplinary group of scientific, clinical and community experts, to help us rethink how clinical research could better reflect the populations most affected by disease. Together, the EC challenged us to move beyond individual studies and build a more systematic approach to advancing inclusive research (AIR).

One of its earliest recommendations was simple and ambitious. If we wanted research to better reflect the communities most affected by disease, we needed to work directly with the sites serving those communities. That idea became the Advancing Inclusive Research® Site Alliance, launching in 2021 and continuing to translate site insights into practice.

Five years in, we’re proving what’s possible. Today, AIR Site Alliance oncology sites enroll underrepresented patient populations at 63% higher rates than peers in the same studies.² And they do it twice as fast.³ The Site Alliance is an active blueprint for how to build trust, remove barriers to research participation, and optimize medicines for everyone who needs them.

At times, study eligibility criteria has unintentionally excluded some communities from participating in clinical studies. For example, when Dr. Gregory Vidal at West Cancer Center and Research Institute, one of our Site Alliance sites, noticed his Black patients could not enroll in a breast cancer trial due to their higher blood sugar markers than allowed.

This triggered a system-wide transformation. Working with Dr. Vidal we re-evaluated study entry criteria with Alliance sites, resulting in more inclusive entry criteria that are now applied across our studies, helping more people participate in research.⁴

The AIR Site Alliance is challenging the long-held belief that enrolling underrepresented populations is inherently difficult or slow. Across our studies, Alliance sites have enrolled underrepresented patient populations more than twice as fast as comparable sites, demonstrating that representative enrollment can go hand in hand with enrollment speed.

While Site Alliance ophthalmology sites make up just 2% of total U.S. enrolling sites, they contributed 17% of Black and 12% of Latino patients across 23 vision studies.⁵ In one diabetes-related study, just three Site Alliance sites enrolled 57% of all Black participants, and the study enrolled a full month ahead of schedule, proving that this approach works.⁶ By enrolling more representative study populations, accelerating enrollment and expanding participation globally we gathered better evidence.

These results are driven by models like Montefiore Einstein Comprehensive Cancer Center's BRAID (Bridging Research, Accurate Information and Dialogue) site, which uses community “Conversation Circles” to convene community leaders, clinicians and scientists to co-create culturally relevant patient materials.⁷ This evidence-based approach helps ensure health information is accurate and delivered by voices the community knows and trusts.

Distance, transportation and cost remain participation barriers, especially for rural and other underrepresented communities. To help close these gaps, some sites in the Site Alliance leverage Mobile Research Units, state-of-the-art clinics on wheels designed to meet patients where they live and work.⁸ We're also sharing many of these learnings more broadly through our publicly available educational materials, which provide practical guidance to help researchers, healthcare providers and patients advance more inclusive clinical research.

Rather than requiring patients to travel to distant medical hubs, these units bring free screenings and routine care directly to neighborhoods. By meeting immediate wellness needs first, our teams build a seamless and approachable bridge to clinical research education, screening and enrollment for patients who might never have known these opportunities existed.

For example, unlike traditional health systems, Pinnacle South Texas leverages relationships with local promotoras and community engagement staff to provide free screenings in non-clinic locations, like the Mexican consulate and agricultural fields, directly reaching local Hispanic families and agricultural workers. For example, the Wagner Kapoor Institute has used its "Vision Van" across bridges to reach patients in new areas.

AIR Site Alliance sites work in collaboration rather than in isolation, leading to sharing, adoption and scaling across sites. Together, these sites created an Alliance community outreach toolkit providing guidance to sites to dismantle barriers to ophthalmology care. By sharing proven models across sites, the AIR Site Alliance enables successful innovations to spread more quickly and reach more communities.

Enrolling a patient in clinical research is just the beginning of the work; keeping them there requires patient-focused support to overcome practical barriers. Some Site Alliance sites, like Mays Cancer Center at UT Health San Antonio, integrate non-medical patient needs assessments directly into electronic health records, allowing providers to proactively screen for and help solve hurdles related to food, clothing, and transportation.

Even though we have accomplished a lot, many patients still face stark realities, like a patient in California who used their last dollar on transportation to get to their study site, leaving them unable to afford a meal that day. To address the true out-of-pocket costs of participating in clinical research, the Site Alliance developed and embedded an innovative cost calculator into U.S. study budgets. This cost calculator is now used by the majority of study teams to ethically reimburse patients and, when appropriate, their care partners, helping ensure that participation in clinical research doesn't result in personal financial hardship.⁹

Traditionally, it is not uncommon for clinical research sites to compete for resources and participants. The Site Alliance flips this script. Our sites trade operational and programmatic successes across the network, even if ideas go beyond clinical study recruitment into addressing social determinants of health.

For instance, a community garden connecting patients to healthy food in Boston grew out of Boston Medical Center's broader "food is medicine" approach, which integrates fresh produce and nutrition support into patient care.¹⁰ It directly inspired a similar project at a partner site in Florida, showing how successful ideas can be adapted to meet the needs of different communities. 

Our AIR Site Alliance acts as a brain trust, helping us gather early feedback on research plans and amplify local learnings across all sites. By converting these insights into actionable development decisions, we can ensure every site is equipped with the best tools to build trust and eliminate barriers.

Five years ago, the industry viewed inclusive research as too difficult, and too slow. Today, the Site Alliance challenged those assumptions, demonstrating that broader participation can accelerate enrollment, strengthen evidence, and build trust with the communities clinical research is meant to serve.

Our work is far from finished. The Site Alliance provides the industry with a scalable blueprint for a future where inclusive research is the standard, and life-changing medicine belongs to and is available for everyone.

My pulse quickens as my eyes scan over the PET images, focusing on the dark patches where none should exist.

Anxiety rises as I brace myself for the conversation: how do I tell the patient and their family?

I walk into the room. My usual cheerful demeanor is absent – they can already sense something is off. I apologize, inform them of the terrible news, and let them process.

I immediately feel empathetic. My mind turns to how much their lymphoma may impact their day to day. What will they do with work? The upcoming graduation? Wedding planning? That cruise they just booked? In that moment, lifetimes flash by me.

My mind jolts back into “doctor mode.” Answers whiz across my mind. I’m searching for a solution – striving and hoping for whatever treatment will allow them to reach these milestones.

Together, we form a treatment decision.

This is the sobering reality of the moment a patient learns their blood cancer has relapsed – a moment defined by uncertainty, and the urgency that comes with needing a new treatment when options may be limited.

For patients with relapsed or refractory blood cancer, the return of disease doesn't arrive with a clear roadmap. Relapse often brings uncertainty, new decisions, and a different set of challenges than the first diagnosis. Even with advances in treatment, approximately 20-30% of patients with blood cancers like diffuse large B cell lymphoma (DLBCL) relapse or are refractory after first-line therapy, and survival outcomes decline significantly with each subsequent line of treatment.^1,2 When that happens, the path forward isn’t always straightforward – and the impact goes far beyond the disease itself.

What does relapse mean for blood cancer patients?

When a blood cancer relapses, it means the disease has returned. While some slow growing lymphomas may not need immediate treatment, the faster growing subtypes including DLBCL need immediate treatment. Many patients will require additional lines of therapy as their disease progresses. When this happens in lymphoma, care decisions often become more complex, and options may become more limited.

For patients whose disease does not respond to initial therapy (aka refractory disease), prognosis is particularly poor, with historical data showing median overall survival measured in months in the relapsed or refractory setting,² underscoring one of the highest unmet-need populations in lymphoma. At the same time, many currently available regimens rely on intensive chemotherapy or inpatient treatment, which can be physically demanding and disruptive to patients’ daily lives, often requiring time away from home, work, and family during an already difficult phase of care.

While advances in care have expanded what’s possible, each treatment approach can come with different considerations that may affect if – and how – patients are able to receive them.

What treatment options exist for relapsed or refractory blood cancer?

There are several types of treatments used for relapsed or refractory blood cancers. While they work in different ways, each comes with its own benefits and challenges.

Chemoimmunotherapy:

This approach uses a combination of drugs – some that kill rapidly dividing cells directly, and others that help the immune system find and attack the cancer.³

• Potential benefit: Chemoimmunotherapy is widely used and can be used at different stages of treatment.
• Considerations: Side effects can build over time, making it harder for some patients to tolerate treatment.

Targeted therapies:

These treatments focus on targeting specific signals that the cancer cells rely on to help them grow.⁴

• Potential benefit: Targeted therapies can be more precise than chemotherapy. Some are designed to be used as continuous, long-term treatment, while others are designed to be used over a fixed period of time, to help keep the cancer from progressing.
• Considerations: This refers to a broad group of treatments, including antibody-based therapies (such as monoclonal antibodies, bispecific antibodies, and antibody-drug conjugates), as well as other therapeutic approaches. Each works in a different way and has its own safety profile, so treatment decisions depend on individual patient factors and where they are in their care.

CAR T-cell therapy:

This treatment involves collecting a patient’s immune cells, training them to recognize cancer, and returning them to the body to fight it.⁵

• Potential benefit: CAR T-cell (CAR-T) therapy can provide a deep, lasting response for some patients, offering a treatment option that does not require continuous, ongoing dosing.
• Considerations: It is a complex process that may require travel, coordination, and time in specialized care settings.

For many patients, choosing a treatment is not just about what works medically – it's also about what they can realistically manage and what they have access to, especially when time matters. Each relapse can narrow treatment options and shorten the window to act. Time in the hospital, travel requirements, side effects and time away from work or family all play a role in treatment decisions. While current therapies hold promise, innovation in relapsed or refractory blood cancer remains an area of significant unmet need for highly effective, well-tolerated treatment options that allow patients to spend less time in the hospital and more time living their life outside of it.

What challenges do relapsed or refractory blood cancer patients face when options become limited?

When treatment options narrow, the challenges patients face often extend into every part of their lives.

Managing care can become complicated

Patients may need to see multiple specialists, travel to different treatment centers, or wait anxiously between appointments and decisions. Coordinating care over the long run requires meticulously managing calendars while feeling entirely unlike themselves. The workup itself can be a marathon; even an optimally scheduled day could involve hours of back-to-back appointments with medical staff poking and prodding a weary body. This means long days, time away from home, and immense stress.

The emotional weight can grow heavier

Relapse is fundamentally different from an initial diagnosis. Patients have already endured treatment once, and they may now face more uncertainty with fewer clear answers. The emotional toll of “starting over” can feel overwhelming, and it can be particularly depleting when the last treatment was emotionally, logistically or medically challenging.

Barriers to access for blood cancer patients

For patients treated outside large academic or specialty centers, access itself becomes a barrier. Therapies that require inpatient care, specialized monitoring, or referral to distant sites can narrow options during an already urgent phase. Practicing in Manhattan, I see this firsthand. Practically every patient must travel over an hour to see me, but many commute much further – driving three to eight hours from upstate New York or deep within Pennsylvania. By the time they finally sit in the exam room, I can see their sheer fatigue from hours of navigating tunnels, bridges, potholes and traffic lights.

Caregivers feel the strain, too

Family members and caregivers often step back into intensive support roles – helping with appointments, travel, and daily needs. Over time, this can create added emotional and physical strain for everyone involved. I frequently see the extremes: some patients fiercely shield their family members, wanting to carry the heavy burden alone to protect them, while other families respond with an almost overwhelming, all-hands-on-deck level of involvement. Though these approaches look completely different, they are both deeply loving in their own ways. Underneath the logistics and the questions, I can see the raw grief, the profound love, and the desperate worry driving them. In those heavy moments, sitting across from families trying to process the unimaginable, I often find myself just wishing I could hand them an "easy button" – some magical way to instantly ease their pain and make the road ahead simple.

“In my role, I bear witness to the profound heaviness that follows a relapse diagnosis. But as deeply as I feel that weight with my patients, it is always eclipsed by my awe for them. It takes unimaginable courage to face down a returning cancer and still say, 'let's try again.' We mourn the setback together, and then we look to the horizon. Those exact moments are what shape my work at Genentech. When patients are brave enough to endure another round of treatment, we owe it to them to develop therapies that don't just fight the cancer, but actively remove the logistical and physical stress from their lives.” – Connie Lee Batlevi, Senior Medical Director, Genentech

How can we better support relapsed or refractory blood cancer patients?

For patients living with relapsed or refractory blood cancer, better support isn’t just about new treatments – it’s about improving the overall experience of care.

For some, it could mean receiving treatment closer to home, without the need for long-distance travel or extended hospital stays. For others, it may mean clearer next steps, more coordinated care, and less time spent waiting between decisions.

Better support also includes treatment options that are not only effective, but easier to tolerate and fit more naturally into daily life. Managing side effects, maintaining routines, and staying connected to family and work can make a meaningful difference.

Just as important, patients want to feel heard. When patient experiences are consistently reflected in how care is designed and delivered, progress becomes more than clinical – it becomes meaningful in the context of real life.

Moving forward

Progress in blood cancer research continues to open new possibilities. But improving outcomes also means improving how patients experience care every day.

Patients are not only navigating treatment. They are balancing work, family, and the realities of life that continue even when cancer returns. Supporting them means truly listening to their needs and making care more accessible, coordinated, and less disruptive.

Genentech, alongside the blood cancer and lymphoma community, is working to push science forward, improve treatment options earlier in the disease journey, and support more equitable access, recognizing that delays and limitations in care can have real consequences for patients facing relapse. There is more to be done, and continued commitment is needed at every level of care. The goal is clear: to move toward a future where blood cancer care is not only effective, but easier to live with.

Setting a marathon record, flying through the air with nothing but silks to support you, drumming in front of a packed arena, or sharing your life with millions of TikTok followers would be an impressive (maybe even daunting) feat for anyone. But imagine doing so with blurred vision, tingling feet or brain fog. That’s what makes Elizabeth, Cecelia, Rob and Demi’s stories so extraordinary. All four are exploring their passions while also living with the daily and unpredictable challenges of multiple sclerosis (MS).

Yet, even as their disease has progressed, they’ve found new ways to adapt, rebuild and thrive. Now, they are sharing how their lives have been uniquely shaped by MS, and how they have found purpose and resilience through their talents and passions – or hidden geMS. Their personal stories illustrate the importance of advocating for their health and navigating MS disease progression with a healthcare provider.

Meet Elizabeth

Elizabeth is a Massachusetts-based reporter and 17-time marathoner living with relapsing-remitting MS.

Elizabeth was four days away from running the Berlin marathon when she was diagnosed with MS. With her doctor’s encouragement and her now-husband running beside her, Elizabeth found a way to not only cross finish line, but to continue to make distance running part of her routine.

Her dedication ultimately led Elizabeth to a Guinness World Record for the fastest woman with MS to complete a marathon. Though her journey hasn’t always been easy, she embraces the unique position of being an elite athlete living with MS. An important part of her training is working to find the balance between pushing herself and listening to when her body needs rest and support from others. Discover how Elizabeth keeps pace with MS and what’s she learned along the way about the mind-body connection:

Meet Cecelia

Cecelia is a New York-based aerial artist living with relapsing-remitting MS.

Through aerial arts, Cecelia has found a way to connect to her body with movement she loves – a practice that emphasizes patience, prioritizing rest and embracing modifications. Although MS can present challenges for her on the ground and in the silks, Cecelia is frequently checking in with her body, adapting her movement and recalibrating when something isn’t within her ability that day.

Cecelia’s story is one of many examples of overcoming the daily challenges of living with MS and the unique beauty of finding strength through peace. Dive into Cecelia’s story:

Meet Rob

Rob is a musician and content creator from Tennessee, living with relapsing-remitting MS.

Rob feared losing his ability to drum when MS put his music career on hold. But he's since learned that just like in music, these moments of pause can give way to what really makes us shine.

After finding meditation helpful for managing the stressful moments following his diagnosis, he realized he could use his talents to create his own meditation music. Through his music and the love he has for his furry companions (Doug the Pug and Dory), Rob embraces life at a new tempo. One that now emphasizes the value in slowing down and working through the moments of stillness.

Dive into Rob’s story to learn more about his return to music, unbreakable spirit, and how in the face of MS progression, he’s redefining what it means to live the rockstar lifestyle:

Meet Demi

Demi is a mother of two and a social media creator based in Pennsylvania living with relapsing-remitting MS.

Alongside her husband and their two young kids, Demi has built a vibrant social media presence where authenticity thrives. What started as a playful TikTok account filled with dances and pranks evolved into a support network as Demi navigated the ups and downs of parenthood and her MS journey.

For Demi, it’s all about controlling what she can, whether it's her attitude, her actions or how she chooses to live. She listens to her body, knowing that some days demand rest and others allow for a faster pace.

Learn more about Demi’s story and her joyful, intentional approach to each day:

For Elizabeth, Cecelia, Rob and Demi, life didn’t stop after an MS diagnosis. Their stories are powerful reminders of the importance of proactive disease management and discussing treatment options with healthcare providers. For people living with MS, high-efficacy disease-modifying therapies (DMTs) can play a crucial role in slowing disease progression, helping them continue doing the things they love.

For more tips and information on how to discuss treatments with your doctor, see Navigating Multiple Sclerosis Disease Progression With Your Healthcare Provider.

Are we on the cusp of a new era in scientific research? This episode delves into the transformative potential of the "lab of the future." Host Danielle Mandikian and guests Margaret Porter Scott, VP, Biochemical Cellular Pharmacology, and Andy Lash, Head of Engineering, Lab Automation, discuss how AI and automation are set to tackle major bottlenecks in science, from fragmented systems to the physical and time-intensive toll of repetitive lab work. Discover how these advancements will lower the activation energy for groundbreaking research and empower scientists to better tackle some of humanity’s most pressing medical needs.

If you would prefer to read a transcript of this episode, please click here.

SUBSCRIBE BELOW TO CATCH EACH EPISODE

If you want to learn more about the groundbreaking science happening in our labs, click here. To learn more about the jobs in our research and early development group, click here.

Transcript of Two Scientists Walk Into A Bar: “Lab of the Future” with Margaret Porter Scott and Andy Lash

Maria: I’m Maria Wilson.

Danielle: And I’m Danielle Mandikian.

Maria: And we are scientists. We. Love. Science.

Danielle: Yeah, we do. So, when we aren’t doing it, the next best thing is to talk about science! And what’s really awesome is that we’re surrounded by some of the most brilliant minds in research!

Maria: We are going to step away from the labs today to talk to other scientists about the cool stuff they are thinking about, working on and imagining . . .

Danielle: . . . as well as how some of these discoveries just might lead to new medicines. So, grab your favorite drink, get ready to unlock your science brain and join us for Two Scientists Walk into a Bar…

Maria: The show for scientists, science geeks, and the people who love them!

Danielle: What comes to mind when I say lab of the future?

Employee responses:

Lab of the future. I think it will be a more fun and interesting place to experiment. I think ideally for experiments that have really well-established protocols, you know, you can click a button and automation will do it for you, and then we can focus less on these manual and routine walk-throughs and more on thinking about and executing on unique experiments that only we are prepped to do versus the machines.

Shiny.

More automation.

I want RFID benches to track where compounds are.

You can watch a fully grown system from tiny cell all the way up to medicine.

Fully automatic.

You know, people have really big ideas and most of them we cannot do because you need an army of, like, scientists to do all of that. And, you know, at least the hypothesis is a bit like automating a lot of, like, the hard parts. Maybe we are enabled to do, like, you know, combinatorial screens. We'll see. It’s hard work to get there.

The lab in the future will allow us to get to scales that we cannot do right now, and that means not just the throughput, but less mistakes, less errors that are introduced, more standardization, and that will help us unlock a next step.

Danielle: Hi, everyone. Welcome to Two Scientists Walk into a Bar. I'm your host, Danielle Mandikian. And I got to say, I am over the moon about today's episode. Because last season, we asked everyone about AI, and most of the perspectives were on data science. But at the end of the season, we had another guest, Aviv, who went further and gave us a bit of a teaser about how things are actually gonna change in the lab. So we decided for this season to dedicate a whole episode on what the lab of the future could actually look like. Now, as an often burnout lab scientist myself, I'm absolutely ready for this. So in order to bring us all into the lab of the future, we've invited two special experts, Margaret Porter Scott and Andy Lash. Welcome to the bar!

Margaret: Nice to be here.

Andy: Hello. Yeah.

Danielle: Let's get into it. So out the gate, lab of the future, what is the, like, visual or song that you hear in your head?

Margaret: I'm a Futurama girl, so I see, like, the guy walking him out of the cryo-chamber. [Laughs]

Andy: Yeah. I think it definitely has that, like, sort of 1950s sort of, you know, what the future will be. Sort of, Jetsons-esque.

Margaret: Lost in Space, like, yeah.

Andy: Yeah. Yeah.

Danielle: Okay. So, alright. So from y’all’s perspectives, thinking about lab in the future, what do you see as some of the bottlenecks that we absolutely need to just deal with?

Margaret: I would say a lot of it is around fragmentation. So today, while, you know, scientists exist, and a drug-discovery process that is integrated across a lot of labs and a lot of disciplines – we use a lot of separate software, a lot of separate automation systems that today are not very uniformly connected to one another. And our future lab, I think, will have fully integrated and connected processes where things will flow a lot more smoothly.

Andy: Yeah so like, when we think about lab automation in a pharma setting – so first, there's the whole world of industrial automation like we do in manufacturing, and that world is totally different, right? You wanna make sure that every dose of XYZ drug is precisely the same as every other dose. So, you know, those systems, the way they're programmed, they're kind of designed to be inflexible by their nature, and a change to that line is a really big deal – and that's not how it is for the stuff we do, right? So we're really talking about lab automation in a research context, and in a research context, you need flexibility. And there's a whole cottage industry of, you know, tooling to be able to do this, and software that lets you kind of reconfigure this stuff. And I think for me as someone with, like, a more traditional software engineering background, not like a coming at it from a life sciences background, the way this software works is, like, quirky and weird. The way you program these systems has – I think it's – I think they're intended to be easier to program, but I think the kind of reality of the way they approach this makes it, like, very different than how you program other things, and because of that, the like, easier-to-program stuff actually ends up being this hyper-specialized skill and making changes is slow, and I just generally think that that sort of lab automation software being this, like, special, quirky thing is ultimately going to go away, and it'll be just much more like how we approach any software engineering things. And, you know, a lot of the sort of software engineering best practices like automated tests or, you know, version control, which are like broadly not used in lab automation, will be. And so then once that happens, I think that'll make it, like, a lot more accessible to folks.

Margaret: Yeah. I know one of the things I've been reflecting on a lot is how I'm really excited about the way AI can enter the scene at a lot of those interfaces and make them a lot smoother and faster for us. And what I think this will mean is that automation will maybe finally reach its potential. Because I think when we've thought about automation in the past, we've thought of, like, fully autonomous processes, and we've sort of, you know, have seen these, you know, things in action and assumed that they were working really smoothly – but a lot of times, there are people on the back end trying to find bespoke solutions, there are lots of little disconnects and a lot of times scientists are doing things that are really not all that intellectually stimulating or interesting or amazing, but just kind kind of, like, kludgy copy-pasting, you know, making connections or waiting for vendors, those kind of things. And so, you know, my belief – and I think what, you know, a lot of people are gonna be working really hard on – is to get AI systems that can just fill those little gaps.

Danielle: I mean, that's really, like, a thing, too. And sometimes we don't talk about the tax of this, but, like, when you're the person at the bench, there's, like, a mental cost from, like – and I don't know if this is just the way my brain functions or the people I know, but – you know, you have to, like, switch topics, and you get really into it, and then having to stop and stall and do some of those kind of like clerical things, really trying to do this communication stuff really interrupts the creative process. It's –

Margaret: That's a great point.

Danielle: – it's a really big problem sometimes.

Margaret: Yeah.

Danielle: Yeah.

Margaret: And I've reflected on that. Like, it's probably not what you and others went to school for, was to, like, you know, fix a clog. You know? [Laughs]

Danielle: Yeah. Yeah.

Margaret: To download data onto a –

Danielle: Or – yeah. Or why does – why can't I get the math to math – right? – on certain things. And you think that you transfer it to one system, it should be easy. So that's actually something I wanted to ask y’all about. Like, what are things you think might help this kind of integration between two worlds?

Andy: Yeah. I think – you know, I think the challenge there is when you have these different backgrounds, you need to develop some trust because trust is this sort of magic currency that makes great teams great. And so, you know, I think, like, a piece of that is being able to, you know, on the software side, when we're building stuff with scientists, that we can bring them along, produce stuff that's useful. But then also, you know, it's sort of trying to solve some of our broader goals in the area. You know, I know – you know, for – I think for a lot of times, we, when we're building a system that maybe has, like, a UI component that a scientist needs to interact with, you know, in like the back of my mind, the real goal is get good, clean data to do data science on, like, later. But the – in order to get some, you often need to kind of get data out of people's heads. And since we can't yet do that in a magical way – right? – you know, so if we have to get them to enter stuff. And so then the – I think the trick is, how can you make the software easy enough to use, or not just be easy to use but also give something back to the scientist, like, save them time in some way so it's not just the, like, enter metadata text, it's like, "Oh. This is helping me, it's providing a calendar of what I need to do," or it tees up the next step, or, like, however that system solves the problem, you know, that's like a big piece of it.

Margaret: And I think on the side of the experimental scientists, I think we need to be able to abstract one level from what we're doing. So rather than say, you know, we need to measure, you know, this antibody for this project, and this solution for this reason, you know, using these reagents, we need to say something like, "We need to move liquid from, you know, these tubes into these tubes, dilute it in these ways." And we also need to be able to bring that into the paradigm of other things that are similar.

Danielle: How do you all see that future looking like for, like, bench scientists and also from, like, the computational side?

Andy: So, you know, unless you've been living under a rock, you're aware that AI has, like, grown by leaps and bounds over the last several years. And also, just in general, like, if you've played with ChatGPT when it came out and then haven't really touched it in a while, all of your opinions about it are so wildly out of date. It's really changed a lot in the last year, and one of the things that it has gotten much better at is writing code for you. And I would say in something like November of the past year, it got generationally better again, and I can now see that a lot of the software engineering we'll be doing going forward will really be, like, written by AI. So that's crazy. And so, like, then when I think about this, you know, so what's that going to mean for automation? Because, like, automation, or the lab automation like I was describing before, is like – it just has this, like, weird environment, but if it was more like software engineering, then you could imagine getting these, like, potentially huge gains from, you know, the same AI-assisted coding stuff happening here. There's a ton of lab automation that doesn't happen today, because some scientist rightly concludes that, like, the effort in terms of setting the whole thing up is, like, not gonna save them time because they're not doing a repetitive process, or it's – they're only repeating it once, or whatever it is. So if we, like, greatly lower the activation energy, then I mostly just think we'll see lots more of it.

Danielle: I appreciate that you put that in a biochemist's framework – lowering the activation energy. I just want you to know, I picked up what you put down.

Margaret: The delta G. [Laughs]

Danielle: [Laughs] Yes. So, like, so just to make sure I'm, like, getting this, like, you see, like, how even for, like, people in the software engineering space that AI is gonna be helping them shift over, like, some of the, just, nitty gritty of the coding by assisting them, and that might help them envision how to put everything together in a more – I don't even know the right word for it – but, like, way to make it easier to apply to multiple systems or something.

Andy: When you're using these tools and when they're working well, you're really operating on a much higher level. You're not – you're really describing much more what you want and the kind of broader software architecture of how you want it to be in like, in the framework, but you're not saying, "Write the code this way." I mean, it's really different. It's not – again, if you, like, played with the previous generation of tools and it was sort of like auto-complete, and you, like, want it to write this function and it, like, writes a function for you. It's like, it's not that anymore. Now it's – you write these, like, long, detailed prompts, and then it goes off for 20 minutes and writes, like, a large amount of code.

Danielle: It frees you up to be a conductor.

Andy: Yeah. It frees you up to be a conductor, is the – you know, and I'm like, I like, my instinct is to try to, like, insert more tentative language here because, you know, I just feel like every day, we're learning a little bit more.

Danielle: Yeah.

Andy: And it's not – these tools are, they can be still be, like, drive you crazy. They get stuck in ruts, but also, like, oh my gosh, just the last less than half a year, the changes have been crazy.

Margaret: Mhm.

Andy: So I think we'll see a lot of that, and I think that that will start to have a significant effect on lab automation, if we can get there.

Danielle: How do you see this, like, changing the role of folks in your world?

Margaret: Yeah. Well, I mean, I think one thing it may do is really allow them to increase their productivity in a very satisfying way. I mean, if you think about what it takes to make a medicine, you know, it's the confluence of several miracles. You know, you have to be potent and selective and safe. You have to be, you know, on a correct therapeutic hypothesis which you, you know, may not know until you get to the clinic. So there's so many things that we have to do and test to make medicine, that if we can improve our productivity, we could, you know, make more medicine for more unmet need. And we're not out of ideas or nor is the world out of unmet medical need. So in, what that might look like in a tangible sense is rather than if we use AI for what I think it might be good for – so for example, to program a robot – right now, we need specialists for some automation to program it. And if you wanna do something different tomorrow than what you did today, you need to call on that specialist again. So, I mean, what's AI good at? It's pretty good at translating one thing into another. So it can translate a human protocol into an automation protocol. I mean, we can do that, you know, already today in some ways, and I think, you know, this is what's going to, you know, as Andy’s mentioning, it's – this is gonna get more and more sophisticated and really speed up, so now, you know, our scientists don't have to spend as much time on programming and planning their automation runs. They also, you know, in terms of errors, people are spending a lot of time babysitting –

Danielle: Oh yeah.

Margaret: – automation. And, you know, waiting to see is a plate gonna be dropped? Is there gonna be a clog? Is there gonna be a bubble? These kind of things. And so if AI can do what I think it can do really well in terms of pattern recognition, and, you know, detect these things, maybe even eventually correct them without somebody having to babysit – now, all of a sudden, maybe instead of doing one project, you can do two projects. And, you know, because of what we know, because we know there is such a great need for additional medicines for unmet medical need, this is a boon for our scientists and ourselves to be able to just deliver more into the world.

Karen: Hi Danielle!

Danielle: Hey Karen!

Wellington: Hey Danielle!

Danielle: Hey Wellington!

Karen: So tell me, what is AI really good at now at the lab bench?

Danielle: So even though we've been talking about the future, if we just focus for now, I would say AI is really great at helping us dig through literature, and the other kind of thing that I think has been really helpful is helping to give different perspectives on how we might approach something.

Wellington: You brought up the analogy of it freeing you up to be a conductor. What specifically does that mean for you at the bench?

Danielle: To me, that means that I'm still in charge of what science is happening, but I'm being able to point to different instruments, if you will, or different tools or assistance, however you wanna frame it. But each of these kind of tasks that I could offload to AI are things that I'm still putting together for an overarching symphony or an overarching scientific goal, but I'm just kind of directing what needs to happen in order to bring it all together for the bigger picture.

Danielle: Okay. So kind of a weird question, but, like, if I wanted to, like, super-duper distill this down, you know, it seems like there's things y’all have to think about from, like, theoretical, hardware and software. Like, how do y’all rank order where your power is going – your brain power is going towards these?

Andy: The hardware outside of the software, it sort of doesn't get you anywhere.

Margaret: Mhm.

Andy: So when you're thinking about automation, you kind of have to do a good job of both. So you know, I guess that's kind of how I'm thinking of it, but, I mean, again – and, you know, maybe this is kind of playing off a little bit of what Margaret was saying, like, you do these things for a reason, right? We don't – we're not interested in just, like, building cool automated processes that are cool –

Danielle: Yeah.

Andy: – right? We want to do something that is, like, gonna drive the science forward in some, you know, meaningful way. And, you know, I mean, that is the really cool part about being in this industry is, like, this idea of like you could actually do something that would actually help people. And also, there's a history of it working. Like, look at cancer in my lifetime –

Danielle: Yeah.

Andy: – how it went from a death sentence in most cases, to not, in many cases. That's like a huge change. So anyway, accelerating that seems like, yeah, a very worthwhile thing to do.

Danielle: I agree with that vibe. [Laughs] What about you, Margaret?

Margaret: Well, I think at the end of the day it's not an either-or.

Danielle: Yeah.

Margaret: It's really, this is a partnership, right? I mean, just like no one's ever made a drug by themselves, I also think that this, you know, kind of creation of the lab of the future is going to be a pretty tight partnership between, you know, people who are good at software and automation, and the scientists who know exactly what they need their experiments to deliver and how they should be done. So, I mean, I think that – but that's actually the joy of the whole thing, is that, you know, being smart with smart people. This is really – [Laughs] – this is really one of the great joys of being a human on planet earth. So I think that's part of the – the fact that it's not one or the other, and they're not sequenced, but that it's a partnership, is probably the best part.

Danielle: In terms of, like, lab automation in the foreseeable future, like, what is the weirdest one that you're most excited about and why?

Andy: You know, right now, the way lab automation works is it's like industrial robots, right? Like a car factory. And everything, it's all based in absolute positioning in 3D space, and so, and, you know, sensors aren't being used. Now, we're now seeing – actually, on the way here, Margaret and I were chatting about her, you know, the latest generation of the most standard robotic arm we use in lab automation has an integrated camera, and it will adjust itself and self-teach positions in 3D space. So that's, like, already, you know, that's already happening. I think on the humanoid robots side, you know, to me, the implication is that they have more sensors, more decision making. Their humanness, I think is like a bit up for debate. For example, like wheels seem better than legs.

Margaret: One hundred percent, yeah. [Laughs]

Andy: But, like –

Danielle: Wheels seem better than what?

Margaret: Legs.

Andy: Legs.

Danielle: Oh yeah. [Laughs]

Andy: But also, I think some of the reason why they want to make robots more like humans is actually to make it easier to make the robots, because then you – so LLMs are, like, amazing and magic and they work because you gave them all the text in the world, right? So where would you get a similarly-sized training set for robots?

Danielle: I see. Okay.

Andy: Okay? So if they're shaped like humans, you can teleoperate them, which means you're in, like, a VR set, a headset, and then you do actions and the robot mimics your actions, and you do that enough over time that it can learn those actions. And so that's a way to get training data. Someone else was telling me there's research right now into can you – can you use, basically, YouTube videos as a way to teach humanoid robots?

Danielle: That was the first thing that came to mind when you were saying that –

Andy: Yeah. It said, like, how to be humans –

Danielle: – yeah.

Andy: – and I don't actually know where that has gone and how effective that is, but, I think that's sort of another reason why, sort of, the humanoid shape is interesting. But, like, for us, I think you could see robots like that to try to move things between disconnected stations, that's one sort of obvious use case. Another case is that some of the things we do happen at volumes that just your, like, traditional lab automation is not good at –

Danielle: Mhm.

Andy: – you know? Your trad–

Danielle: Like tiny volumes?

Andy: Yeah. Traditional lab – no, it's mostly bigger.

Danielle: Oh okay.

Andy: So traditional lab automation is, like, oriented these, like, rectangular plates.

Danielle: Yeah, yeah.

Margaret: The SBS format.

Andy: Yeah. Exactly. But, like, we do some stuff in five liter flasks. Or, I'm not even sure what those bags are called –

Margaret: Wave bags.

Andy: Yeah. Wave bags. They're, like –

Margaret: Forty liters.

Andy: Yeah. So they're, really, they're hard to use, right? [Laughs]

Danielle: More than five? [Laughs]

Andy: You know, and they're, and even just picking one of those up is, like, a classically hard problem for a machine –

Margaret: Right. Right.

Andy: – because the weight moves as you're holding it.

Danielle: Yeah.

Margaret: Mhm.

Andy: But if we could solve that, that would be great. Because otherwise, that's all just sort of – there's a lot of drudge work in maintaining that kind of stuff.

Margaret: Mhm.

Andy: So I think something will happen there. But like when I'm envisioning these robots, it's much more like they are autonomous, or at least autonomous to some degree, right?

Margaret: Mhm.

Andy: And if you just imagine, like, to get a robot to do something, there's, like, many little motors that have to fire.

Danielle: Yeah.

Andy: And so you're, if you’re, like, one version is you explain the like, detailed steps of, like, how to do that, but I think those days are over. I think now it's more like you're explaining the task and it figures out the details.

Margaret: Yeah. Yeah.

Andy: So to what extent that gets used in labs or not, I guess we can talk.

Margaret: So I wanna add to what you were saying, Andy. I think, you know, humans are probably gonna continue to do the job of humans, but what may be interesting is that our labs will – we'll have the opportunity to reconfigure our labs so that they are on the scale that the scientists and the automation needs to really function well. And so as, you know, as scientists are, you know, reviewing data and designing new experiments, they may not actually need to get in there with their own body and hands to do the work, we may be able to optimize the way the labs are configured to run autonomously. This is what we can really be excited about, which is that we can be more productive. We can even explore, you know, some of the lower – the kind of higher risk, higher reward kind of hypotheses that we haven't had the time and resources to do. It may come in scope if we have a lot of automation that was exploring, let's say hypothesis A. Now maybe because we have more time freed up from the excellent automation, we could maybe also explore hypothesis B, and once in a while, those are gonna pay off and those are gonna be the real transformative therapies.

Danielle: I don't think there's a scientist out there that that doesn't hit real for, right? [Laughs] I mean, seriously. But you know, the cool thing is that even though y’all are coming from such different worlds, really what you're talking about is, again, this idea of, like, having all of this extra automation frees up the scientist to explore more –

Margaret: Mhm.

Danielle: – and conduct higher theory projects.

Margaret: Yeah.

Danielle: Which is what we all want.

Karen: Danielle, what's your weird prediction? Are aliens involved?

Danielle: Well, ancient astronaut theorists would suggest that they've already been involved, Karen. [Laughs] I don't know. I feel like my weird prediction would be that instead of us always focusing on the way that we're trained to have a very defined, small, incremental question, we're able to orchestrate much larger questions and get to bigger picture answers much faster. So not that weird, it's kind of like our goal.

Wellington: Describe for us what you do in a day and how you would leverage, you know, this lab of the future to be more creative.

Danielle: So, I'm a wet lab scientist and I really try to understand where therapies distribute in a living organism. What's the biology and chemistry behind that? And, you know, as focused as that might sound, it actually means that the types of techniques and lab-based experiments that I do are actually very diverse. And some of them are kind of dangerous. Like, I was joking with someone, like, I can't wait for the lab of the future to come handle all my liquids because I'm always terrified I'm gonna get osmium on my skin. And like, when I think of the lab of the future and what it would do to take care of things, I could offload some of these really tedious, experimental things where I'm at the bench for hours, manipulating things slowly and incrementally – having a robot do that so I could then think about the biology behind it and where I'd want to go afterwards.

Margaret: I also want to talk directly to the young scientists out there. I think as we're able to get a lot of the mundane jobs in the lab done in the future – the repetitive things, all the silly copy-paste things done by AI and automation – this will really make a lot of space for innovation. We may be able to do higher risk experimentation than we've been able to do in the past because we just have that much more productivity at our fingertips with the same amount of resources.

Andy: Something that comes to my mind is the need for software in this biological context is so high, and we're able – and we've just, just we've been never able to even come close to meeting the need – that the thing that's, like, most appealing to me is if we imagine that we're giving, you know, each software engineer this, like, you know, science fiction power suit that makes them five times stronger, right? What I see is we're going to then be doing five times more, and maybe then we'll catch up to where the actual needs are.

Margaret: Yeah. Yeah. Because the work is waiting, right? I mean, if we look at the things, the needs that we have in informatics and software to support lab operations, I mean, at our current rate, you'd never get it done. So yeah, it'd be great to have an AI assist on that.

Danielle: With this really fast-changing technology, what do you think scientists need to do to change their perspective or their skill sets to really stay relevant and integrate fully into this, like, crazy freeway of change?

Andy: It’s like a whole new world. It seems like so many things are changing. And the good, the opportunity there is there's very few experts because all of this stuff is so new. So, you know, dive in, become an expert. You know, I think if your mode is, like, "These tools don't affect me," then you're probably setting yourself up for a tough future. But if instead you're like, "Oh. These tools, you know, make me better and I'm extremely good at using them," then I just think that – those are the people we'll need in the future.

Margaret: I think it's imagination, right? Because we've been in one paradigm for actually quite a while. You know, I mean, I saw the first automation land in the mid-90s, and things haven't changed that much since then, but I think they're about to change pretty quickly. And so what the challenge before us that I think is great is to be able to think, you know, much more broadly at what, you know, each scientist can do, what we can accomplish, how we can do things in new ways. I mean, there's a real unlock about how we're practicing science in the lab that's, you know, I think about to happen. And so the way to ready for it is to think about, like, you know, what would I do if I could? How could I make a bigger impact? What other ideas, you know, do I have? And I think those are the kinds of things that I would really encourage people to, you know, start imagining to be ready for the moment.

Danielle: Yeah. I mean, so kind of a funny question close to this came from my co-host Maria, and she was telling me that she remembers when molecular biology was a lot like cooking. Like, there were tons of buffer recipes, and they were super finicky –

Margaret: Mhm.

Danielle: – she spent a lot of time wasting, like, weighing things out, pH-ing them perfectly, autoclaving them when necessary. And then she remembers when, like, the DNA prep kit started coming out. Like, precast gels and stuff like that. So she wanted me to ask y’all, in the lab of the future, do you think molecular biologists will still need to know how to pipette, or will they need to develop a totally different skill set? I mean, that's the dream, right? Screw carpal tunnel. [Laughs]

Margaret: [Laughs] Well, yeah. I do think – I mean, in real life, repetitive stress injuries, I think, are something that we could hopefully put in the rear-view mirror for sure. I mean, to be honest with you, I think today, molecular biologists do already do a lot of things with automation. But so they're probably more in the space of getting stuck on, you know, what are those few manual steps – either on the data side or on the automation side – that they need to do. And I think those are the kinds of things that could probably, you know, get evolved even further.

Danielle: I think that's, like, really good news. Like, I've known in my lifetime, I mean, more scientists that I can count on a hand who have actually had their careers ended because they got such severe carpal tunnel that they couldn't do this. And now, like, as things are transitioning, they – it opens them up to be able to do so many more things they wouldn't have been able to do.

Margaret: Yeah.

Danielle: Right? I think that's something that sometimes we ignore, or we don't realize –

Margaret: Yeah.

Danielle: – how prevalent it actually is. Yeah.

Wellington: So Danielle, when you mentioned repetitive stress injuries, I quickly got my phone out and did a little search and it turns out there are studies that show that between scientists and laboratory workers, up to 77% of scientists suffer from work-related injuries.

Danielle: Yeah, I think sometimes when people think about scientists, they think about someone lost in the books, really thinking deeply about stuff, but it's actually quite physical and very tedious. And so when I'm training early-in-careers scientists, part of what we have to go through is just building up the tolerance to be able to manipulate, you know, fine tune motor type skills for hours on end, where you're not actually spending that time thinking deeply. Just imagine if we were able to offload and automate some of these mechanical, tedious tasks, how much more time we could spend on real fundamental questions that drive the science versus just getting it done.

Danielle: We've talked a lot about what this could look like in theory, but can y’all give me, like, a concrete example of something that's already there or that you think the field is currently working on and is, like, gonna be achievable soon?

Margaret: So what we've been able to do recently and have some really nice proof of concepts are on having AI program automation to do the exact protocol that scientists want to do. And we have this in, you know, little mini bite-sized versions now, and this is gonna be able to be something that we're gonna be able to do, I think, you know, better and better, and more and more easily. So the way that scientists are struggling to program automation, I think that's, you know, I think it's pretty much a breakthrough to be able to have AI do that for folks.

Andy: And not have to like, go to some highly specialized person to –

Margaret: Who doesn't have enough time. Yeah.

Andy: Right. Yeah. So, I mean, I think that just lowers the bar and makes this stuff, like, a lot more available to people.

Margaret: Yeah. Being able to, kind of, democratize automation, the programming of sophisticated automation, I think that's a breakthrough that is, you know, just beginning to start to roll through. I'm excited about it.

Danielle: That's one of my favorite things that I've heard people talking about, is, like, as new technologies are developing, democratizing that for other scientists is –

Margaret: Mhm.

Danielle: – like, I feel like, fundamentally important as a scientist, and it's great to see that these tools also enable that.

Margaret: Right.

Danielle: Okay. So this has been a lot of fun. I gotta ask before I let y’all go – what is the thing you're most excited about in your work and in the field?

Andy: Like, for me, it's just this sense of, like, here I've had this decades-long career in software engineering and suddenly we're in this time period where, like, a week is like what two months used to feel like. So that part is crazy and interesting and exhilarating and scary and all of those things. The way in which we want to harness it is to, you know, do more discoveries faster. And I think if we are doing way more experiments, which is the implication of doing a better job with automation, then we will get more insights faster, and when we get more insights faster, we will invent new drugs faster, and that's ultimately what we're here to do.

Margaret: And if we can invent them, you know, faster and more of them in parallel, we also have the potential to reduce the cost of making drugs, which I think could be really exciting. I think similarly, you know, after many decades working with automation, I think it's really about to deliver its promise of really supporting the scientist and really being walkaway. You know, really completing their experiment, either without errors or being able to figure out its own errors, and, you know, correct them without somebody having to drive back in and move things around. So I think we're about to see it really start to work.

Danielle: I think for every scientist who ever had to take a nap in lab, thank you. [Laughs]

Margaret: Yeah. [Laughs]

Danielle: This has been really wonderful. It's been so nice to meet y’all. This has been fun. Thank you so much for joining us.

Margaret: Thank you.

Andy: Thanks for having us.

Karen: Great episode. You started the show talking about Futurama though. If you could jump in a cryo chamber and wake up in a lab 100 years from now, what is the one thing you hope hasn't changed about being a scientist?

Danielle: Access to coffee, full stop. More seriously, I would say it's the back and forth kind of arguments that we have. I know this sounds, I mean, y’all know I'm a peach, but my favorite thing is actually arguing with other scientists and figuring out new ideas and new directions to go, and so I think that even as we automate things, bring in AI like out the wazoo, that's still going to be like a core part of being a scientist. Over coffee.

Wellington: What's one crazy experiment you would have done that you didn't do, or like, or failed at, or thought it was too impossible or it just took too much time?

Danielle: Hmm. I think I would've done a lot more in neuronal signaling mapping where I would have, like, much more accurate patches into much more complex larger organotypic cultures and kind of done more at like a system-to-system communication aspect. I mean, what's really crazy to me are that the new scientists entering into the workforce now are going to have so many more tools that are going be available and they're going to be able to modify and grow with them so much faster. I cannot wait to see what happens.

And that's our show! Thanks so much for listening. If you haven't already, rate our podcast wherever you listen. It'll help new people find us. And make sure to subscribe! If you have any questions about the show, you can contact us at podcast@gene.com. And now for me, it's back to stalking cells!

The name Two Scientists Walk Into A Bar is under license and used with permission from the Fleet Science Center

For millions of people across the U.S., vision loss is more than a condition - it’s a life-altering obstacle that often strips away independence and cherished experiences. Among the many causes of vision loss, retinal diseases can be particularly detrimental, affecting people’s livelihoods, family relationships and social interactions.

While treatments are available to help manage retinal diseases such as wet age-related macular degeneration (AMD), diabetic macular edema (DME) and retinal vein occlusion (RVO), continued research and innovative thinking are still needed to better preserve and even fully restore people’s vision, and in doing so, to improve the lives of the millions of people impacted by these conditions.

With Genentech at the forefront of biopharma research for 50 years, and over two decades of ophthalmology innovation, we’re proud to have taken a leading role in the significant scientific advancements that have transformed the treatment of eye diseases. Our work in ophthalmology began with leveraging translational medicine to address the role of vascular endothelial growth factor (VEGF) in the major retinal vascular disorders, including wet AMD, DME and RVO. Today, we are building on that foundation by leveraging emerging expertise in inflammation-driven pathways, including those mediated by interleukin-6 (IL-6). This scientific focus enables us to expand further into retinal and ophthalmic diseases with significant unmet needs, including uveitic macular edema (UME) and thyroid eye disease (TED).

Parallel Drivers of Disease

As early as the 1980s, we isolated, cloned and then targeted VEGF, a protein responsible for blood vessel growth that is produced in excess by tumor cells. Such research enabled us to develop targeted anti-VEGF cancer treatments that blocked blood vessel formation and prevented tumor growth.¹ Following the discovery of VEGF's role in retinal disorders, we developed a humanized antibody fragment designed to inhibit this seminal pathway, creating a novel therapeutic approach for serious ophthalmic conditions.²

As Genentech celebrates a half-century of redefining biotechnology, we are also taking a moment to reflect on our own milestone: 20 years since this foundational anti-VEGF research launched our ophthalmology portfolio and revolutionized the treatment paradigm.

I will never forget the first patient I saw with a new wet AMD lesion that had bled late one Saturday night, causing a complete and sudden loss of sight in the patient’s eye. Telling someone who is already tired and afraid that their vision will never return is terrible - for the patient and their caregiver, (in this case, his wife). The introduction of anti-VEGF therapies fundamentally shifted that reality, offering hope for patients who previously had very few options.

It was exactly that commitment to advancing care of retinal disease that drew me to Genentech. I was driven by a desire to be part of a team fiercely dedicated to research. I wanted to help connect those scientific dots, because I knew firsthand the profound human stakes of untreated vision loss. My badge actually features a simple motto: 'Eat, Sleep, Fix Eyes, Repeat.', and if you include vacation within ‘sleep’, it is something I definitely live by. It is a real honor to be able to work at Genentech – alongside many collaborators across the world – which strives to cure or prevent these blinding diseases.

Targeting Pathways That Lead to Inflammation

Building on our expertise in applying research from other therapeutic areas, we explored IL-6-mediated pathways as a potential driver in ophthalmic conditions where inflammation is a key driver of disease, like UME or TED.^3,4 IL-6 is a cytokine that coordinates immune responses and is implicated in inflammation in autoimmune diseases. Both UME and TED have limited treatment options and profoundly impact the patient’s quality of life. For instance, TED is a rare condition predominantly affecting women that can disrupt daily tasks, social interactions, and mental health, with few treatment options.⁵ Similarly, UME disproportionately affects working-age adults, often leading to significant disruptions in careers and family life, and whose treatment is predominantly based on steroids with significant side effects.⁶ Addressing debilitating conditions with limited treatment options drives our tireless work in translating science into meaningful innovation.

Looking Ahead at Genentech’s Ophthalmology R&D

Translating complex science into approved therapies has never been a solo endeavor; it requires decades of relentless collaboration, with hundreds of colleagues across many countries. When I look at photos of my colleagues over the years, I’m reminded of the lifelong friendships and shared curiosity that have defined my time here. It is that exact spirit of teamwork—built day by day, year after year—that continues to drive our expansion today as we tackle new frontiers like UME and TED.

For over 20 years, our anti-VEGF breakthroughs have transformed the field, and now our IL-6 research holds the potential to drive the next wave of innovation. Year after year, our focus remains steadfast: to save sight, improve lives, and bring hope to millions living with vision-threatening conditions. Follow our journey in advancing ophthalmology at gene.com/ophthalmology.

2026

The Next Frontier: Degrader-Antibody Conjugates (DACs)

Building on a decade of trust and shared scientific ambition, Roche and C4 Therapeutics (C4T) have now entered their third collaboration, taking a step into the rapidly emerging field of degrader-antibody conjugates (DACs).

ADCs have made important contributions to cancer therapy, but their clinical utility has historically been challenged by a limited therapeutic margin. Degrader-based ADCs (DACs) represent a potential step-change in this modality approach. By utilising degrader payloads that target specific cancer cellular dependencies, DACs offer a superior therapeutic index. Degraders are characterised by a catalytic mechanism of action—a feature unique to this payload class— rendering them exceptionally well-suited for targeted antibody delivery.

In this collaboration, C4T will leverage its TORPEDO® platform to develop the highly potent, conjugation-ready degrader payloads. Roche will then take over to generate, optimise, and conjugate the antibodies, ultimately delivering molecules for the clinic.

"In less than a decade, C4T has become an expert in discovering and designing highly catalytic, specific and potent degraders in addition to more recently building capabilities to advance the emerging field of degrader-antibody conjugates." - Paige Mahaney, Ph.D., Chief Scientific Officer of C4 Therapeutics

“We are honored to work alongside Roche to rapidly progress this new modality by relying on our commitment to learning from one another and advocating for new, exciting ways to advance scientific discovery. We’re hopeful this shared mindset will help us develop yet another new modality that could offer transformative medicines for patients.”

2018

Expanding the Foundation: Maturing into Oral Heterobifunctional Degraders

The original partnership expanded as C4T developed into a clinical-stage company, which included establishing its proprietary TORPEDO® platform and strengthening its pipeline.

"As C4T grew as a company over the years, it became essential that we stayed flexible and adapted the way we were collaborating." - Barbara Lueckel, Head of Research Technologies at Roche Corporate Business Development

The operational responsibility for drug candidate identification was placed entirely in C4T's hands, with Roche scientists staying heavily engaged in project discussions. This led to expanded agreements between Roche and C4T focused on creating orally bioavailable heterobifunctional degraders. 

2015

The Initial Leap: Entering the world of Targeted Protein Degradation (TPD)

In late 2015, a publication in Science magazine showed for the first time in a successful preclinical in-vivo experiment that binders to an E3 ligase named cereblon could be exploited for targeted protein degradation (TPD). Roche established a relationship with the corresponding researchers just prior to this publication and learned that this new discovery was underway to become the nucleus for a new biotech: C4 Therapeutics (C4T).

When Barbara Lueckel, now Head of Research Technologies at Roche Corporate Business Development, visited the then-empty C4T labs back in 2015, there wasn't much to see. “It was all hope and aspiration at the beginning,” she recalls. However, Roche scientists were eager to work on this new approach, making Roche the very first pharma partner of this promising new biotech.

Given C4T had to build its team and platform, the collaboration was set up so that the research plan was jointly executed by scientists at Roche and C4T. “We really developed the C4T platform together,” noted Stew Fisher, former Chief Scientific Officer at C4T.

For five decades, Genentech has been fueled by a passion for science, driving us to explore uncharted territory and deliver groundbreaking discoveries that have fundamentally changed patients’ lives and reduced the burden of complex health challenges on society. As the original biotech company, we honor our 50-year legacy by applying the same innovative spirit—the one that creates life-changing therapies — to improve healthcare access and outcomes, ensuring our scientific breakthroughs have a real and positive impact on people’s lives.

Defining the Next 50 Years

We are now leading the next revolution in human health, ensuring we remain at the forefront of scientific progress for the next 50 years. Today, we are pioneering the integration of AI and data science into every facet of drug discovery and patient care. This new technological frontier will accelerate our ability to solve the most complex, urgent health challenges in the world.

Defining Moments

Learn more about the inspirational people, scientific discoveries and transformative partnerships that have shaped our 50-year journey.

SEE ALL

Employee LinkedIn Series:
50 Posts for 50 Years

Throughout the year, leaders and employees across Genentech and Roche are taking to LinkedIn to share the stories that defined our past and will shape our future. From scientific breakthroughs to personal reflections on our culture and our deep-rooted commitment to patients, we are telling our story, one post at a time.

SEE ALL

Resources

Inflammatory bowel disease (IBD) is a debilitating, lifelong condition that changes how people plan their entire lives. In this episode, KT Park, Global Head of Gastroenterology and Hepatology, and Seppi Lin, Head of OMNI Early Clinical Development, explore the complex biology behind IBD. They discuss the role of genetics, the gut microbiome, and an individual’s environment, as well as the exciting future of "immune reset" therapies that could offer hope for people with IBD.

If you would prefer to read a transcript of this episode, please click here.

SUBSCRIBE BELOW TO CATCH EACH EPISODE

If you want to learn more about the groundbreaking science happening in our labs, click here. To learn more about the jobs in our research and early development group, click here.

Transcript of Two Scientists Walk Into A Bar: “The ABCs of IBD” with KT Park and Seppi Lin

Maria: I’m Maria Wilson.

Danielle: And I’m Danielle Mandikian.

Maria: And we are scientists. We. Love. Science.

Danielle: Yeah, we do. So, when we aren’t doing it, the next best thing is to talk about science! And what’s really awesome is that we’re surrounded by some of the most brilliant minds in research!

Maria: We are going to step away from the labs today to talk to other scientists about the cool stuff they are thinking about, working on and imagining . . .

Danielle: . . . as well as how some of these discoveries just might lead to new medicines. So, grab your favorite drink, get ready to unlock your science brain and join us for Two Scientists Walk into a Bar…

Maria: The show for scientists, science geeks, and the people who love them!

Maria: I am wondering if anyone knows the difference between IBD and IBS?

Employee responses:

IBD and IBS? IBS is...oh, I have no idea.

Not really.

A disease is not quite a symptom, where a syndrome is a little bit more of a symptom. I'm totally guessing!

One is a symptom and one is a chronic condition, right?

I'm gonna guess that IBD is something bowel disorder.

So one sounds more permanent, versus one that can be like treated or is just kind of like an onset that you can kind of treat, yeah.

Maria: Welcome back! Today, we're going to be talking about inflammatory bowel disease. Now, this is a really scientifically interesting area of biology, and it's also a very personal topic for me because my partner suffers from Crohn's disease, a form of IBD. He's had it since he was a child. People in the IBD patient community and scientists in the field are really, really interested in potential breakthroughs. With me today are two guests, KT Park and Seppi Lin, who are both experts in this disease area. KT treats patients and Seppi develops medicines. We're going to talk about a lot of things in this episode, but let's start by defining what is IBD, inflammatory bowel disease, and what it isn't. It's not IBS. This is something that confuses people.

KT: Yeah, I mean, I could literally speak for the next hour on this but I'll try to keep it brief, and Seppi, please help me. So inflammatory bowel disease is not irritable bowel syndrome, IBS. It consists of Crohn's disease and ulcerative colitis. They are immune-mediated conditions and once diagnosed, it's lifelong. It's chronic. It's relapsing and remitting. And I think the best way to describe IBD so that the audience here just kind of gets it right away is that – imagine yourself as a young adult trying to get your first job or trying to finish school and you are waking up in the middle of the night – 10 to 20 times a night – with bloody diarrhea, abdominal cramping, urgency, hesitancy, rectal blood, anemia, fatigue, you know, fevers and joint pain, hair loss. So it is an absolute devastating chronic condition if you have moderate to severe.

Seppi: I have friends who have IBD, I have family members who have IBS. There may be an overlap in symptoms. However, the biology behind IBD is very different than the biology behind IBS. And KT spoke to some of that around, hey, IBD is associated with spondyloarthropathies, with other conditions, and the risk of other things like colon cancer are much more elevated in patients with IBD. And that necessitates a completely different form of treatment, right?

KT: Yeah.

Seppi: And you know, I – my friends with IBD, one of them can't think about driving. They limit how far from home they drive because they never know when they're going to have another symptom, right? So it really isn't just an inconvenience.

KT: That's right.

Seppi: It's life changing –

Maria: Yeah.

KT: Yeah.

Maria: And very, very unpleasant. What is going on? What's happening in the gut that's causing all of these pain and symptoms and problems?

KT: Well, I mean, you know, in medical school, we used to focus back then on genetics. We used to focus on known gene associations like the NOD2. But now, it's gene plus environment plus some trigger. So –

Maria: And NOD2 is a gene involved in immune function.

KT: That's right.

Maria: Yeah.

KT: So if you imagine this overlapping three-way Venn diagram, you know you are genetically susceptible, but interestingly, in the – in modern era, I would say especially in the last 20 years, you realize that there's an interplay with environment, foods you eat and your microbiome, in particular, whereby folks who we never thought would have been highly susceptible to IBD, such as underrepresented minorities, in particular, they end up presenting with new-onset IBD. And then, there's that trigger, whatever the trigger may be, but it's almost always invariably due to some stress factor that the patient might be going through. It's often sleep deprivation or studying for a major exam. [Laughs] It happens in the worst time. I remember several patients, you know, the week before their MCAT or their LSAT, the months leading up to their wedding, you know, trying to study for the bar – there are so many stories of that stressful event creating absolute chaos.

Seppi: I think what's been challenging for the field is we haven't gotten a completely good handle on the different causes of IBD. Clearly, it's about a dysregulated immune system, but it's not just a single pathway.

Maria: And it's not like an autoantibody, autoimmune disease like type 1 diabetes exactly, right?

Seppi: It is not. Like, so if you compare, let's say, IBD to another inflammatory disease like psoriasis or Th17 cells, and the IL-17 and IL-23 pathways are central in the pathogenesis of psoriasis. That is not the case –

KT: Yeah.

Seppi: – for IBD, and it's because it's multifactorial. There's more than one pathway involved. There's more than one cell type involved. That leads to difficulty in diagnosis and treatment because different patients have different cardinal pathways that are involved in what's going on with them.

KT: Yeah, Seppi's right on point here. There are so many pathobiology mechanisms of action that are at play and you don't even know at the time of diagnosis or after two years into a stable therapy which one is dominating, you know, the actual presentation.

Maria: So we talked about the genetics and the environment. When you think about those two things, you tend to think about epigenetics as well. Like what else is going on? Is it true that the disease is increasing in incidence? That seems to be the case. You were saying it's increasing in ethnic groups that didn't typically have the disease –

KT: Yeah.

Maria: – in the past. What is this?

KT: Yeah. I mean, it's got to be epigenetics.

Maria: Yeah.

KT: There is some epi – [Laughs] – or other or non-primary gene that – genetics that's driving the phenotypic overexpression towards immune dysregulation of a Crohn's patient or an ulcerative colitis patient. And it's just fascinating to think about because the immigrant from, let's say, Asia, who you would not think would be naturally predisposed to having increased incidence of Crohn's disease and ulcerative colitis compared to a Caucasian in New York, for example, in the Northeast America, they come to the States, they are now, after 10 years of being here, have the same incidence rate as someone living in that particular geographic area.

Maria: And that's true of people who are born in this – in the U.S. – as well, right?

KT: That's right.

Maria: Interesting. They take on the risk of the location –

KT: That's right.

Maria: – rather than the risk of their ancestral genetics.

KT: We do think it's diet related.

Maria: Yeah.

KT: We also think there's some other component of the environment. We don't know.

Seppi: And in terms of response to therapy, we know that patients who see a TNF agent – their immune system, in many ways, is permanently different after that.

Maria: Even if you stop taking it?

Seppi: Yes.

Maria: Wow.

Seppi: And that's why people switch, right?

Maria: Yeah.

Seppi: And so epigenetics, or non-nucleic-acid-based modification of the genome, is probably one of those factors driving that different type of immune memory, if you will.

Maria: And that might be good because it's therapeutic? Or it could be bad?

Seppi: I think it might help us define what maladaptation is happening to the immune system in terms of losing response to therapy and what other therapies may be warranted down the line. To geek out even more –

Maria: [Laughs]

Seppi: – we can tell to some degree, just by the transcriptome in the gut, what therapies a patient may or may not have been on, or whether they may or may not have responded to that therapy in the past or the future.

Maria: And that's from the – from the gut line, the cells – just a biopsy from the gut?

Seppi: Yes.

Maria: Yes, or a whole bunch of different cells in there?

Seppi: Yes.

KT: And speaking of cells, I just have to comment that, if you look at the microbiome changes that happen when you go to a whole different area of the world, it's amazing how you become your microbiome fingerprint of the people you are living with, right? And that's been studied quite a bit in microbiome research particularly with the Hadza Indians of sub-Saharan Africa. And it's just fascinating to think about that somebody from a whole different part of world can go to another part of the world, and their microbiome imprint is now very much that of, you know, another ethnicity.

Seppi: Yeah and I think the microbiome intersection with diet is really interesting right, because there's a lot of work looking into the diversity of what we eat matters –

Maria: Well, that's feeding the bugs, right –

Seppi: – matters to the diversity of what's in your gut.

Maria: – and you feed the bugs you want. Yeah.

KT: Eat your kimchi and kefir. That's what I say.

Seppi: If we look at folks who are classic, let's say, hunter-gatherers – right? – their diet, which is no processed foods, their microbiome is completely different than, let's say, a super Western diet, which contains a lot of processed foods. And so there is an angle there. It definitely contributes.

KT: And I love the idea that your other self, the microbiome that's living inside of you and on you, is actually inducing change to your genetics.

Maria: Mhm. Yeah.

KT: Isn't that fascinating?

Maria: It is fascinating. Yeah.

KT: And so Seppi, who are you? Are you the microbiome Seppi? Or are you –

Seppi: [Laughs] No, that's super interesting.

Karen: Hey, Maria!

Maria: Hi, Karen!

Wellington: Hi, Maria!

Maria: Hi, Wellington!

Karen: I was really fascinated by the epigenetics conversation and how that impacts IBD, but I'm realizing that some of our listeners might not know what that term means. Can you give us some more detail on that?

Maria: So, epigenetics is the study of how non-sequence, non-mutation modifications to DNA affect the expression of genes. This is about whether, in a particular cell type, a bit of chromatin is open or closed. Or it's often methylation you'll hear about and histone modification. So while the sequence of the genome is unchanged, the way those genes are expressed in a cell can be modified by how the chromatin is accessed, and environmental factors can influence epigenetics. So over the course of a person's lifetime, while the underlying DNA sequence hasn't changed, the way genes are expressed in a particular cell might change over time.

Wellington: Maria there was the mention of an immune memory. We've talked a little bit about this when we talked about obesity and I asked you about the set point. Is this at all related?

Maria: Ah, not really, I would say. There's sort of physiological set points around things like blood pressure, temperature, body weight – these are these homeostatic processes that go up and down. You think about it like a thermostat. Whereas the immune system – and I am not an immunologist – is much more complicated, so there is not like a one set point for inflammatory status or immune status or something like that. It's a milieu of cells and cellular and tissue interactions that make up your overall immune system health. But it's not as simple as a binary set point for your immune system.

Maria: Could you tell us a little bit about the classes of medicines that are available at the moment and what biologies they target as well so that people can understand that a little bit better?

KT: Sure. Many people know about anti-TNFs. It is – it has been our backbone biologic for several decades. You have IL-23s. You have JAK inhibitors. You have one particular gut-selective anti-integrin. And there are a couple of other players, like an S1P inhibitor.

Maria: But they're all different mechanisms of suppressing a particular subset of immune cells or preventing cells from trafficking to the gut, right? That's the kind of main mechanisms that we're looking at.

Seppi: Yes. And initially, the targeted therapies that were approved were borrowed from other disease areas. So the TNFs were borrowed from rheumatoid arthritis, etcetera. It wasn't until the integrin therapies came into play that there was really a fit-for-purpose drug dedicated for IBD. And I think that's part of the challenge in terms of the science because I think we're only beginning to delve into the specific biologies around IBD. It's really only in the last five to 10 years that we've seen therapies specifically designated towards inflammatory bowel disease as opposed to just an immunology drug, if you will. And I think we're, through the same work, understanding that, hey, single therapies aren't enough, right? Because a lot of the work we're doing now is around combination therapy, right? Because we were talking about the patient journey earlier. Even patients that are getting good biologics – and KT, correct me if I'm wrong on these numbers – maybe half of them are, let's say, in remission at the end of eight to 12 weeks. But if you look at that remission rate out for a year, the minority of patients are in remission. It's like 20-25 percent ish, right?

KT: Yeah.

Seppi: But if I think about that as a patient, that sucks. [Laughs] Right? I have a one-in-two chance of being in remission at 12 weeks. And I have a one-in-four chance of being in remission at a year. That's not great. We have done better in other diseases. And these patients deserve something different.

KT: My goodness, Seppi, I might have to get you through a GI fellowship because you're talking like a gastroenterologist. [Laughs] I love it. I love this. But I, you know, Seppi is absolutely right. The primary non-response, basically those patients who are not responding after that initial induction period, which is around week 12, is already pretty high, you know, compared to other diseases, right? And then, you talk about secondary loss of response, which is either antibody-drug mediated, or you have an unknown loss of response, namely because the immune system is so darn smart. And that's why you have such a minority of patients at the end of one year holding on to clinical remission. Yeah. It's dismal.

Seppi: Yeah just, you know, I brought up psoriasis earlier as an example of another immune disease, right? Over 90 percent of patients can expect primary response and clearing of skin with the current standard of care. If we could get there with IBD that would be incredible.

Maria: How do we do that? There clearly isn't such a one clear biologic pathway like in psoriasis. Well, maybe there is, but we haven't found it yet. What are the other approaches that we are considering to get to these breakthrough therapies for patients with IBD?

Seppi: There's more than one approach.

Maria: Yeah, let’s hear them.

KT: Yeah.

Seppi: So, I mentioned one of them, which is combination therapy.

Maria: So existing therapies and putting them together.

Seppi: Or taking mechanisms that we know are important and putting them into a combination in a molecule –

Maria: I see.

Seppi: – which will be like a bispecific antibody or something like that. And precision medicine, right? Precision medicine is common in oncology. And yet, we haven't deployed that adequately in inflammatory bowel disease yet. If we can define which pathways are important for one particular patient, we could do a much better job at giving them therapies that we think will benefit them the most. But I think it's going to be more than one potentially.

KT: And by the way, for the last 20 years, we've been talking about personalized therapies. We just haven't been able to get there. It's just been a very difficult path to take.

Maria: So while we don't have personalized therapies yet, and we are, as a field, developing newer therapies that may have future breakthroughs, I've also heard you say that of the therapies that are available to patients, it's really important to pick the right one first.

KT: You want to hit hard and hit right in that window of presentation, in that first six months to one year max – and if you're not getting that patient into deep remission, that patient has now a different kind of relapsing remitting trajectory years later. So, you know, obviously, combination, but you have to hit hard in the beginning. So if your monotherapy is able to aggressively reset better than another monotherapy, absolutely, we have to be able to pick that one as opposed to an inferior one.

Maria: But you need to know from that patient's biomarkers –

KT: Right.

Maria: – biology, biopsy, whatever, which therapy it would be –

KT: That's right.

Maria: – because at the moment, you're just essentially, you step through the levels of care that the guidelines suggest.

KT: One more thing – and that’s absolutely right, Maria, so thank you for bringing that up – I just want to make sure that everybody understands how different IBD is because it's on the inside. Psoriasis – you can see it. You can see your healed skin. IBD – there is no way to tell unless you do a scope. So I would not stop treating aggressively until I saw what's called mucosal healing, meaning if I went in there with a scope and I saw the mucosa of the intestinal lining looking pretty much back to normal, that's when I could back off. But that is not what we do. The clinical symptoms often dictate, you know, pulling off on the gas pedal and it's way too premature. And that's what we do like 95 percent of the time in this disease. The patient feels better with steroids. Oh, okay. So let's just get you off of them and let's see how you do. Well, guess what? If you were to scope that patient when they said they "felt better," their colon does not look right. And that's been a challenge, right? It's not like the skin that you can just like, hey, let me see your skin.

Seppi: And to restate what KT is saying – right? – I think patients want to feel better, and we've been treating symptoms as opposed to an immune dysregulation. We need to be treating the immune system and not just symptoms alone. And I think that is the breakthrough that we're looking for. And fortunately, unlike, let's say other diseases where we can't biopsy a tissue, whether it's a brain disease or a lung disease, here we have access to the gut pretty easily. And we've –

KT: With a cleanout though. [Laughs]

Seppi: Yes. And –

KT: I don't know how easy that is.

Maria: I was going to say for people having a regular colonoscopy, how easy is that? [Laughs]

Seppi: – we can get to the molecular mechanisms behind what's happening in the gut. And that's pretty unique I think. And so I think we have an opportunity to do what we've been trying to do for the last few decades in terms of personalized precision medicine, in terms of understanding the biologies that are driving disease in that particular patient.

KT: One other thing that I think deserves some sort of a place in this conversation of what do we need to do differently – we have to think about immune reset.

Maria: Yes. I was hoping one of you would get to this fascinating topic. Tell us about immune reset.

KT: In this field, we have thought about immune reset in the context of B-cell therapies. While B-cell therapies, we know, is at play – B-cell dysfunction, dysregulation is at play in IBD – it is not the dominant player. In fact, when you look at patients with aggressive disease or very much relapsed disease, it's your myeloid cell lines. It's your fibroblasts. It could be your plasma B cells, but you still have the background of your neutrophils and monocytes and, you know, the standard players that have been dominating research in IBD. Seppi, what do you think?

Seppi: Well, I think it's super exciting in terms of what the immunology field has done in immune reset. And yes, a lot of it has been based on B-cell-directed therapies.

Maria: And basically wiping out the B cells and letting them start over.

Seppi: Yeah. And wiping them out deeply enough that the rest of the immune system resets. And that's what we mean by immune reset. You know, something like that must be possible with the remainder of the immune system, whether it's T cells or myeloid compartment, etcetera. I used to have a joint appointment in the bone marrow transplantation division and we used to do that with bone marrow transplantation, which is a highly morbid and risky thing.

Maria: Yeah.

Seppi: But that would reset the immune system. We're looking for ways in a less aggressive, less morbid way to get the immune system to go back to “normal.”

Maria: Is there any anecdotal evidence of people with IBD having –

KT: I have some, yeah.

Maria: – bone marrow transplants?

KT: Yeah, absolutely. There are some good case studies in the New England Journal. In fact, I had a patient who underwent a hematopoietic stem cell transplant – very well matched donor, 11 out of 12 – and we were able to get that patient in a curative state.

Maria: And that was specifically for the IBD?

KT: For Crohn's. Yeah, because –

Maria: Yeah. Yeah. Okay, it wasn't a transplantation for cancer.

KT: We were looking at a transvaginal fistula. This person had a complete anal-rectal occlusion from the inflammation and had five surgeries.

Maria: Wow.

KT: And so there was no way we could keep doing what we were doing. And so we had this massive conference just on her and we decided to go for the stem cell transplant. She responded beautifully –

Maria: Wow.

KT: – for a year and a half. The disease did come back, but interestingly, not to the same degree of severity. The fistulas went away but the inflammation stayed.

Seppi: Yeah. I think immune reset – and it's actually much closer in terms of being able to achieve that than we've ever been before. And in the meantime, there are plenty of other things that we're doing to try to get patients to a better immune state.

Karen: Maria, how's the field looking for non-invasive procedures or newer biomarkers to measure IBD, or if someone has IBD?

Maria: So I think at the moment, the gold standard of understanding the state of somebody's bowel is still using colonoscopy because that's relatively – it sounds silly to say this – but that's a relatively non-invasive procedure. It's quite safe. It's not like a liver biopsy or something like that, for example. For example, in the fatty liver disease field, there's been a huge, huge effort to develop non-invasive biomarkers because there's a non-zero chance of death from a liver biopsy, whereas in IBD, well, yes, it makes trials easier for patients, it does make treatment less onerous for patients to have biomarkers. There's been I think a little bit less of a push to get really good biomarkers that would replace imaging, you know, direct imaging. But that being said, for novel mechanisms of action, developing a good biomarker that tells you whether you engage that target is really, really important for understanding whether a new drug worked or not. Did I have a good drug that engaged the target and it didn't work? Or did I have an inferior drug that didn't engage the target properly, so I need to make a better version of that drug? That's always a really important question in diseases like this.

Maria: And when we're doing our clinical work, there's a – across the industry – several new agents in trials for this indication, And are people mostly focusing on the people who have failed the other lines of therapy? Is that how we're developing the new drugs?

KT: Well, the unmet need is in moderate to severe, right? So that's where most drug discovery – drug trials and discovery is happening. But I think there's great unmet need for the milder patients because that represents a massive unmet need where patients don't even get on anything and they just sort of cycle through steroids here and there.

Seppi: Yeah. Maria, to your question around what are the patient populations that we're studying the most for new therapies – yes, a lot of those patients have failed what we would call targeted therapies. It doesn't matter if they're an oral small molecule or an injected large molecule, they target specific immune mechanisms. And patients who fail those end up doing much worse. And we're always looking for therapies that can benefit those patients. But to KT's point, patients at the beginning of their journey, patients who haven't started targeted therapy yet – those are the patients who may have the best opportunity to be in remission for longer also.

Maria: Yes.

KT: Hundred percent.

Seppi: And we need to think about what are the best strategies for them.

Maria: Yeah. And that's not how the current medical system is really set up –

KT: That's right.

Maria: – to aggressively – it’s like this is a chronic disease. Yes, you're not super sick right now. But we want to make sure that you stay healthy, so we're going to aggressively treat your disease earlier, and that's not how we do it.

KT: Yeah. That's right. The lack of education. Patients’ perhaps trepidation in the beginning to get on targeted therapies and, obviously, our healthcare system – they all contribute to the fact that 70 percent of IBD patients are either not on medicines or are not on the right medicines.

Maria: And despite the fact that they're advertised all over the television. It seems you can't –

KT: Yeah. It's become more –

Maria: – turn on the TV without seeing an advert of them.

KT: Turn on the Super Bowl, and you see.

Maria: Yeah. One of our favorites is that woman – there's a young woman eating lunch on a boat.

KT: [Laughs]

Maria: And only an IBD patient looks at that advert and goes, "That must be a good medicine because she's prepared to each lunch while on a boat, which doesn't have a great bathroom.”

KT: That was very strategic marketing involved there.

Maria: Very good marketing.

KT: Yeah.

Seppi: Well to be fair, I think that strikes at what is the patient experience like, right? Because some of the work that the field is trying to do is come up with different and new endpoints that better capture what the patient experience is, right? So nothing will ever replace endoscopy – that's super important for looking at the tissue, getting samples of the tissue. And we have pretty antiquated instruments in terms of measuring how a patient was responding or what level of function they're getting back, and not every patient is the same. A lot of these instruments or endpoints are tuned to look for a signal in an entire population – they're not necessarily designed to get at what each patient is experiencing. And I think, with the advent of wearables and different types of software, I think we're in a new age where those kinds of assessments are possible. And –

Maria: Sounds like you're going down the AI route, Seppi –

Seppi: Well, I think –

Maria: – and how AI is going to help us. [Laughs]

Seppi: Well because a lot of these instruments are capturing a lot of data passively, right? And to sort through those mountains of data, you need some kind of agent to do that. And most of the time, that's some kind of AI agent, right? And so it's the combination of the hardware, the software and the analytics that's allowed us to think about this a bit differently and to capture the patient experience a little bit differently.

KT: I'm just trying to pinch myself because I have a critical care doc sitting next to me talking about colonoscopy.

Seppi: [Laughs]

KT: I love this. But yeah. We can go to data.

Maria: Danielle, my cohost, she has a question that she wanted me to ask you guys, which was, what about the microbiome as a therapeutic target or engineering strategy? Do you think there's anything to this? Or is it just a load of –

KT: [Laughs] What did you say, Maria?

Maria: [Laughs]

KT: So there is something to fecal microbiota transplantation, FMT. We did trials on this not too long ago before the pandemic. It was quite in vogue, but people are now starting to realize that there is no permanence to microbiome therapy for longer-term remission. In fact, there was data presented not too long ago that showed that you literally have to get an FMT every two weeks.

Maria: Because the natural microbiome just overgrows anything that you transplant.

KT: Yeah. And the treatment effect was not so spectacular.

Seppi: It's a two-way street. The microbiome affects the immune system. The immune system affects the microbiome. And no one therapy is going to completely normalize the other. And I think that's why we're seeing these temporary effects from trying to reset the microbiome. You haven't fully reset the immune system either, right? So there's no question the microbiome is important. However, will a microbiome-directed therapy cure IBD? Never say never, but the chances that that's going to be the case are very low.

KT: But it's a super interesting topic. I love talking about microbiome and poop.

Maria: I've known several gastroenterologists in my life and it does seem to be a theme.

KT: [Laughs]

Maria: Great dinner guests. [Laughs]

KT: We're weird.

Seppi: Well so one of the – your questions earlier, Maria, was around what else should we be doing? So just by the nature of who we are, we – KT and I focus on therapeutics. However, to KT's point earlier around early therapy matters, a big part of that is diagnosis. There are a lot of patients out there suffering with symptoms for months, years before –

Maria: It's embarrassing, right? It's not something you talk about.

Seppi: – before they are appropriately diagnosed.

Maria: Yeah.

Seppi: And some of that is education around symptoms. Some of that may be around access for patients to medical care. But I think we need to think differently about diagnosis as well, because the earlier we can diagnose patients, I think the better they'll do.

KT: Yeah. And I can't tell you how many times I saw patients who thought that for decades they had IBS. Which is such a shame, right? You could have had a better life for a decade or more.

Maria: Last lightning question for both of you: what's one thing that you think we will see or you would like to see 10 years from now for IBD patients? And I'm going to start with Seppi.

Seppi: Curative therapy. And that could be anything that resets the immune system permanently. So you get treatment. It could be a one-time thing, it could happen over a few weeks, but after that, the immune system resets to normal and you don't need chronic therapy after that.

KT: Yeah I, you know, I can't top the cure, but I will say, in the 10 years, which can go pretty quickly, I would say aggressive reset, which is really about trying to be – trying to give the therapy that is so good in the beginning at the time of diagnosis that the patient now goes into deep remission, and we're not just treating the surface-level symptoms.

Maria: Thank you so much, both of you. That was a fantastic conversation.

Seppi: Thank you. It was a great discussion.

KT: Yeah, thank you guys. This was so fun!

Wellington: Maria, this was a great episode. I learned a ton of things that I had never known and you started off the episode talking about, you have some personal experience. I thought when you guys were talking about the – and laughing about the commercials and the worrying about things that – I didn't quite realize how sort of moment-to-moment this disease is.

Maria: Yes, so I think one of the things that impacts your life when you have somebody in your life with IBD is really about when you're thinking about planning for the future – just simple things like where am I going to go on vacation in 12 months – and you're just always making assumption that I'm still going to be okay, that my medicine is managing my disease. We always take out travel insurance, for example, because we may have to cancel a vacation at the last minute because of a health issue. So I think for us, if there was a therapy that you could just believe in that it was going to work forever – you know, how much confidence would that give you just to live your life as if you didn't have the disease. That would be a real breakthrough.

Karen: Maria, what was the most interesting thing you learned on this episode that you didn't know before?

Maria: I think I didn't really appreciate the impact of that true immune reset that has been observed in certain cases with stem cell transplants and stuff. So I think understanding that it is, at least in some patients, possible to cure their disease – that was really exciting for me to hear.

And that's our show. Thanks so much for listening. If you haven't already, rate our podcast, wherever you listen – it will help new people to find us. And be sure to subscribe. If you have a question about the show, you can contact us at podcasts@gene.com. That's G-E-N-E dot com. And now for me, it's back to wrestling with data!

The name Two Scientists Walk Into A Bar is under license and used with permission from the Fleet Science Center