Oby: From the campus of Harvard Medical School, this is Think Research, a podcast devoted to the stories behind clinical research. I'm Oby.

Brendan: And I'm Brendan, and we're your hosts. Think Research is brought to you by Harvard Catalyst, Harvard University's Clinical and Translational Science Center.

Oby: And by NCATS, the National Center for Advancing Translational Sciences.

Brendan: The complex nature of cancer tumors presents many challenges for understanding disease and treatment. Researchers spend weeks or months growing cells in lab settings and developing cell lines. Now, researchers at the Muthuswamy lab are developing organoids that can more quickly mimic the three-dimensional properties of a patient's tumor. In turn, a small tumor sample can better inform treatment, resistance, and variants of such heterogeneous tumors.

Dr. Senthil Muthuswamy is an Associate Professor of Medicine at Harvard Medical School and the Director of the Cell Biology Program at the Cancer Center at Beth Israel Deaconess Medical Center. Dr. Muthuswamy, welcome to Think Research. Thank you for joining us.

Senthil Muthuswamy: Thank you very much, Brendan.

Brendan: Your lab has developed a method for growing cancer tumor cells to use for translational and clinical research. Tell us a little bit about the method and this new platform you've developed.

Senthil Muthuswamy: Yeah, happy to. So traditionally, we grow cells from a culture and establish cell lines that are allowed to develop into cell lines over a period of sometimes months or weeks. What we decided to do some time ago is to develop culture conditions where we can keep the patient tumors alive without minimal evolution in culture. So we relied on developmental biology principles and came up with media that is likely to support the growth and survival and expansion of tumor cells in culture without having the need for them to go through a long-term adaptation to culture.

And the second thing we also did was instead of putting them on a plastic dish, we tried to grow them on a matrix. That way, they maintain some of the three-dimensional properties that are reminiscent of what is happening in a patient. So with these two nuances, we are able to now take patient tumors, digest them, and then put them in these culture conditions in the proper media. That allows them to slowly adapt and grow much faster than a cell line but slower than a patient tumor. And eventually, they will adapt into culture into three-dimensional organoids.

Now, we use two criterias to make sure we are doing the right thing. One is that they have to divide and multiply and all those things, and that's how we can establish them in culture. But more importantly, we want to make sure the way they're growing in culture is they are still able to retain some of the histological features of how the patient tumor looks in the patient. And that parallel is important because only then we can be sure that we are actually growing a mini tumor in the lab without significantly changing it from how it was in the patient.

Brendan: Great. So that word you just used, mini tumor. So you're taking a portion of a patient's tumor, growing it in a lab, and that tumor is basically identical to the tumor in the patient's body?

Senthil Muthuswamy: Yes and no. Yes in the sense that it is a sample of the tumor in the patient's body, so it is similar to the patient tumor. No because we have learned over the past couple of decades that a patient tumor is not one tumor, it's very heterogeneous within the tumor. Portions of the tumor are very different from each other. So we are only getting a sample from the tumor, so we don't know if this will truly represent the entire tumor. But whatever we get is still part of the tumor. So it is a sample, but there is a chance that we are only taking a sample and it doesn't represent the entire tumor.

Brendan: And you talked about the difference between the cell lines and the time it takes to grow cells in traditional cell lines and contrasting that with what you're doing. Tell us a little bit more about that contrast and why this organoid platform is going to be helpful for research.

Senthil Muthuswamy: Sure. I think there are two salient differences between a cell line and these organoid culture differences. One is the cell lines usually tend to suffer from what is called a culture-induced drift. Because of the long-term culture, the long-term maintenance, and also taking a long time to establish them into a line, they are not exactly the same as what it was in the patient where the cells were obtained from, so they've gone through a drift. So they are genetic drift, phenotypic drift, however you want to do it. All of them actually happen during this time.

The organoids, on the other hand, we try to maintain them as close to the patient tumor as possible. So they're both genetically and phenotypically closer to the patient tumor than they are to a cell line, so that's one big advantage. The second advantage is that since we are able to maintain a patient tumor, if you have a panel of tumor organoid lines from one particular cancer type, be it pancreas or breast or lung or colon, those panels are representing the heterogeneity of tumor models between patients.

Because we know today that one drug, a cancer drug for one particular type of cancer, not all patients respond the same way. Patient-to-patient variation in drug response to the same approved drug is significant. We don't really know what the differences are. It could be physiological, it could be their body weight and obesity, it could be their age, it could be their gender, it could be their race. It could be many factors. Nevertheless, we don't know what the contribution is from.

So with having these models that are representing a patient population, we can call it as a mini tumor population model, so then we can understand population responses. And that is really unique to organoid models. That cannot be done effectively.

Brendan: Yeah. And I think we should probably step back a little and talk about what the purpose of these organoids is. We were talking about cancer research and you just mentioned evaluating drugs, so maybe tell us a little bit about the why. Why are you doing this?

Senthil Muthuswamy: Well, I think we have now the knowledge in genomics and all the omics platforms that's told us a lot about what cancer is. We know they are highly heterogeneous, there is a lot of mutations, there are a lot of changes in the epigenome, a lot of changes in the proteome, and there is a tremendous amount of heterogeneity within a tumor of our patient. So we know all that now, but now we need to know how do we actually use this knowledge, use the models that we have to understand how to treat them?

So this is where I think patient-derived models come into play because if you know that the cancer is a highly complex disease for one cancer between patients, the current cell lines do not do justice to represent that entire population of the models. So we need better ways to actually capture the variations that we know exist in a cancer population. So we can now use that knowledge, A, to find better drugs, to find out what treatment works better and what combinations work better, and B, also to find out how a given drug resistance is developing. What we do when a patient develops resistance? And then also to find and validate new targets that can potentially be used in a clinical setting.

So I think this organoid platform is now a clear evolution in terms from cell lines to here because it now matches what we have learned from all the omics approaches, and we know we can now match the omics to these new generation of models that are more similar to the patient than the cell lines. That is the biggest advantage in my opinion.

Brendan: Yeah. So I guess the overarching goal is to simulate as close as possible the characteristics of the patient's tumor and their specific tumor biology.

Senthil Muthuswamy: Exactly. Correct.

Brendan: So you've been studying cancer in humans for many years, but your early training was in plant genetics and you studied plants. But you moved to cancer research for your graduate studies. What made you interested in plant genetics in the first place and why did you change your research focus in graduate school and throughout the rest of your career?

Senthil Muthuswamy: Sure. We are an agricultural family. My grandparents were farmers, my dad is an agricultural scientist. So I was drawn to agriculture as a field for undergraduate education. That's how I got into it.

But then once I was in my agriculture program, I realized that genetics and molecular genetics is really cool. It's really interesting. I think these were the days when recombinant DNA technology was the heyday. That was the talk of the thing in science, the next wave of science, so I really wanted to do that, and that was best done in the genetics and molecular genetics context.

So I went into that field initially for a masters in India, and as I was learning more about genetics and molecular genetics, I realized the biggest advancements in this field happen in the mammalian context, not in the plant context, and then they are adapted into the plant context. And since I was fairly young and very new in my path in terms of career, I said, this is the time to switch if I want to switch. So that's what prompted me.

And I moved to a PhD into cancer biology and molecular genetics because most of the active research in this field happens in North America, not in India. We usually tend to be a decade or so behind at that time. So that's partly my motivation to move into this and the rest is history. But I then moved into molecular biology, cancer biology type of research.

Brendan: And talk about your training. And what was the introduction to cancer biology like? And how did you navigate starting in a whole new field in graduate school?

Senthil Muthuswamy: Yeah, it sounds crazy even to me now. How would I make that change? It was not easy.

It was quite challenging in the beginning because I was not able to connect with any of the terminologies, the concepts, the background that you're supposed to have. But given that I was fairly young at the time and you are still a student, I was able to dedicate more time to learning some of the basics. But I think what really helped is the education system that was there allowed me to initially balance my courses so that I learned broadly first before I go into the more mechanistic and advanced courses. So the time to allow myself to learn broadly helped a lot.

And during that time, I was simply learning what is cancer. So I would take pathology courses or audit pathology courses and other basic courses, and that really helped me refresh my understanding what it is and also learn some of the basic principles before I went into the advanced courses. And I think that process was what really helped me adapt and move in from completely one background in plant genetics into this.

But having said all that, I will say that some of the principles of molecular genetics are the same in plants and mammalian, so that part was OK. I think it was to cell biological and physiological concepts that I had no idea about. I had to learn through some of these broad courses.

Brendan: You talked about taking pathology classes and auditing those courses. Is that typical? If you're studying cancer biology, would people typically take a pathology course? Or what did that give you?

Senthil Muthuswamy: Great question. I don't think most cancer biology graduate students today take pathology courses. I'm not aware of any. When I go and give seminars in departments and other institutions and when I meet with the graduate students, I actually ask them to go do it because you learn a lot by doing it.

And I would say that the principles I learned in that course during my graduate program actually is what is shaping what I do today, the organoid idea. The idea of growing cells in 3D and the idea of trying to have culture conditions where the growing cells has a lot of resemblance to the way the patient tumor looks in the patient all came because of that sort of understanding I had before. Otherwise, you don't really have to worry about the cancer cells forming anything that looks like a mini tumor. We can just grow them on a flat monolayer culture and use them for all the studies we want, because it still allows us to ask a lot of very important, interesting questions.

But I think if those are the questions you're interested in, there is no need to create a mini tumor-like culture condition. But because I learned that, it biased my thinking to some extent, and that actually was really helpful for me in that sense. I'm actually glad I did this because it makes me think about whatever we do in the lab more in a physiological context.

During my PhD I was doing two things. One, we were using moss models to study cancer, and we were growing cells on a flat dish to study some of the basic mechanisms. When I moved into my postdoc, this was with John Brugge's lab here at the Cell Biology department, that's when I started asking questions like can we adapt three-dimensional culture methods to grow cells in a manner that it looks like a tumor-like, three-dimensional growth?

And there were a few people doing that at that time, maybe one or two, but the leader in that field was this woman Mina Bissell. She's a pioneering scientist in Lawrence Berkeley National Lab in California. She's still active today.

So we established a collaboration between John Brugge's lab and Mina's lab. I went to Mina's lab to learn how to grow cells in 3D, how do we use this platform to understand some of these things or study some of these things, brought it back, and then I started actually establishing a culture platform where we can grow cells. These were all done in breast cancer at that time. Breast epithelial cells, so it looks like a tissue structure that is resembling some aspects of human breast tissue. That's when we started thinking, OK, can we now use this to study some of the oncogenic processes? And we started asking questions.

Then we started getting insights that were not apparent in a flat, monolayer culture. For example, in a flat, monolayer culture where cells grow, they grow to come to confluence. They fill the plate and then they stop normally.

But in a three-dimensional culture, if you put them, they will form these little ball-like structures. And if you activate oncogene, they continue to grow and they will look like a mini tumor. So something about the three-dimensional allows them to do more things that we didn't see in a flat dish. That's what started my thinking into this field.

So I think it was an eye-opening experience at that time for us to see these morphology change, and then that led me to thinking what is so unique about 3D? What is happening there? What sort of biological principles can you learn?

And there are some biological principles like cell polarization and directionality that is really uniquely better-suited to study in a three-dimensional context than in a flat monolayer, if you will. And how the organ size, the structure size is maintained, what happens when you activate cancer-causing genes, how it loses the structure, loses its shape. Those are all the things that we start to see.

And today, you may have heard pathologists diagnose cancer by first looking to see is there a disturbance in cell and tissue structure? That's how they diagnose cancer. They look at a tissue, a slice, and they say, this doesn't look normal. So there's purely morphology-based diagnoses, and then all the subsequent things happen.

When you grow cells in 3D, you begin to see some of those features show up in a culture system. So that's when I think you realize that OK, you're getting into aspects of biology that is traditionally ignored by growing cells on a flat monolayer.

Brendan: Great. And so right now you're running a clinical trial to evaluate this organoid platform. Tell us a little bit about the trial and what you hope to learn.

Senthil Muthuswamy: So I think when I moved here, we can generate mini tumor organoids from a patient's tumor and grow them in the lab, and we can use them to test for drug response. But the big question then is is what you see in the laboratory indicative of what happens in the patient? And if so, and if you truly believe that you are able to maintain the tumors closer to the patient, can you use it as a tool to predict or help patients get a better treatment? That's the big holy grail question that we are interested in.

And I think the reason this becomes a feasible question today is if you backtrack a little bit and think for a minute, if a patient walks in with a bacterial infection, and if you don't know anything about what the infection is, the way it is done today, and it's been done like this for decades, is that you take a swab and you send it to the lab, they grow it in a Petri dish, and then the bacteria grows into a lawn and they'll put small disks of 5 different, 10 different antibiotics and then see which antibiotic kills the bacteria better.

Once that is known, they relay that information back to the physician. The physician then prescribes that antibiotic to the patient to control the bacteria infection. That's typically the best that may happen even today, right?

Why can't you do that for cancer? The simple reason is that you cannot do the same thing that we do in a bacteria. You cannot take a piece of tumor, expand it in the lab, and then test the same way. And I think this is where organoid platform comes in, because since we have now crossed that bridge of being able to take a patient tumor and grow them in a lab as a mini tumor, now it creates a dish with a lot of tumors. Can you now use them as a screening platform?

So the HOPE trial that we are doing, which is called Harnessing Organoids as a Potential for Treatment, is really designed to test two things. One is simply ask, is this feasible? Can you take a patient tumor, can you grow it in a lab, can you test them for the drugs, and can this all be done in a timely manner that will be beneficial for the patient? So that's the first goal, the major goal.

The minor goal of the trial is that if you do accomplish that first goal and if you begin to get some responses in the culture, is that response indicative of what the patient response is? Those are the two goals. I think we achieved both these goals.

We know it is feasible and we know the criterias that we need to do to make it more effective in terms of being able to establish organoids and test the responses in a timely manner. And the initial results for comparing the response in culture and the response in a patient also sounds extremely promising for us. It looks like we know how to model the drug treatment platform in a culture dish and how to interpret the results so that we can predict what may happen in the clinic.

So those two look like we are able to achieve those goals, and I think that's the biggest step for us. And if you ask, what does it really mean in the clinical setting? I think there are two potential outcomes.

One is maybe this will help us find the best possible treatment for that patient. So let's say you have a patient tumor growing and you want to know among these 10 different drugs that you can give to the patient which one works best. So we can now screen and pick out the one that works best. That is one possibility.

The second possibility we can do is if we know among the 10 drugs 4 drugs do not work at all and six of them work in different degrees of effectiveness, we know for sure we can avoid those 4 drugs for the patient. So avoiding drugs that will not work is also a big help for the patient because it minimizes side effects and toxicities. So preventing toxicities and side effects is actually a big problem for many of the patients, so that also is a second advantage of using this approach.

Brendan: Right. Yeah, I was going to ask you about that. Maybe just briefly tell us a little bit about how, when somebody is diagnosed with cancer and they start a treatment regimen, the drugs are chosen now and how this platform could improve that process.

Senthil Muthuswamy: So how drugs are chosen, it may be very complex. And since I'm not a medical oncologist, I won't be able to give a precise answer, but I'll give you a broad answer.

Brendan: Yeah.

Senthil Muthuswamy: Usually it is chosen from past record of how those particular cancers are treated from past clinical trials, the drug effectiveness, the medical oncologist's knowledge on that topic. Some of the drugs are chosen by their molecular signatures. So the simplest example I can give you would be if a breast cancer patient is known to have an estrogen receptor-positive, HER2-positive disease, then the likely choice of drug would be if it is a very early disease, something that interferes with the estrogen receptor function. Tamoxifen, a Fulvestrant, or any of those drugs.

But if it is clearly a HER2-positive disease, then you would give a HER2 inhibitor, whether it is an antibody or a small molecule inhibitor. And that is what the clinical trials have concluded and shown, and that is the standard practice that people use. There are guidelines that are published for medical oncology folks that they follow and give.

The real problem comes when the patient's tumor relapses after two or three times and all the existing guidelines have been exhausted for meeting the criteria what drugs to give. So the initial choices can be based on age, the type of tumor, the stage of the tumor, pathological criterias, molecular criterias, the patient's health. All those things are taken into account when they make the choice.

But when the patient now has an advanced metastatic disease and the tumor has now become resistant to all the treatments, at that point there is no real logic to choosing anything, so that's where I think it becomes really a hodge-podge game of saying let's try this, let's try that. Let's try this, let's try that because there is no precedent to know exactly how to treat that disease. And different patients' tumors are extremely different in that way because they may have had different evolutionary paths before they get to there. So the oncologists usually are at loss into making the most effective possible combination at that time.

So perhaps organoids can help at all levels, especially in cases where the treatment and the response is not very strong. Then you can help even in the early cases. An example of that is a pancreatic cancer or even head and neck cancer where the treatment choices are usually limited. There are only a few drugs they give even for a patient who's diagnosed with an early disease. And there, perhaps organoids can help in choosing better drugs and more drugs.

But that logic will not be very effective for a breast cancer patient with an early disease because the current way they treat breast cancer prolongs survival [inaudible] for 10 years. So you don't want introduce an organoid into a setting where you already know the treatment is pretty good for a long-term response. Whereas if the breast cancer patient has an advanced metastatic disease that has relapsed after all the treatments, there an organoid may be useful if we can achieve enough tissues to do that. In the pancreas cancer setting it will be useful in any cases, whether it is early stage or if the patient's relapsed after the initial treatment. We can still probably try to get an organoid, test them for different drugs, and then potentially use that information to match the best possible drug responses.

Brendan: Great. And so just to finish up, which phase of the trial are you in now?

Senthil Muthuswamy: We finished the first phase, which is the feasibility and initial testing. And so this is a simple correlation testing. We are not actually telling the oncologist to change any treatments because we are not approved to do that. So the phase II that we are in the process of putting together is really to get the proper FDA approval and then do the testing and use the testing results to inform the oncologists to consider changing the treatment on the basis of what the results tell us.

In that setting, then it's a choice the oncologist has to make, whether they would believe in the organoid testing to change what they think should be given into a different combination, or they wouldn't want to change. That's a choice they have to make. But at least the FDA approval and the trialist will be designed to inform the oncologists of what they can do.

And if the oncologists believe in it, they want to try it, they can try it. And then we can assess whether the response that is tailored based on the organoid treatment is going to actually help prolong the response for these patients. If that's the case, that actually will be fantastic because then it means you really are the sensitivity testing base result to change the way clinical practice is done.

Brendan: Dr. Muthuswamy, thank you very much for joining us. It was great to have this conversation with you.

Senthil Muthuswamy: Thank you. It was a true honor and pleasure to do this with you. Thank you, Brendan.

Brendan: Thank you for listening. If you've enjoyed this podcast, please rate us on iTunes and help us spread the word about the amazing research taking place across the Harvard community.

Oby: To learn more about the guests on this episode, visit our website, catalyst.harvard.edu/thinkresearch.