From Tumor to Treatment: How PDX Models are Powering Cancer Breakthroughs

Julia Schüler:
So the PDX based in vitro assays, they can support the full spectrum of drug development from target discovery over hit ID to efficacy testing. They can support the in vivo model selection afterwards for preclinical pharmacology, but can also be performed in parallel to an in vivo study, for example, for the development of a translational biomarker. And what is nice about this is that the use of the PDX sourcing material increases then the productivity in depth of the assay and thereby improves the drug candidate selection.

Mary Parker (00:03):
I am Mary Parker, and welcome to this episode of Sounds of Science. I’m joined by Julia Schuler, Research Director, Therapeutic Area Lead, Oncology at Charles River. Today, we'll be diving into the world of patient-derived xenograft models, or PDX models, a cutting-edge cancer research tool helping bridge the gap between preclinical studies and clinical trials. These models provide more accurate insights into how treatments will perform in patients. Join us as we explore how PDX models transform drug development, improve treatment outcomes, and shape personalized medicine’s future. Welcome back, Julia. We’re happy to have you back on the show. Can you tell us a little bit about your role at Charles River?

Julia (01:02):
Yeah, of course, Mary. Thanks a lot. I'm happy to be here. I work out of Charles River Germany, based in Freiburg. After my PhD in veterinary medicine, I completed a postdoc at the Max Splunk Institute in Immunobiology and then went back to oncology. I had several leading positions at OncoTest in Freiburg, which is a small SME offering drug development services for oncology research. In 2015, we've been acquired by Charles River, and our services are now integrated into the discovery portfolio. As a tumor biologist, my expertise lies in preclinical oncology drug testing platforms. I have the pleasure and the honor to lead a multidisciplinary team with the aim to expand our current portfolio of PDX models and also implement new technologies such as small animal imaging, multiplex omics analysis, and image analysis workflows. And within our scientific team, I bring the oncology expertise and represent Charles River in multiple international consortia focusing on different aspects of the disease.

Mary (02:09):
So I think it's fair to say that your oncology experience is vast. You're the perfect person to talk to about this particular topic. Before we get into the nitty gritty, I'm just curious, what are some of the coolest advances you've seen in the industry, whether in research or in business models since you've started studying?

Julia (02:31):
I would say the development of humanized mice is the most interesting new discovery in this field. So the possibility not only to have human tumor cells into immune compromised mouse , but also the human immune cells interacting with them. This is really driving new developments and making these models even closer to the clinic is really fascinating.

Mary (02:58):
I love that. So, getting into patient-derived xenograft or PDX, they're a type of preclinical model, I guess, used in cancer research. So how are they used in drug development, and how do PDX models help in simulating human cancers in preclinical trials?

Julia (03:17):
Most of the time PDX are used in later preclinical stage, for example, to selecting the best suitable drug candidate or also an optimal patient cohort for a later clinical phase. The reason for this is that the PDX maintain a lot of the characteristics from the tumor they originate from and then thereby increasing their translational value.

Mary (03:39):
So basically you can take these xenografts from the patient and use them to see what treatment will work for their specific cancer.

Julia (03:49):
Yeah, exactly. So the original thought of a PDX would be a precision medicine one so that you have a patient in a mouse, but due to several reasons, that's not an optimal option for the patient. But the models developed represents still the patient. Right. So if you have a large enough collection, you have a patient cohort.

Mary (04:11):
So what makes PDX models more reliable than traditional cell line-based models in preclinical testing?

Julia (04:18):
So based on the method, they are developed, never seen plastic. The clonal selection is avoided. That is taking place when you establish a cell line. So the PDX will represent they preserve the histology and also the heterogeneity of the donor tissue and provided that the panel of the models has a certain size, they largely represent the molecular landscape of a respective disease.

Mary (04:42):
So I know that in cancer, the tumor microenvironment can play a really big role in determining whether a drug is going to work or not. So how do PDX models maintain the original tumor architecture and what advantages does this provide?

Julia (04:59):
So this three dimensional orientation and the phenotype of the cells seems to be a model imminent feature that is preserved during the process of the model establishment from donor tissue to model. And there are different sub clones of the tumor cells that coexist in one tumor as well as the stromal compartment that is represented then by mouse fibroblasts, but mimic in percentage and also in shape exactly the human homolog from the donor patient tissue.

Mary (05:29):
And that's not really something that you can do in a Petri dish or in vitro type model because you don't have the biological structure to make the tumor more similar to what it would be in a human.

Julia (05:41):
Yeah, exactly. And for some reason you don't see that this cells are self-organizing themselves in a way they would do in a 3D dimension in an animal.

Mary (05:55):
Yeah. So when we're thinking about the advantages of PDX models in drug testing and the advancements in vitro, how can you explain the role of in vitro PDX assays in characterizing oncology therapies and how do they compliment in vivo studies? Not like they're useless, but bringing them together seems like it would be the most efficient way to get a full model.

Julia (06:19):
Yeah, absolutely. So the PDX based in vitro assays, they can support the full spectrum of drug development from target discovery over hit ID to efficacy testing. They can support the in vivo model selection afterwards for preclinical pharmacology, but can also be performed in parallel to an in vivo study, for example, for the development of a translational biomarker. And what is nice about this is that the use of the PDX sourcing material increases then the productivity in depth of the assay and thereby improves the drug candidate selection.

Mary (06:57):
Can you think of any examples off the top of your head of specific types of cancer and A PDX in vivo and in vitro model that has been used to model it really accurately?

Julia (07:11):
Most of them come from solid cancer, and this is very well proven, for example, for colorectal cancer or for non-small cell lung cancer that there is the possibility to have a sort of biobank of PDX, but as well as the same biobank in vivo assay. And they perform very well with regard to predict sensitivity in a clinical cohort, but also to the patient they came from.

Mary (07:40):
Okay, that makes sense. So what are the benefits of integrating ex vivo 3D tumor models with traditional 2D assays and how does this integration enhance drug efficacy studies?

Julia (07:55):
So by complimenting ex vivo 3D tumor models, with the 2D assays, we can leverage the speed and scale of a 2D system while benefiting from the higher biological relevance of the 3D systems. And the combined approach then improves the predictive power of the drug efficacy studies at that stage, and then in the end increasing the chances of a clinical success for the oncology therapeutics we are investigating.

Mary (08:21):
And I guess if you were using it for a sort of personalized medicine, time would be of the essence, you'd want to figure out what treatment works best for the patient as soon as possible. Since cancer obviously gets worse the longer you leave it alone.

Julia (08:33):
Yeah, exactly. So the ex vivo part is the one platform that is suitable for precision medicine in contrast to the in vivo part, which simply takes too long.

Mary (08:42):
So to ensure that the data generated from PDX models is trustworthy and meaningful, obviously we need to maintain high standards throughout the process. So how does Charles River ensure the accuracy and reliability of data generated from PDX models?

Julia (09:01):
So the most important phase is the establishment phase of the models. And all our models undergo a thorough quality control process. In the first passages, we take histological samples and compare them to the donor patient tissue in each passage. In addition, we run molecular tests to determine the percentage of human versus mouse tissue in the tumor model. And this ratio is very specific for each model and remains stable across all passages. So once established, we then do a whole-exome sequencing alongside with RNA sequencing to characterize the models more in detail, but also to compare them to a patient cohort in clinical phase. And if we add these data altogether in the database, we can then have the possibility to compare the new models with our already established ones and see whether they make a good fit in our database and if they reflect the disease we think they should represent.

Mary (10:02):
Okay, that sounds like a giant pile of data. So, how does Charles River use machine learning and live cell imaging to analyze data from PDX assays?

Julia (10:15):
Currently, we are focusing more on live cell imaging technologies in our PDX-based assays. We apply this for our in vitro assays, which we can run in 2D as well as in 3D. And the importance of live cell imaging comes into play when we add immune cells to the mix to test oncology drugs that need to involve the human immune system to tackle the cancer cells. And for these core cultures, the image analysis tool are super helpful as they can measure features that are beyond a simply tumor cell killing assay. And so we can determine in which timeframe and in what kind of spatial relationship the immune cells are attracted to the tumor cells and then kill them. And we are able to quantify these movements and also the time in which the immune cells are attracted towards the tumor cells.

Mary (11:11):
This might be a kind of off the wall unanswerable question, but on average, what do you think is the percentage of the treatment of killing a cancer that comes from a drug versus comes from an immune system? So I know that they can help each other, but is it more efficacious to get the immune system to do most of the work or do you mostly rely on a drug

Julia (11:37):
As always, it depends. I would say currently it’s still the drug that kills the tumor cell directly. All the targeted drugs, the cytotoxic drugs we have established in the last 50 years, and for the ones that enabled the immune system to kill the tumor cells, we just started with 10 years ago. So I think we did make some breakthroughs, but I still would rely on drugs that target directly the tumor cell.

Mary (12:08):
Yeah, it might help prevent it from coming back, though, if the immune system has been trained. Yeah, that makes sense.

Julia (12:13):
Yeah, exactly.

Mary (12:14):
Are PDX models customizable for specific research needs, and how does this flexibility benefit different therapeutic areas?

Julia (12:26):
We can adapt PDX models to tailor to specific scientific or therapeutic questions, for example, with new technologies. And what’s important is that their ability to closely replicate a human tumor genetically, biologically, and also pharmacologically. And this basic characteristic can then be used to model other diseases and also to go into detail like resistance mechanisms or rare diseases that might not be the main focus of cancer research, but go into the pre-cancer stages, or something that is super rare and driven by only one genetic mutation. And for these, we can also develop and use PDX.

Mary (13:13):
So, would PDX models be helpful in developing cancer screening methods as well? Is that kind of what you’re saying?

Julia (13:24):
Yeah, for example, so if you would develop a PDX panel for a specific disease from precancerous lesion until late stage, full-blown disease, that’s possible. And then you can use that as a, for example, biomarker platform.

Mary (13:39):
Okay, that's great. Because obviously the sooner the better.

Julia (13:42):
Yeah, yeah.

Mary (13:44):
So PDX models are known for closely mimicking the human tumor environment, offering researchers a better understanding of how therapies might perform in clinical settings. Can you share examples of where PDX in vitro assays have led to significant advancements in oncology drug discovery?

Julia (14:03):
There's one example among many other papers which stood out from my perspective where it was published a couple of years ago by a scientist group across the US and Europe, and this says a lot because you need a lot of labs for that. And they developed a workflow for personalized in vitro and in vivo model pipeline to guide precision medicine. So, for a specific patient and what the ultimate result of that paper was that you need the in vitro and in vivo model plus the genomic data of the patient tissue to make the best prediction, which compound might help that individual patient best. So only in combining these three analysis platforms, you get the best answer for a specific patient.

Mary (14:53):
It sounds like it would also help with dealing with the problem of patient diversity where maybe you have a clinical trial run in one country with a kind of homogenous group of people, but you want to roll it out into other areas because obviously we know genetic factors can contribute to a patient's response to a drug. So this sounds like it would help with that.

Julia (15:14):
Yeah, absolutely. And what I also liked at that paper is that the labs were spread across the world so that also epigenetic influences are covered by those both different tissues coming from different countries and different environments.

Mary (15:30):
So kind of in that same vein, how do PDX models impact the design of clinical trials or the likelihood of success?

Julia (15:38):
So drug testing in PDX models, specifically those executed in the format that we call single mouse trials, deliver data sets that have a high translational value. So by testing different drugs across a broad range of tumor types or also molecular subtypes, it is possible to investigate a large number of compounds in a very large diverse tumor population that mimic a patient cohort. So you might be able to identify patient cohorts in preclinical stage, which weren't necessarily your main target when you designed the drug. And with these kind of screening formats, we can then explore one histotype or screen a variety of range of histotypes that are based on one molecular characteristic, like a shared specific mutation. And with this, it's possible to identify or expand a clinical indication and thereby shaping than the design of the clinical trial.

Mary (16:34):
Alright, that's excellent. So what are some of the challenges researchers face when using PDX based in vitro assays?

Julia (16:53):
The goal of these assays is to mimic critical features of the tumor microenvironment, the tumor, the stroma, the immune cells, and there are spatial dependency in a 3D architecture. And this then leads to a certain complexity in the setup of these assays. And then in combination with the heterogeneity, that’s the main feature of A PDX. The major challenge is to develop something that is robust and scalable for a drug screening application because time is of the essence. So at Charles River, we have thoroughly investigated the different parameters that influence that robustness and reproducibility. And we are now using standardized culture condition for our tumor lines, and also used for the healthy tissue part for the immune cells and also for the fibroblast, we are using well characterized donors that guarantee us a certain robustness and reputability from assay to assay.

Mary (17:56):
And does that kind of standardization also help with regulatory acceptance of whatever drug is being put forward?

Julia (18:04):
Yeah, absolutely.

Mary (18:05):
That makes sense.

Julia (18:05):
Absolutely.

Mary (18:06):
As we continue to see significant advancements in preclinical research, PDX models have become an invaluable tool for oncology providing more accurate insights into drug efficacy and patient response. So looking ahead, how do you envision the role of PDX models evolving in oncology research, especially with advancements in personalized medicine and targeted therapies?

Julia (18:36):
So with the new possibilities around lab automation and more specific readouts enabled, for example, by image analysis, the PDX models will be increasingly important source for ex vivo assays that will then expand in earlier phases of the drug development, for example, even for target identification. And with that, we can then improve further our conversion rate of the oncology drugs by improving the very early stages of drug development.

Mary (19:02):
You mentioned image analysis. I imagine that’s an area where machine learning can be extremely useful. It's very good at analyzing images.

Julia (19:10):
Yeah, absolutely. This is not possible in a non-automated way.

Mary (19:16):
So for researchers and companies looking to implement PDX models into their drug development process, what are the first steps they should take?

Julia (19:26):
So they should carefully evaluate where in the process to best integrate these models as no model is a perfect fit throughout the complete drug development workflow. And once they have identified the scientific question they want to tackle, then it’s easier to identify the best-suited PDX model.

Mary (19:43):
So what advice would you give those looking to integrate PDX models into their pipeline?

Julia (19:49):
The most important part is that the PDX models of choice have to withstand the quality criteria we have talked about earlier today. And for the optimal model selection, a good characterization is key. So However, sometimes a wider screening effort across models that would not necessarily fit exactly the intended patient cohort can also lead to new insights that might be then helpful when designing a clinical trial.

Mary (20:13):
All right. Julia, thank you so much for being part of the show and sharing your expertise with us today. It's been a pleasure having you.

Julia (20:21):
Thanks a lot for having me, and it was my pleasure to talk to you.

Mary (20:25):
Julia Schuler, Research Director and Therapeutic Area Lead Oncology at Charles River. Stay tuned for the next episode of Sounds of Science. Until then, you can subscribe to Sounds of Science on Apple Podcasts, Spotify, Stitcher, or wherever you get your podcasts. Thanks for listening.