How The Parker Institute Addresses Tumor-Specific T Cells

July 17, 2019 | While Daniel Wells was in his post-doc at the University of California, Berkeley, a friend of his recommended he meet with Nikesh Kotecha, VP of Informatics at the not yet launched Parker Institute for Cancer Immunotherapy, to discuss his interest in the immune system and data science.

"I started talking with Nikesh about his vision for the team, and this belief that an integrative data strategy, where you're bringing multiple data types, clinical data as well as many different forms of molecular data, together with next generation machine learning, is really going to be a game changer, a transformative tool for advancing and accelerating immunotherapy," says Wells. "That vision was extremely compelling to me."

Now, as a Senior Data Scientist at the Institute, Wells sees how discoveries in immunotherapy are rapidly changing how researchers develop treatment strategies for patients by integrating multiple molecular and clinical data types.

On behalf of Bio-IT World, Mary Ann Brown spoke with Wells about the Parker Institute for Cancer Immunotherapy, his recent work addressing tumor-specific T cells, and what analytics need to be pursued for the sake of data.

Editor's note: Brown, Executive Directors of Conferences at Cambridge Healthtech Institute, is planning a track dedicated to Informatics for Cancer Immunotherapies at the upcoming Immuno-Oncology Summit in Boston, August 5-9. Wells will be speaking on the program. Their conversation has been edited for length and clarity.

Bio-IT World: The Parker Institute for Cancer Immunotherapy was launched in 2016. Can you tell me just a little bit about the institute?

Daniel Wells: We were launched in April 2016. Our founder is tech entrepreneur Sean Parker, who was the first president of Facebook, and has since been an entrepreneur on many different start-ups such as Airplay, Spotify, and others. Sean, for many years has had a very deep interest in the immune system and how it works. The story that I've heard is many years ago, he was looking to do something big, and really wanted to build an institute that was focused on understanding the immune system. At that time, this was just around the time of the initial approvals for pembrolizumab. Sean saw a real opportunity to finally take cancer immunotherapy, which is something that people have wanted to do for years. It goes back to Coley in the early 1900s.

He saw a chance, that maybe this is the time for immunotherapy to really happen, but he felt that it needed a boost, and really, it needed some kind of central place where leading academic researchers could partner with pharma, could partner with nonprofits together and have, as he would say, a sandbox to play together in, and really pursue the boldest research that is maybe trying to break down some boundaries and break out of the traditional ways of thinking.

So that was my understanding of why Sean decided to found the Parker Institute, and pretty early on he met up with our now CEO Jeff Bluestone, who is a professor at UCSF in immunology, and has also worked for a long time building institutions that bring researchers together to unite around common goals that are larger than one lab, so Jeff founded something called the Immuno-Tolerance Network (ITN). It was the same idea—How can we help researchers come together and be more than they could be apart? That's really what the Parker is about, is collaboration.

You have a paper recently published, "Clonal Replacement of Tumor-specific T cells Following the PD-1 Blockade" (DOI: https://doi.org/10.1101/648899). This addresses tumor-specific T cells and how they're immobilized following a checkpoint blockade. Could you share some of these findings?

Absolutely! I'm really excited about this paper.

So, I think there's a core question in the field that many people have tried to address, which is, how does anti-PD-1 therapy work? And in particular, is PD-1 reinvigorating T cells that are already present in the tumor? Or is it enabling peripheral recruitment of T cells to the tumor? And those T cells may not have been previously engaged in tumor killing.

In this paper, we were very lucky to work with several great researchers at Stanford, including Anne Chang, Howard Chang, and Ansuman Satpathy, who are really leaders in immunogenomics, and in particular, the single cell sequencing. What we decided to do here was on a trial being run at Stanford using PD-1 to treat basal cell carcinoma. We were able to actually get pre- and post-treatment biopsies of these patients and do a single cell dissociation there, so we actually could measure the different immune cells present in the tumor microenvironment pre and post.

The way we decided to assay these cells is using an assay from 10X Genomics that allows us to measure, not only the RNA content of each immune cell present in the tumor microenvironment but also to capture the TCR, the T cell receptor. With that, we were able to ask a very simple question: Are the cells that we know are tumor-specific by their particular cell subset, CD-3, CD-8, CD-38, high T cells, were the cells that came in after PD-1 with that particular subset, present before? The answer turned out to be not that much. We saw that many of these so-called, tumor-specific T cells, were in fact, coming from outside of the tumor.

This is pretty important because it's indicating that in fact yes, when you treat with PD-1, there's very strong peripheral recruitment of T cells, and it's not just about reinvigorating the T cells that are there. It could be that those T cells are nonspecific or not specific to the tumor. They're viral or they're autoimmune associated, or they're just so completely exhausted, terminally exhausted and unable to be revived.

With that study now, I think it helps support further study into how we enhance peripheral recruitment of T cells and other important immune markers to the microenvironment. That will hopefully lead to enhanced tumor killing and ultimately improve patient response.

Traditionally, the complex data from cancer informatics and immune response have been siloed. And now with the emergence of cancer immunotherapies, how are these silos being broken down?

I think this is a great question, and a really important question. As you eluded to, with immunotherapy, and really trying to design rational combination immunotherapies, it's absolutely paramount that we are able to look across multiple modalities, to be able to understand how individual patients are responding to treatment. We have to look at the immune compartment, we have to look at what the tumor's doing and the microenvironment and all of these different things, so it's absolutely important that we have these data integrated together.

There's a lot of solutions out there that people are building. We're actually building one in-house right now, and this tool is actually a very large triplet data store. It's a very particular type of database that stores triplets. We've been working very hard for quite a while to be able to store all the different types of molecular data we see next to a very refined schema for clinical data collection, as well. When we are doing these translational analyses, we ultimately have all the data physically living together in the same place. This has been a transformative tool for our team because we no longer have to run around and talk to the sequencing lead and the flow cytometry lead and the CTR lead, etc. Every day, the scientist on our team is empowered to ask their own questions on the particular trial that they're working on. It's been absolutely amazing. I think having an integrated data platform where everything can live together and be queried together has been really great.

Yes, I agree with that one completely. Where do we go from here? What technologies need to be developed to provide more data? Or alternatively, if we have enough data, what analytics need to be pursued?

I think that that's a great question, and there're a couple of things I'll say. So, on one hand, I do think that we're still working on developing the right assay technology, and there are a couple that I'll highlight. We all know that the tumor microenvironment is this extremely rich, complex structure that is unique for most patients. I think technologies that allow us to assay that really deeply—I'm thinking particularly of high-dimensional imaging technology, such as multiplex ion beam imaging or codex or others—are going to be really important as they begin to scale because they're going to allow us to really understand which tumor cells are present, what they are doing, what immune cells are present, what they're doing, and how they are talking or connecting with each other. I'm really excited to see those technologies continue to grow and be adopted in the field more broadly.

The other thing I'll say, is that I think that there's a growing understanding that collecting particular data sets with an eye towards using those data sets with machine learning and advanced statistics is a very powerful strategy for drug discovery and rational combination identifications. So for example, I'm thinking of the recently signed deal between GlaxoSmithKline and 23andMe, which Parker Institute is not a part of, but this deal is really beautiful to me because you have a company here that's collected genetic information on millions of people, and then you have a company with a very long history of developing best in class drugs for a wide range of different conditions, and as we're able to develop the analytics to leverage that data, to identify the patients who have particular conditions that may be associated to response to checkpoint inhibitors and what's their genetic makeup. That's extremely powerful and so I think we're going to see a rise in companies, and institutes and labs, building large data resources where machine learning can really be applied to identify these kinds of targets.