Behind the clinical trial: Advancing CAR T-cell therapies for kids with fatal brain cancer
A pediatric oncologist and computational immunologist at Stanford University combined their skills to use single cell immune profiling to answer pressing scientific questions about an incurable pediatric cancer—diffuse intrinsic pontine glioma (DIPG).
Sneha Ramakrishna, MD, came to Stanford University after completing her pediatric hematology/oncology fellowship at the National Cancer Institute/National Institute of Health to access the “incredible multiomics set of platforms that researchers can use to understand the drivers of immunotherapy activity or failure in patients.” She worked on one of the first iterations of CAR T therapy as a medical student, and after seeing its powerful potential, she dedicated her career to immunotherapy development and treatment.
Scientific sparks flew when she crossed paths with Zinaida Good, PhD, who came to Stanford University to attend a program in computational and systems immunology after working as an oncology researcher at Genentech. Good was the first person Ramakrishna met in the lab of Crystal Mackall, MD, where the two worked together on a clinical trial using CAR T therapy to treat DIPG. Ramarkrishna recognized Good as an “incredible force.” And Good admired Ramarkrishna’s enthusiasm and dedication to patients.
The two combined their complementary skill sets to apply Chromium Single Cell Immune Profiling to analyze cerebrospinal fluid (CSF) samples from pediatric patients with DIPG enrolled in the clinical trial to get a clearer picture of the effects CAR T therapy had on patient tumors in collaboration with the Parker Institute for Cancer Immunotherapy and 10x Genomics (1). We talked with Good and Ramakrishna about the results of their study, recently published in Nature, and how Chromium Single Cell Immune Profiling—and eventually Visium Spatial Gene Expression and Xenium In Situ—helped them address questions about a devastating cancer with no available treatments.
Why is it important to develop new therapies for pediatric gliomas?
Ramakrishna: DIPG is every pediatric oncologist’s worst nightmare. There's literally nothing we can do. There have been hundreds of clinical trials with chemotherapy agents and small molecules with limited improvement in either quality or quantity of time for these patients. The opportunity to change the paradigm with a new treatment strategy was incredible. When I saw the potential that immunotherapy was something that could really treat this tumor, I wanted to get it to the patients as quickly as possible. If it even moves the needle a little bit, that is a huge step for this particular disease.
What was the first result of this study that excited you?
Ramakrishna: The first thing that struck us was that patients were regaining neurologic function. DIPG is a really interesting tumor—it doesn’t ruin the neural networks, it just infiltrates them. We didn’t know if the neural networks would function normally once you got rid of the tumor. Would the patients lose nerve function and maintain that loss forever? The first thing that struck us was that the patients were regaining neurologic function. Our third patient in the trial had significant difficulty walking and left-sided weakness, which made it difficult to smile or open his mouth. But as the CAR T-cell therapy started working, he regained function. From a clinical standpoint, this was clear evidence that not only was the treatment working, but it significantly improved quality of life. We had pediatric patients who were unable to stand, walk, or do anything who became ambulatory, were able to eat, and gained significant weight and height in a matter of months. The quality of life improvement has been the most inspirational part of this trial.
Why did you incorporate single cell immune profiling into your study?
Ramakrishna: When I first set up the correlation matrix for this study, I reached out to a number of individuals who were doing brain tumor immunotherapies and asked them, “Can you capture cellular populations from patient CSF samples?” Very consistently across these individuals, they said there were not enough cells in the CSF to capture for analysis. I started to explore more sensitive techniques we could harness and use to look at the cellular makeup of the CSF we collected from these patients. We turned to single cell RNA sequencing based on Zinaida’s prior work and optimization done by the Cancer Correlative Science Unit here, at Stanford University, for other cancer studies.
Good: From a computational point of view, there are no comparable platforms. We consistently obtain massive amounts of high-quality data from very limited samples. With these data, we can trace clones of CAR T cells through time using endogenous T-cell receptor sequences, which every CAR T cell inherits from its parent T cell. This approach enables us to identify clinically useful infusion CAR T cells because, after a while, we don’t see the majority of the CAR T cells we started with—only a few persist through time. A completed dataset is just an amazing resource that we and other researchers in the field can go back to for decades.
How do you collect samples for immune profiling analysis?
Ramakrishna: We literally take CSF samples from the bedside, walk them to the lab, and process them fresh. If you freeze down the sample, there are not enough cells to do anything with after thawing, so fresh processing was a critical step to sort of obtaining immune profiling data. Because this CAR T-cell therapy is clinically active, we had a lot of cells that we could capture at clinically distinct time points. For example, if a patient had evidence of clinical inflammation, we would check the intracranial pressure and then I would say, “Okay let's take that CSF and do a single cell RNA sequencing analysis.” As we treated even the first four patients, we realized that we very consistently can capture the cellular populations over time, and that the capture of those cellular populations is dynamic over time. Having the various time points gave us insight not only into what CAR T cells look like before we give them to patients, but what they're doing in the CSF as they are killing tumor cells.
Where does the clinical trial go from here?
Ramakrishna: We are in the process of writing another paper that demonstrates clear evidence of continued therapeutic activity in a subset of patients. But that’s only a subset of patients. The goal is to figure out how to make that consistent across our patients. Correlative studies like ours are needed to figure out how to optimize treatment and really improve patient outcomes. This is the first time we have ever moved the needle in treating DIPG. Now that we’ve moved it, let’s push really hard. Let’s figure out how we can get to a—I’m going to say it out loud—a curative option. Our patients have already benefited from this. Our goal is to continue to improve upon this treatment by understanding the biological mechanisms using single cell immune profiling.
Good: We are getting an amazing amount of sequencing data from this first patient cohort that showed great responses. Once we have all the data, we will construct a unified CAR T-cell atlas that will provide insights into the mechanisms of response and resistance. It may become the biggest CAR T-cell atlas constructed yet. We’re now doing all of these analyses we’ve been dreaming about on the first half of the data.
Could the results of your single cell immune profiling studies impact designing T-cell therapies in the future?
Good: First, we want to know which gene expression programs and transcription factor networks help a particular CAR T-cell clonal subset become clinically dominant over time. Then, we can engineer cells with this profile in the lab and test them in various assays and animal models, and ultimately test them in a clinical trial.
Ramakrishna: I think Zinaida said it beautifully. The goal is twofold. One is to be able to identify patients who will benefit from this treatment because this is a universally lethal tumor. If the therapy is not going to help them, clinicians want to know in advance. But subsequently, we want to make sure we’re giving patients the best product possible. I’ve been thinking about the diversity of CAR T cells recently, and, as Zinaida said, there is only a small subset that persists, and we still don’t know which cell populations provide a durable response. As we characterize this further, it will be helpful to know what those cell populations are and, then, understand if manipulating the manufacturer product upfront would be helpful. Or would having a more polyclonal and complex product be more advantageous? These are the types of questions that are open ended and our dataset will, one day, be able to answer. We’re really excited about it.
We know you are planning to incorporate spatial transcriptomics and in situ into your workflow. How will this impact the course of your study?
Good: We are very excited that the 10x team has included spatial and in situ technologies into our pilot project. The first technology that we will apply is Xenium, which will hopefully be available at Stanford within a month or two. With Xenium, we will be able to interrogate 300–400 genes in the DIPG tumor biopsies from our patient cohort. We also look forward to employing Visium HD, which captures the highest resolution picture of the whole transcriptome compared to other competitor technologies.
Ramakrishna: Exactly. This technology will provide insight into immune cell interactions within the tumor microenvironment. The tumor microenvironment and the spatial orientation of immunosuppressive populations with CAR T cells and tumor cells will be super important to understand treatment response. I think an important question to ask is whether the myeloid or regulatory T cells that we capture in the CSF are reflective of the cell populations in the tumor itself. And are there things that changed within the tumor before and after treatment? These tumor tissue samples are limited and rare because DIPG is located in the brainstem, a very dangerous location to access for tissue acquisition. It’s an incredible privilege to have these patients and families allow us to learn from them and with them.
This interview has been edited for length and clarity.
Discover how Chromium Single Cell Immune Profiling can transform your translational, immuno-oncology research →
- Majzner RG, et al. GD2-CAR T cell therapy for H3K27M-mutated diffuse midline gliomas. Nature 603: 934–941 (2022). doi: 10.1038/s41586-022-04489-4