publication date: May. 1, 2020
Conversation with The Cancer Letter
NSCLC patients getting checkpoint inhibitors more likely to develop pneumonitis, Syapse-FDA-Aurora study finds
Thomas D. Brown, MD, MBA
Chief medical officer,
Founder and president,
This story is part of The Cancer Letter’s ongoing coverage of COVID-19’s impact on oncology. A full list of our coverage, as well as the latest meeting cancellations, is available here.
Patients with lung cancer and a history of pneumonitis are more likely to develop treatment-associated pneumonitis later, especially in the course of receiving immune checkpoint inhibitor therapy, according to a new study by Syapse, FDA, and Advocate Aurora Health.
The study, which relies on real-world data, examines the frequency of treatment-associated pneumonitis in patients with advanced non-small cell lung cancer who were treated with immunotherapy or chemotherapy. An abstract was presented April 27 in the Clinical Plenary Session at the American Association for Cancer Research’s virtual annual meeting.
In the COVID-19 era, real-world data are playing an increasingly significant role in helping researchers and physicians understand the risks and benefits of their interventions, as well as how their clinical operations are being affected by broader trends.
The Syapse-FDA-Aurora findings are all the more important, because checkpoint inhibitors—e.g. ipilimumab, atezolizumab, durvalumab, nivolumab, and pembrolizumab—have become the standard of the care in the treatment of many cancers and disease subtypes, especially NSCLC.
Given that the COVID-19 pandemic will likely increase the percentage of the population with a history of pneumonia, studies designed to characterize the safety outcomes of these patients will be critical to informing future clinical practice.
“In neither datasets were there deaths that were attributed directly to the pneumonitis,” Thomas Brown, chief medical officer of Syapse, said to The Cancer Letter. “Obviously, pneumonitis can be a serious complication, but the prior history of pneumonitis should not automatically exclude a patient from consideration of receiving immune checkpoint inhibitor therapy.
“We’re intent on expanding the analysis very rapidly to include analyzing patients regarding their past history of pneumonia. This has always been of clinical interest, but increasingly so with the COVID-19 pandemic.”
The Syapse-FDA-Aurora project helps define the utility of real-world evidence in oncology. Using data generated in real time, RWE researchers are demonstrating that they can now rapidly inform the standard of care by eliminating the knowledge gap between outcomes in the real world vs. outcomes from clinical trials used to support drug approval.
“This is the first time that someone has done a comprehensive look at the prior medical history of pneumonitis, and looking at the impact of that prior medical history on post-treatment-associated pneumonitis for both therapeutic agents,” Jonathan Hirsch, founder and president of Syapse, said to The Cancer Letter.
“To be honest, it was the first time that we had really considered a question like this,” Hirsch said. “We were as eager as the FDA to look at the feasibility of an analysis like this in the real-world dataset. Together, we believed that there was room to provide practice-changing knowledge through an analysis like this.”
The study, which was completed within 2.5 months, compares data (N = 1,262) from Advocate Aurora Health in Milwaukee—derived from patients with advanced, stage III and IV NSCLC—with results from clinical trials (N = 6,491). The latter are pooled data from eight randomized aNSCLC trials comparing immune checkpoint inhibitor treatment, both with and without chemotherapy, to chemotherapy.
With conventional retrospective data collection practices, it would’ve taken up to two years to find the sites, wrap up contracting processes, find patients, and obtain and analyze the data, Hirsch said.
“If you look at initiating the research collaboration in mid-August 2019, we were able to have a very impactful project completed and submitted for late-breaking to AACR in mid-January,” Hirsch said. “That’s a pretty rapid turnaround for what turned out to be a high-impact project selected for a clinical plenary, which we were very happy about and honored by. We certainly were not expecting that at the time. I think the topic was highly relevant given the current global pandemic.”
The long-term safety profiles of checkpoint inhibitors are often not fully understood, because of speedy FDA approvals and equally swift uptake by physicians.
“The association between immune checkpoint inhibitor usage and pneumonitis has been reported in a range from 1% to 7% of patients with immune checkpoint inhibitor therapy,” Brown said. “We’re using the classical definition of pneumonitis, meaning non-infectious lung inflammation, and effectively seeing if that prior history of pneumonitis increases the risk of immune checkpoint inhibitor-associated pneumonitis.
“The answer in both the clinical trials and the real world dataset is ‘Yes.’”
A past history of receiving radiation therapy appears to be a crucial predisposing factor for the development of pneumonitis, Brown said.
“It turns out that, amongst the pneumonitis patients, whether you’re talking about a past history of pneumonitis or immune checkpoint inhibitor-associated pneumonitis, a majority of patients in both those groups had a prior history of radiation,” Brown said. “Radiation appears to be an important factor in the story.
“On review of the individual patients, when I say a majority of the patients received prior radiation therapy, in most of those cases, the radiation was felt to be the cause of the pneumonitis, i.e. radiation pneumonitis.
“There’s also a phenomenon called radiation recall that can be triggered by certain drugs. The interaction between the history of radiation therapy, immune checkpoint inhibitor therapy, and chemotherapies, all has to be further sorted out.”
Brown and Hirsch spoke with Matthew Ong, associate editor of The Cancer Letter.
Congratulations on being selected for the plenary at AACR. This is a milestone in characterizing treatment outcomes with real-world evidence, and benchmarking your findings against results from traditional clinical trials. Can you describe the significance of your project?
Thank you. Matt, you have touched on some of the key issues. The overarching issue was to examine an impactful question and compare both the evidence found in clinical trials with that in our real-world dataset.
The specific question is of importance. One way I like to look at this, stepping back a little bit, is that the FDA has been very successful in increasing the efficiency of new drug approvals. There are several paths, as you know, to getting a drug to market nowadays, especially in the cancer realm.
The challenge is that when drugs enter the market, there’s a greater burden on the post-approval period to further clarify safety issues. In particular, especially for those safety issues that play out over time, and also to further clarify the broader therapeutic applications of anticancer drugs.
In this case, we wanted to address, even prior to the pandemic coming into full form, an important topic, which is pneumonitis associated with therapy, with immune checkpoint inhibitors. That association has been well-recognized. We wanted to, firstly, examine that question overall, in the real-world dataset as compared to clinical trials, but then to also utilize the depth and breadth of the real-world dataset that we’re working with, to look at a subpopulation of patients who had a prior history of pneumonitis.
Now in this case, we’re using the classical definition of pneumonitis, meaning non-infectious lung inflammation, and effectively seeing if that prior history of pneumonitis increases the risk of immune checkpoint inhibitor-associated pneumonitis. The answer in both the clinical trials and the real-world dataset is “Yes.”
We were able to to tease out some issues from the real-world data that are of interest. One is, and this was addressed by the discussant during the AACR presentation, that in the real-world dataset we were able to examine patients who had any prior radiation therapy.
It turns out that, amongst the pneumonitis patients, whether you’re talking about a past history of pneumonitis or immune checkpoint inhibitor-associated pneumonitis, a majority of patients in both those groups had a prior history of radiation. Radiation appears to be an important factor in the story. Though this needs further study, we were able to identify this issue through the plumbing of the real-world data.
So, the risk for treatment-associated pneumonitis is higher, in general, for patients who receive checkpoint inhibitors, compared to chemotherapy alone, regardless of prior history of pneumonitis. But what’s new here is that patients with a past history of pneumonitis are at risk of developing it again?
Correct. The association between immune checkpoint inhibitor usage and pneumonitis has been reported in a range from 1% to 7% of patients with immune checkpoint inhibitor therapy.
The data shows in both the clinical trial set as well as the real-world dataset that there is a similar association with classical chemotherapy and pneumonitis, albeit at a lower level.
In addition to the novelty of the clinical trial and real-world comparison on the post-treatment pneumonitis, the novel factor about our work is that this is the first time that someone has done a comprehensive look at the prior medical history of pneumonitis, and looking at the impact of that prior medical history on post-treatment-associated pneumonitis for both therapeutic agents.
As Jon alluded to, during the AACR presentation, there was a discussion session and the discussant labeled this as the first recognition of that association, that is that a past history of pneumonitis increases the subsequent risk of immune checkpoint inhibitor associated pneumonitis.
How does past history of radiation therapy worsen the risk for recurrence of pneumonitis in patients undergoing immune checkpoint inhibitor therapy?
That’s another important feature of our real-world dataset. On review of the individual patients, when I say a majority of the patients received prior radiation therapy, in most of those cases, the radiation was felt to be the cause of the pneumonitis, i.e. radiation pneumonitis.
There’s also a phenomenon called radiation recall that can be triggered by certain drugs. The association between radiation and pneumonitis is well known. It’s something that has a variable time course. The interaction between the history of radiation therapy, immune checkpoint inhibitor therapy, and chemotherapies, all has to be further sorted out.
How much does the past history of radiation increase the risk of immune checkpoint inhibitor pneumonitis?
There’s been some study of this in the literature, but it’s a subject that we plan to further explore. The important elephant in the room, so to speak, is that this particular study focused on pneumonitis and did not include patients with pneumonia, with an infectious cause of lung inflammation.
One would assume that infectious causes of pneumonitis would likely also be associated with an elevated risk of immune checkpoint inhibitor-associated pneumonitis, although our study did not address this. That’s certainly a further analysis that we’re pursuing.
Did you first see the signal in your real-world datasets, ahead of FDA’s analyses of the results from clinical trials? How was the research question formulated?
This is a very interesting example of collaboration between us and the FDA. As you know, we signed our research collaboration with the FDA in August of 2019. Shortly thereafter, the Office of Clinical Pharmacology team members came to us and said that they had this question about pneumonitis in chemo versus checkpoint inhibitors with patients who had a prior medical history of pneumonitis.
They wanted to compare the population of what they’re seeing in clinical trials versus real-world data. There were several reasons why they had this question. A lot of it had to do with the fact that this was an unanswered question pertaining to the use of checkpoint inhibitors
Together, we believed that there was room to provide practice-changing knowledge through an analysis like this. Part of why they approached us is because, in our discussions with the FDA, one of the attributes that we had highlighted about our real-world data work is the fact that we are working with health systems that see more than just the outpatient cancer journey and health systems, that typically have a relationship with the patient that dates back years, in fact, including before their cancer history.
One of the important facets of something like pneumonitis is the potentially slow onset, or the long development period. In order to really look at this question, you do have to have quite a bit of longitudinality to your data, which also means, for something like metastatic lung cancer, being able to look way back in not just the patient’s treatment journey, but potentially to look back before the patient was diagnosed with cancer.
Because we had discussed with the FDA on this facet of the real-world data we integrate, Qi Liu and her FDA OCP colleagues thought of us when they had this question and approached us about answering it.
To be honest, it was the first time that we had really considered a question like this. We were as eager as the FDA to look at the feasibility of an analysis like this in the real-world dataset. We developed a joint analysis plan and proceeded with joint methodology. One of the really important facets about the analysis in the real-world data was the ability to look at the radiation, and specifically the ability to look at the imaging, to assure that the association of radiation with pneumonitis could be actually verified via the imaging.
It was really a collaborative effort, and one where full credit should go to the FDA for generating the research question and bringing it to us.
The FDA appreciates, as we do, the value of real-world data and real-world evidence in clarifying safety issues, in particular in the post-approval realm. As Jon said, the depth, breadth and longitudinality of the real-world dataset that we have at Syapse is a powerful tool for evaluating a phenomenon like pneumonitis where there is a variable clinical course. Sometimes it appears relatively early, sometimes it appears relatively late.
So, Aurora Health in Milwaukee is your data partner for this project. Did all of the use cases in your dataset come from Aurora Health? I happen to know them and Milwaukee quite well, having lived there.
Yes. In this specific analysis, we wanted to look at the patients being seen at one health system, even though our network is much broader than that.
One of the reasons for this was that we wanted to assure that, with a complicated question such as this, we had clinical partners on the ground who were able to collaboratively work with us to address any specific details with analyzing the population. One of the facets of our work on real-world evidence is that we collaboratively engaged the key opinion leaders at the health systems we work with. We will involve them in the research and analysis that we do.
This is very important in this case because Dr. Mike Thompson, who runs the Precision Medicine Program at Aurora, was a key thought partner in this project, providing guidance to the analysis. We certainly plan to scale this analysis up to the rest of the network, but in this early analysis, it was very important to be collaboratively working with a KOL at a health system who provided very important guidance as the project shaped over time.
For oncologists reading this story, what do the results of the study mean for the treatment and management of patients with non-small cell lung cancer?
First of all, if you look at the the data and evidence within the clinical trial realm, and look at the data and evidence within the real-world realm, we can say that, while there is an increased risk of immune checkpoint inhibitor-associated pneumonitis when you’ve had a prior history of pneumonitis, this risk—it appears in both datasets—can be managed.
In neither datasets were there deaths that were attributed directly to the pneumonitis. Obviously, pneumonitis can be a serious complication, but the prior history of pneumonitis should not automatically exclude a patient from consideration of receiving immune checkpoint inhibitor therapy.
What are the mechanisms of action associated with an increased risk of pneumonitis—specifically, with immune checkpoint inhibitors, either as monotherapy or as combination therapy?
The running assumption has been that the mechanism of action is immune-based, i.e. relating to cellular and cytokine dynamics. The mechanism of action is not completely clear.
As we have discussed, there are multiple factors that can impact the incidence of pneumonitis: the immune checkpoint inhibitor; the classical chemotherapy; past radiation; and, of course, there may be other factors to include any history of pneumonia. Teasing out the mechanism of action is a work in progress.
How should health care providers be thinking about this information as well in the context of COVID-19?
We’re intent on expanding the analysis very rapidly to include analyzing patients regarding their past history of pneumonia. This has always been of clinical interest, but increasingly so with the COVID-19 pandemic.
For now, it’s important for clinicians to assume that a past history of pneumonia may well have an impact on the incidence and severity of immune checkpoint inhibitor-associated pneumonitis. They should continue to exercise caution in treating patients with a past history of pneumonia, to include a past history of significant respiratory compromise. We hope to further elucidate this topic in the short-term.
You’ve mentioned probably about a dozen important things that you’d like to build on with this study. What are the next steps?
I think the next steps are basically in three categories. One is to expand the analysis of pneumonitis to include pneumonia, to include infectious causes. The second is to further elucidate the role of radiation therapy by further comparing those patients who received radiation therapy in the past and those who did not.
Then thirdly, looking at a broader range of treatments when one considers treatment-associated pneumonitis. For example, some of the other targeted therapy classes include the tyrosine kinase inhibitors, and beyond.
Did we miss anything?
The importance of this project at a high level is that it’s a demonstration of how real-world data and real-world evidence can supplement and add to clinical trial information, particularly in the safety realm, in this post-marketing era.
One of the drivers of our collaboration with the FDA is our mutual recognition that there’s an opportunity for real-world data to elucidate safety issues for all the reasons that we’ve stated.
Jon, do you have anything to add? Also, any updates about your ongoing work on validation of real-world endpoints?
One of the things that excites us about the research collaboration with the FDA is how rapidly we’ve been able to generate meaningful results. If you look at initiating the research collaboration in mid-August 2019, we were able to have a very impactful project completed and submitted for late-breaking to AACR in mid-January.
That’s a pretty rapid turnaround for what turned out to be a high-impact project selected for a clinical plenary, which we were very happy about and honored by. We certainly were not expecting that at the time. I think the topic was highly relevant given the current global pandemic.
We are moving at a similar pace with our other efforts. As you well know, we have been working on several programs that were announced in the press release, including work on real-world endpoint development and methodology. We always say “validation” in quotes because we’re still trying to figure out what the right word is, working in conjunction with our colleagues at the FDA. That project continues at a rapid pace. We hope to be sharing results on that soon.
We continue to participate in and support the efforts from groups like Friends of Cancer Research, who are working on real-world endpoints as well.
I will say that this has taken on an increasing importance as you look at what’s happening now with clinical trials in oncology. What we’re seeing from the industry, in general, and certainly from the health systems and the life sciences companies that we work with, is that the global COVID-19 pandemic has impacted clinical trial operations.
It has impacted things in such a way that real-world data—and the need to be able to use real-world data in a regulatory setting where one is confident in the underlying methodology and data quality—has taken on increasing importance.
We think that this pandemic is going to accelerate efforts that were already happening, and the use of real-world data in the clinical trials and regulatory realms is one example.
TB: One thing I’ve learned, to extend on Jon’s comments, is the important aspects of utilizing real-world data for safety issues. My background is in developmental therapeutics, early phase clinical trials, specifically phase I trials.
Even today, when we tend not to have an upper age limit for adult patients’ clinical trial eligibility, older patients are not as likely to participate in clinical trials. Real-world data then gives you the ability to look at the entire age spectrum.
Likewise, on the fringes of organ dysfunction, whether one’s talking about hepatic, renal, or even bone marrow dysfunction, one can evaluate those patients within a real-world data set, patients who would usually be excluded from a clinical trial.
That is what the real world looks like; right? That broader range of individuals with a wider age distribution, with a wider range of organ function. That’s so important as drugs are introduced to the general population at an increasingly rapid pace.