Transcriptome and ctDNA Associates with Pembrolizumab Benefit
JCO PO authors Dr. Philippe Bedard (Staff Medical Oncologist at Princess Margaret Cancer Centre and Professor of Medicine at University of Toronto) and Dr. Alberto Hernando Calvo (Medical Oncologist at Vall d´Hebron University Hospital) share insights into their JCO PO article, “Combined Transcriptome and Circulating Tumor DNA Longitudinal Biomarker Analysis Associates With Clinical Outcomes in Advanced Solid Tumors Treated With Pembrolizumab,” one of the top downloaded articles of 2024. Host Dr. Rafeh Naqash and Drs. Bedard and Hernando Calvo discuss how combined transcriptome and ctDNA longitudinal analysis associates with pembrolizumab outcomes.
TRANSCRIPT
Dr. Rafeh Naqash: Hello and welcome to JCO Precision Oncology Conversations where we bring you engaging conversations with authors of clinically relevant and highly significant JCO PO articles. I'm your host, Dr. Rafeh Naqash, podcast editor for JCO Precision Oncology and Assistant Professor at the OU Health Stephenson Cancer Center at the University of Oklahoma.
Today we are excited to be joined by Dr. Philippe Bedard, Staff Medical Oncologist at the Princess Margaret Cancer Center and Professor of Medicine at the University of Toronto, as well as by Dr. Alberto Hernando-Calvo, Medical Oncologist at the Vall d'Hebron University Hospital, both authors of the JCO Precision Oncology article titled, “Combined Transcriptome and Circulating Tumor DNA Longitudinal Biomarker Analysis Associates With Clinical Outcomes in Advanced Solid Tumors Treated With Pembrolizumab.”
Thank you for joining us today. Phil and Alberto.
Dr. Alberto Hernando-Calvo: Thank you.
Dr. Philippe Bedard: Great to be with you. Thanks for having us.
Dr. Rafeh Naqash: One of the reasons we do this podcast, as some of the listeners who listen to this podcast regularly may know, is to bring in novel approaches and try to understand how the field is moving towards a space where we are understanding biomarkers better. So your manuscript that was published in JCO Precision Oncology fulfills many of those criteria. And interestingly enough, I was at a conference at the Society for Immunotherapy of Cancer last month earlier in November and a lot of excitement at SITC was revolving around novel transcriptomic biomarkers, proteomic biomarkers or imaging based biomarkers. So could you tell us a little bit about why you started looking at biomarkers? This is an extremely competitive field. Why did you think that looking at the transcriptome is somewhat different from or more interesting from tumor mutational burden PDL-1 than other biomarkers that we currently use? And that question is for you Alberto to start off.
Dr. Alberto Hernando-Calvo: So I think gene expression profiles may have a predictive performance as compared to already existing biomarkers and this was one of the points that we describe in our manuscript. The gene expression signature that we developed back in 2019 at Vall d'Hebron Institute of Oncology was initially developed based on over 45 different tumor types and tested in over 1000 patients treated with antiPD-1 and anti PDL-1. And back then and in this manuscript, we proved that for instance the gene expression signature VIGex that we developed has a potential complementary role to other predictive biomarkers. In this case, we observe this predictive power with ctDNA dynamics and we then see a correlation with other existing biomarkers such as tumor mutational burden. So I don't think we need to use one or the other, but rather they may have additive predictive power. So we need to better individualize predictive biomarkers based on tumor types and select the best combination possible to improve the performance.
Dr. Rafeh Naqash: I completely agree that one size does not fit all, especially in the landscape of immunotherapy. From your perspective, when you developed the original signature, how did you choose what genes to look at? I looked at the manuscript, on the methodology side, some of the signatures are pro-inflammatory STING interferon gamma based, so how did you try to identify that these are the 7 to 10 or whatever number of signatures on the transcriptome side? And then why did you try to combine it with ctDNA based changes?
Dr. Alberto Hernando-Calvo: Back in our initial manuscript, published in Med from Cell Press, we developed the VIGex gene expression signature, as I mentioned, with taking into consideration over 1000 tumor samples from FFPE that we can consider real world samples because they are from real patients coming from the clinic notes as part of real investigational protocol doing or performing biopsies on patients. We did observe after doing a VIGex research and doing different tests, we eventually collected these 12 different genes. Because there is a combination of both genes involved in the interferon gamma pathway, we have genes associated with Tregs as well as T cell memory cells. So it’s not only looking at genes that are associated with T cell activation or CD8+ T cell infiltration, but also looking at genes that may be overactivated, overexpressed, an immunosuppressive tumor microenvironment. So it was both selecting genes, the minimum number of genes to do it more scalable and having the minimum dataset of genes and including in the signature genes that are already at targets for immune sequent inhibitors or are being tested in immunotherapy combinations.
Dr. Rafeh Naqash: Thank you. And Phil, for the sake of our listeners, could you elaborate upon this aspect of using ctDNA? So this was tumor-informed ctDNA from what I understood in the manuscript. You guys basically try to use it to understand changes in the ctDNA with treatment and then try to combine it with the transcriptome signature. How did the idea come up initially and how did you plan on combining this with an RNA-based signature? Because I have seen manuscripts and other data where people are either using one or the other, but not necessarily both together. So how did you guys come up with that idea?
Dr. Philippe Bedard: Well, we thought that this was a great opportunity to look at the combination of the transcriptome as well as the ctDNA dynamics because we had run an investigator-initiated phase 2 clinical trial called INSPIRE at our institution at Princess Margaret from 2016 to 2018, where patients across five different tumor groups received single agent pembrolizumab. And we really did a deep dive on these patients where there were tumor biopsies before and while on treatment. We did exome sequencing, we did RNA sequencing to capture the transcriptome. And in a prior analysis, we had partnered with Natera to look at their Signatera assay, which is a bespoke ctDNA assay, to look at ctDNA dynamics using this test and the association with response outcomes as well as survival outcomes. So we thought that this was a really unique data set to try and address the question of whether or not there was complementarity in terms of looking at the transcriptome and transcriptome signatures of IO benefit together with the ctDNA dynamics.
Dr. Rafeh Naqash: From a patient treatment standpoint, it sounded like you mostly tried to include individuals who were treated with pembrolizumab. Did this not include individuals who were treated with chemoimmunotherapy or chemotherapy with pembrolizumab? Just pembrolizumab alone? And if that's the case, some of the tumor types there included, from what I remember, ovarian cancer and some other unusual cancers that don't necessarily have approvals for single agent pembrolizumab, but perhaps in the TMB-high setting. So can you elaborate on the patient selection there for the study?
Dr. Philippe Bedard: Yeah, that's a great question. So at the time that the study was designed in 2015, this was really the early days of immune checkpoint inhibitor therapy, so we didn't have the approvals that we have now in specific tumor types for immunotherapy and chemotherapy combinations. So when the study was designed as an investigator initiated clinical trial, the idea was really to capture patients across different tumor types - so head and neck squamous cell carcinoma, malignant melanoma, ovarian cancer, triple negative breast cancer, and a kind of mixed histology solid tumor cohort, where we knew that there were some patients who were going to be immunotherapy responsive, where there was already approvals or evidence of single agent activity, and others where the responses were more anecdotal, to try and understand in a phase 2 clinical trial with kind of a deep dive, which patients benefited from treatment and which didn't.
Dr. Rafeh Naqash: Interesting approach. Going to the results, Alberto, could you help us understand some of the important findings from these data? Because there's different sections of how you tried to look at the response rates, the survival, looking at the immune deconvolution, if you could explain that.
Dr. Alberto Hernando-Calvo: So the first thing that we tried was to further confirm the external validation of this immune gene expression signature, VIGex in the INSPIRE asset. So what we observed at VIGex-Hot, the category defined by VIGex-Hot tumor microenvironment, was associated with better progression free survival. After including that in a multivariable analysis adjusted by other biomarkers such as TMB, PDL-1 or tumor type, this was also confirmed for overall survival. So then the next step was to really try to hypothesize if the addition of ctDNA dynamics, taking into consideration the ctDNA quantification at baseline as compared to cycle three, if those dynamics could further improve the predictive performance of VIGex categories taken in the baseline samples. What we did observe was that, for instance, VIGex-Hot tumors in baseline tumor samples that were having a ctDNA decrease, as I mentioned before on cycle three assessment as compared to baseline, were having both better progression free survival and better prognosis overall. Another important finding was the evaluation of response rate across tumor types considering both biomarkers. I would say the most important finding is that when we were considering a cold tumor microenvironment in baseline samples before pembrolizumab initiation plus an increase in ctDNA values, what we observed is that those patients were having a 0% response rate. So this may help as a future strategy either for intensification of immunotherapy regimens in a more individualized way or for an early stop to immunotherapy and try to avoid financial toxicities as well as toxicities for our patients.
Dr. Rafeh Naqash: From the data that you showed, it seems that there was a strong correlation, as you sort of mentioned, between individuals that had ctDNA clearance and baseline immune pro-inflammatory signatures. So do you really need the transcriptome signature or could the ctDNA just serve as an easy quick surrogate? Because from a cost standpoint, doing whole transcriptome sequencing or more RNA sequencing or tissue standpoint, where tissue is often limited, can become a big issue. So do you think that validation of this may perhaps more revolve around using ctDNA as an easier metric or surrogate? Or am I overestimating the utility of ctDNA?
Dr. Philippe Bedard: I think it's a really good question. In our data set which was relatively small, there were 10 patients who had ctDNA clearance, meaning ctDNA that was positive at baseline was not detected. And so 9 out of those 10 patients, as you alluded to, were VIGex-Hot. So the question is a good one, could you do the same with just ctDNA clearance alone, particularly in identifying these patients who really do well, who have long term disease control on immunotherapy? I think it's a tough question to answer because the field is also changing in terms of sensitivity of detection of ctDNA tests. So we know now that there are newer generations of tests which can detect even at logs down in terms of allele variants in the circulation. So I think we need more data to address the question. I think it is important as to what is the best test, what is the endpoint that we should be using from a drug development point of view in terms of really trying to push and understand which treatment regimens are the most effective and have early readouts in terms of activity. Because we all recognize in the clinic that radiographic response doesn't tell the whole story, especially early radiographic assessments using RECIST or other criteria that we apply in clinical trials.
Dr. Rafeh Naqash: From a clinical trial standpoint, we often talk about validation of these studies. You may have heard of other tests where, for example, the NCI iMatch, which is incorporating transcriptome sequencing based approach to stratify patients as an integral biomarker for treatment stratification. Is that something that you guys are thinking of using, this approach where individuals who are signature highly inflamed perhaps get lesser therapies or there's a de-intensification of some sort similar to what people are trying to do with ctDNA-based approaches?
Dr. Philippe Bedard: I think that's a great question. I think it makes a lot of sense. And certainly, with the new wave antibody drug conjugates in terms of identifying patients who have expression of targets for antibody drug conjugates, that's very attractive as an approach because we don't necessarily have IHC markers for all of the different targets of antibody drug conjugates. We don't necessarily have IHC markers to completely understand different contributions to the tumor microenvironment and whether or not tumors are inflamed. But it's also a challenging approach too because RNA-seq currently is not a routine clinical test. Sometimes there are issues, particularly in patients who have stored specimens that are formalin-fixed and paraffin-embedded in terms of the quality of the RNA for RNA sequencing. And it's not always feasible to get pre-treatment biopsies and turn them around in an approach. So I think it is an attractive approach for clinical trials, but it's a hypothesis that needs to be tested. It's not something that is ready for clinical prime time today in 2024.
Dr. Rafeh Naqash: One of the other interesting observations that I came across in your manuscript was that tumor mutational burden, interestingly, did not correlate with signature high tumors. What is the explanation for that? Because generally you would expect a TMB high to perhaps also have an immune gene high signature. Could it have something to do with the tumor types because there was a heterogeneous mixture of tumor type? Or I'm not sure. What else could you possibly think of that you didn't see those correlations or just sample size limitations?
Dr. Alberto Hernando-Calvo: Yes. So our findings are consistent with prior data suggesting for instance T cell inflamed gene expression profile was also not correlated with tumor mutational burden and both biomarkers in a prior publication. So to have additive predictive performance for identifying patients most likely to benefit from anti PD-1 regimen, so we somehow were expecting this observation, the fact that both biomarkers are not very correlated.
Dr. Rafeh Naqash: So given the proof of concept findings from your study, Phil, what is the next interesting step that you guys are thinking of to expand this? Would you think that a nivolumab-ipilimumab treated cohort would have similar findings? Or is this a treatment specific single agent immunotherapy specific correlation that you found versus something else that you may find in a nivo-ipi cohort or a doublet immune checkpoint cohort?
Dr. Philippe Bedard: The findings are really hypothesis generating. They require additional validation. And you're quite right, there may be nuances in terms of specific tumor types, combinations with other immunotherapy or combinations with chemotherapy or other agents. So I think it would be great if there are other data sets that are collecting this type of information that have ctDNA dynamics and also have transcriptome and potentially exome or genome analysis to look at these types of questions because the field is moving quickly and we really need more data sets in order to understand some of the nuances and greater numbers to validate the signals that we see.
Dr. Rafeh Naqash: And one thing, as you said, the field is definitely moving very quickly. I was meeting with a company an hour back and they have an imaging-based approach using fresh tissue to look at pharmacodynamic biomarkers. And I used to work in the NCI with a group that was very interested and they developed an immuno-oncology pharmacodynamic panel that has been used and published in a few clinical trials where they did phosphorylation status. So the final theme that comes out of most of these research based studies that are being done is that one size does not fit all. But the question that comes to my mind is how many things do you necessarily need to combine to get to a predictive biomarker that is useful, that is patient centric, and that perhaps is able to identify the right therapy for the right patient. What is your take on that, Phil?
Dr. Philippe Bedard: Yeah, that's a great question too. The challenge is it depends on the context in terms of what degree of positive predictive value do you need as well as the negative predictive value to drive clinical decisions. So I think in certain situations where you don't have other approved treatment options and with a therapy that is potentially low toxicity and low financial toxicity, then I think the bar is very high in terms of being able to really confidently identify that patients aren't going to benefit. I think the nuance and the challenge becomes when you move into earlier lines of therapy, or when you talk about combinations of agents, or trying to understand within the context of other available options, particularly with treatments that have significant side effect profiles as well as financial risks, then it becomes a much more nuanced question and you really need comparative studies to understand how it fits versus the existing treatment paradigm. So I'm not really answering your question with a specific number because I think it's hard to give you a number. Some of that we also need input from patients in terms of what kind of level of validation do you need and what kind of level of discrimination do you need in order to drive decisions that are meaningful for them.
Dr. Rafeh Naqash: Definitely early days, as you pointed out. More and more work in this field will hopefully lead us in the direction that we all want to go in.
Now, going to a different aspect of this podcast, which is trying to understand the trajectories for both of you, Phil and Alberto. And as you mentioned, this project seemed to have started in 2015. So I'm guessing there's a history there between Princess Margaret and Vall d'Hebron. Could you highlight that a little bit? And then perhaps, Alberto, after that you could tell us a little bit about your career when you worked at Princess Margaret as a fellow and then now back at Vall d'Hebron. Phil, you as well.
Dr. Philippe Bedard: So absolutely. We have a long history of collaborating with Vall d'Hebron in Barcelona. It's really a great cancer institution with a lot of like minded individuals. We have a formal partnership and we have a lot of informal links in terms of scientists and clinicians who we work with and who we collaborate with on early phase clinical trials, as well as through different investigator networks and other translational projects. So this was really how this collaboration came about and we were fortunate to have Alberto, who came to work with us for two years and brought this great idea of looking at this signature they had developed at Vall d'Hebron in their phase one group and applying it to a data set that we had through the INSPIRE clinical trial.
Dr. Rafeh Naqash: Sounds like a very successful academia-academic collaboration, which is very nice to see. So, Alberto, could you tell us a little bit about your career trajectory and how you ended up at Princess Margaret and then back at Vall d'Hebron and what you do currently?
Dr. Alberto Hernando-Calvo: Yes. So I did my oncology residency at Vall d'Hebron in Barcelona, Spain. Then I decided to further specialize in early drug development as well as head and neck cancer oncology. So I decided to pursue a clinical research fellowship under the supervision of Phil Bedard, among others. And so we decided to further validate the signature that we had developed both in the cancer genomic lab at Vall d'Hebron Institute of Oncology and the phase one unit at Vall d'Hebron, and apply the signature that have been originally tested in patients receiving anti PD-1 or anti PDL-1 combinations in early phase clinical trials. In the phase 2 clinical trial of INSPIRE, where we also had ctDNA dynamics and allowed us to test both biomarkers and see that additive predictive power when we were using both. That was one of my research topics under the mentorship of Dr. Bedard and my fellowship at Princess Margaret. And this was one of the manuscripts describing all the findings of this collaboration between Vall d'Hebron and Princess Margaret Cancer Center.
Dr. Rafeh Naqash: And then, Phil, if you could highlight some of the things that you've done over the course of your career and perhaps some advice for early career junior investigators and trainees.
Dr. Philippe Bedard: I finished my oncology, medical oncology training at the University of Toronto in 2008. And then I did a breast cancer fellowship in Brussels at Breast International Group. At the time, I was really intrigued because it was really kind of the early days of microarray and RNA signatures in terms of expressing signatures were being used as part of a clinical trial that BIG was running called the MINDACT Study. And so when I finished my fellowship, I came back to Princess Margaret, started on staff. I've been here now for 15 years. I was fortunate to work with the phase 1 group and kind of my career has sort of morphed in terms of early drug development as well as genomics. I've been involved with the American Association for Cancer Research project GENIE, where I'm the current chair. This is really an international data sharing project with panel based sequencing, which both Princess Margaret and Vall d'Hebron have contributed to. And I’ve been fortunate to work with a number of really talented early career investigators like Alberto, who spend time with us in our drug development program and launched transitional research projects that leverage some existing data sets at their own institutions and also bring together with different research groups at our institution to lead to publications like this one.
Dr. Rafeh Naqash: Thank you so much. This was very exciting. Phil and Albert, thanks for joining us today and thank you for allowing us to discuss your interesting manuscript and hopefully we'll see more of this biomarker work from you guys in the near future, perhaps published in JCO Precision Oncology.
And thank you for listening to JCO Precision Oncology Conversations. Don't forget to give us a rating or review and be sure to subscribe so you never miss an episode. You can find all ASCO shows at asco.org/podcasts.
The purpose of this podcast is to educate and to inform. This is not a substitute for professional medical care and is not intended for use in the diagnosis or treatment of individual conditions.
Guests on this podcast express their own opinions, experience and conclusions. Guest statements on the podcast do not express the opinions of ASCO. The mention of any product, service, organization, activity, or therapy should not be construed as an ASCO endorsement.
Otros episodios de "JCO Precision Oncology Conversations"
No te pierdas ningún episodio de “JCO Precision Oncology Conversations”. Síguelo en la aplicación gratuita de GetPodcast.
