David R. Gandara, MD, discusses the current state of tumor mutational burden in lung cancer, ongoing research further examining the utility of this biomarker, and challenges to address with future research.
Tumor mutational burden (TMB) possesses the potential to be a strong predictive biomarker of efficacy for checkpoint immunotherapy, although it’s quite complex and has shown varied results in different tumor types, David R. Gandara, MD, said during a presentation during the 21st Annual International Lung Cancer Congress, a program developed by Physicians’ Education Resource® (PER®), LLC.
“For TMB and its implications for immunotherapy, non–small cell lung cancer (NSCLC) is a poster child. Lung cancer has a very high number of mutations compared with other cancers,” said Gandara. “That’s largely because of the population of patients with the disease who are smokers or were smokers in the past. In contrast, in never-smokers with lung cancer, such as those with EGFR mutations, TMB is usually low and that helps to explain why those patients do not respond favorably to checkpoint immunotherapy.”
TMB can be measured through comprehensive genomic profiling, with one FDA-approved test, FoundationOne, using tissue and a second test, GuardantOMNI, using blood that is pending approval and has the potential to become a companion diagnostic, according to Gandara. This blood assay is currently being tested in the B-FAST trial, which has already completed patient accrual. In the trial, patients are randomized based on blood TMB levels to receive either atezolizumab (Tecentriq) or a platinum chemotherapy.
“The B-FAST trial is probably the most important [ongoing TMB] trials [in NSCLC],” said Gandara. “Another trial, IMpower110, which is examining atezolizumab as a first-line immunotherapy versus a platinum-based chemotherapy, will be published shortly. We’re awaiting the results of the blood TMB within that trial, but I am optimistic that they will indicate predictive value.”
In an interview with OncLive, Gandara, director of the Thoracic Oncology Program and senior advisor to the director of UC Davis Comprehensive Cancer Center; and 2017 OncLive Giant of Cancer Care in Lung Cancer, discussed the current state of TMB in lung cancer, ongoing research further examining the utility of this biomarker, and challenges to address with future research.
OncLive: Could you start off by speaking about what we currently know about TMB as a biomarker in this space?
Gandara: TMB is one of the lead candidates regarding predictive biomarker for efficacy of checkpoint immunotherapy, along with PD-L1 expression. However, TMB is a more complex biomarker; it’s not a discrete marker like a mutation. TMB is a continuous variable biomarker and it varies tremendously between cancer types or within individual patients.
Additionally, single-agent pembrolizumab (Keytruda) received FDA approval across tumor types for those with high TMB; that’s a step forward, but that approval probably raises more questions than it does answers. In my presentation during the 2020 International Lung Cancer Congress, I discussed the biomarker itself, the studies that do and do not support it, and why.
TMB can be measured in several ways; it can be identified via a research test, such as whole exome sequencing, o, now, a variety of commercially available gene panels that are [done through] next-generation sequencing; this is referred to as comprehensive genomic profiling. Examples of genomic profiling include the FoundationOne test in tissue or the GuardantOMNI test in blood. The blood tests are not commercially available yet; they’re still under development.
My own contribution here is that I am the one who worked to publish the analytical and clinical validation of TMB in blood. Most of the data [available with] TMB [come from] retrospective analyses; however, prospective trials are ongoing. One of those trials, B-FAST, utilizing the Foundation blood TMB test has completed its accrual. In this trial, patients are randomized based on blood TMB levels to either the checkpoint inhibitor atezolizumab or to a platinum chemotherapy. We’ll see what the outcome of that trial is.
As I said, we already have FDA approval for the Foundation test in tissue, and if this works equally well in blood, I’m sure this will become a companion diagnostic, as well. Lastly, like we did with PD-L1, we need harmonization of the various methods for determining TMB and several projects that are seeking to do this; one of those efforts is being led by the Friends of Cancer Research.
Could you expand on the liquid biopsy assay you helped to develop that measure TMB in blood? How was this test validated by the results from the OAK and POPLAR trials?
I had the opportunity to work with Foundation Medicine and Roche-Genentech to analyze data from our previous large phase 3 trial, OAK, which compared atezolizumab with docetaxel in the second-line setting of advanced NSCLC. It was a positive study that [showed a] positive outcome for survival and we wanted to see if we could apply both a tissue TMB and a new, blood TMB assay to those populations. We wanted to use a test and validate procedure, where we used samples from another study, POPLAR, as the test set, and then the samples from OAK as the validation. We developed a computational algorithm on how to measure TMB in blood, which is very close to the tissue, but not exactly the same for technical reasons. We applied this first to POPLAR and then to OAK. We were able to show that, yes, the blood TMB assay that we developed could predict response rate and progression-free survival (PFS). We determined what we thought was the best cut point at 16 mutations per megabase in an analytical process that is highly accepted.
That being said, since this is a continuous biomarker, not a discrete biomarker—not like a mutation—there is no magic level. You decide what gives you the best predictive value, but also what could benefit the greatest proportion of patients. As you go up higher and higher in levels of TMB, the predictive value gets better, but the denominator shrinks, meaning that there are very few patients in that category. At the levels we selected, 16 mutations per megabase, that’s about one-quarter of the population, so that’s a reasonable level.
In this retrospective analysis, we didn’t have a clear predictive value in overall survival (OS). We suspect that was because patients crossed over; we know exactly how many patients crossed over and how they did. [We believe] those patients diluted the value for OS, but that is the reason why prospective trials needed to be done.
The first of those, B-F1RST, has already been completed and met its goals. Results showed a stepwise increase in response rate using this blood TMB assay; this provided proof of concept. Since then, the B-FAST study, which is the randomized phase 3 trial, has been completed. If [the results are] positive, then I feel confident that this test will become a companion diagnostic.
How does plasma-based TMB function as a predictive biomarker for responses to immunotherapy and immunotherapy combinations?
The 1 trial that comes to mind for blood TMB is the MYSTIC trial, which is actually a phase 3 trial of durvalumab (Imfinzi), another checkpoint inhibitor similar to pembrolizumab, plus tremelimumab or durvalumab alone compared with chemotherapy [in patients with EGFR and ALK wild-type, locally advanced or metastatic NSCLC]. That [trial] used a new blood test called GuardantOMNI. This additional test from blood TMB developed by Guardant Health was shown to be highly predictive of survival in patients receiving durvalumab plus tremelimumab compared with chemotherapy. That study and test are still under further development.
How would you say that TMB compares with other biomarkers that are currently under investigation in lung cancer?
Immunotherapy and the immune environment in cancer is extraordinarily complex, so by comparison to a mutation, fusion, or translocation in cancer, such as EGFR or ALK, there is much more to consider here. It’s not only about what is going on in the tumor; you have to also account for the immune system: The T-cell infiltration that occurs or does not occur that facilitates the body’s immune response to that cancer. We know that cancers can produce proteins such as PD-L1. The PD-L1 and PD-L blockers interfere with those pathways, so then there is immune recognition. The problem in many cases is that even though that [interference] occurs, if there are not immune-infiltrating lymphocytes that can be brought in to affect the killing of the cancer cells, just removing that blockade is not enough.
Many biomarkers—literally hundreds—are trying to look at the tumor, the microenvironment, the T cells, suppressive factors, and then the entire issue of the gut microbiome. It’s increasingly important that the gut microbiome can also affect tumor immunity. Data from a couple of interesting, recently published studies recently published have shown that the use of steroids, for example, can reduce the efficacy of checkpoint immunotherapy or antibiotics. With regard to the antibiotics, in particular, [it is] not because of an effect on the tumor or the T cells, per se; it’s because of an effect on the gut microbiome. All of that being said, the 2 most common predictive biomarkers for immunotherapy today are PD-L1 expression, which is, again, a good biomarker in some situations, and TMB.
What are some of the challenges currently faced with TMB?
Some of the challenges are that TMB is being applied in areas where it’s not likely to work. For example, [when] TMB was developed [it was supposed to be used to provide] both the analytical and clinical validation for single-agent checkpoint immunotherapy. However, now it’s being applied to combinations of immunotherapy and, more importantly, it’s being assessed in studies where immunotherapy was given together with chemotherapy. The best evidence available suggests that chemotherapy is agnostic to levels of TMB. If that is the case, then a combination of chemotherapy plus a checkpoint inhibitor would dilute some of the predictive value of TMB compared with studies of monotherapy.
That was shown very distinctly within the past year in several presentations at medical meetings [with regard to] a variety of phase 3 trials. In every trial where checkpoint immunotherapy was given as a monotherapy, we saw good predictive value for TMB. Every single trial where checkpoint immunotherapy was combined with chemotherapy, we saw either borderline or no predictive value. That’s exactly what we would have expected. Unless you realize there are limitations on where TMB should be applied, at least as a single biomarker, you’re left confused about its role.
That being said, at the the AACR Virtual Annual Meeting II, a very nice presentation was given on pembrolizumab monotherapy in a variety of different studies where TMB in tissue was measured by the FoundationOne assay. [These results indicated that TMB] was highly correlated with PFS and OS. Again, those studies are retrospective and they don’t typically lead to an FDA approval, whereas the prospective KEYNOTE-158 trial, support the FDA approval of this assay.