Tumor mutational burden associated with survival on immunotherapy across cancer types
Click Here to Manage Email Alerts
Key takeaways:
- Increased tumor mutational burden levels appeared associated with decreased hazards for death.
- Tumor mutational burden assays are rapidly evolving.
CHICAGO — Higher tumor mutational burden appeared associated with more favorable real-world OS among a large cohort of patients with various cancer types who used immune checkpoint inhibitor monotherapy.
Researchers observed the association independent of microsatellite instability (MSI) status, results presented at ASCO Annual Meeting showed.
Rationale and methods
“Although immune checkpoint inhibitors have revolutionized cancer therapy, the predictive value of current biomarkers is limited, which is in part because they are not discreet biomarkers, such as a mutation, but rather are continuous variables that are measured across a wide range of different tumor types,” David R. Gandara, MD, FASCO, chief medical officer of International Society of Liquid Biopsy and senior adviser to the director of UC Davis Comprehensive Cancer Center, said during a presentation.
In 2020, FDA approved pembrolizumab (Keytruda, Merck) for patients with tumor mutational burden (TMB)-high solid tumors as determined using an FDA-approved test .
“That was based on response rate and duration of response, but not survival,” he said. “The clinical utility of tumor mutational burden has been controversial, which we think is due in part to differences in assays and cut points. It is also important to note that tumor mutational burden has consistently done better when looking at immune checkpoint inhibitor monotherapy rather than combination regimens.”
Researchers sought to evaluate the performance of an FDA-approved TMB algorithm to identify patients with favorable OS on single-agent immune checkpoint inhibitor therapy in a large real-world cohort.
The prespecified analysis included 8,440 patients with 24 different advanced cancer types included in the nationwide combined Flatiron Health/Foundation Medicine clinico-genomic database of nearly 280 U.S. cancer centers.
Researchers benchmarked a composite real-world mortality variable against the National Death Index. They used Cox proportional hazards models — adjusted for ECOG performance status, prior treatment, sex, age, opioid prescription before treatment, genetic ancestry and socioeconomic assessment — to assess relative hazards of death by TMB interval.
Findings
Results showed an association of increased TMB levels with decreased hazards for death independent of prognostic variables, Gandara said.
In individual tumor types of any MSI status, for nine of the 10 tumor types with adequate statistical power, TMB of 10 or greater had a significant association with OS during immune checkpoint inhibitor therapy.
“We also found that for microsatellite-stable tumor types, tracking appeared similar to the overall group with the distinct exception of colorectal cancer, which was anticipated because of the importance of MSI status for that particular tumor type,” Gandara said.
TMB varied greatly according to tumor type.
“For example, the very low tumor mutational burden of less than 5 was only 20% in melanoma; however, in pancreatic cancer it was 90%,” he said. “Conversely, for the high levels of 30 or greater, pancreatic cancer was 0.5%. Results of exploratory analysis also showed a linear relationship between tumor types and the association with real-world OS.”
Limitations of the study included the fact that ORR and PFS could not be assessed in the real-world analysis.
Implications
TMB assays are rapidly evolving and not only becoming standardized, but multiple variations are being explored, Gandara said.
“We feel this will improve its predictive value,” he said. “The TMB reported here will not likely be the TMB assays we use in the future. Those new assays should be compared with this FDA-approved TMB with respect to clinical outcomes and predictive power. TMB deserves investigation in completed and ongoing trials across tumor types to find patients who benefit most.”
Perspective
Back to Top
Jorge J. Nieva, MD
This study is a wonderful exercise in data mining.
We have a plethora of important clinical and molecular data housed within our next-generation sequencing provider company. This comes to us from Foundation Medicine, which is a leader in the field, together with Flatiron Health, which has cooperated in providing the necessary patient clinical outcome measures. Together, the two have made use of the existing clinical data in a very important way.
What we find in this abstract is that TMB is a very good predictor of response to immune checkpoint inhibitors. This association is somewhat dependent, meaning the higher the TMB, the better the outcomes for our patients. This makes sense and is more confirmatory than a new finding.
I think the findings from this study show the FDA was justified in approving pembrolizumab in this cohort. We see that across many tumor types, TMB predicts favorable outcomes. This type of data will be very useful to people who do modeling and can incorporate TMB with other predictors, such as performance status, PD-L1 status and coexistent use of chemotherapy or other therapeutics. I anticipate that mathematicians will grab onto this data and use it to build predictors of prognosis for our patients.
I’d like to congratulate the authors of this manuscript for bringing to light this large volume of real-world evidence.
Perspective
Back to Top
Thomas U. Marron, MD, PhD
The immune system serves to differentiate “self” from “foreign,” and the more “foreign” a tumor looks, the more likely immunotherapy is to effectively activate the immune system to recognize and eliminate cancer. TMB is a surrogate for “foreignness” in that each mutation has the potential to produce a mutant protein that can be recognized by the immune system, leading to immune activation and elimination of cancer cells.
Many trials failed to show utility of TMB as a biomarker, but KEYNOTE-158 enabled us to use TMB as a biomarker to obtain pembrolizumab across tumor types. The optimal cutoff for TMB has been a topic of debate, which makes this a binary biomarker, although it is more of a continuous variable.
This study is notable for its size, the fact that it uses a real-world data set and its use of survival as an endpoint, which is the best measure of efficacy. Interestingly, while the cohort with low TMB (< 5 mutations/Mb) showed no significant benefit of PD-(L)1 agents, there was a significant benefit in the cohort with 5 to 10 mutations/Mb, for whom the FDA label is not applicable.
In the subgroup analysis, the impressive HR for cancers not classically considered immunotherapy-responsive, including breast, colorectal and unknown primary, is particularly impactful and could be used to guide treatment decision-making in these populations. It also may draw into question the utility of universal use of immunotherapy in all patients with tumor types classically thought of as immunotherapy-responsive, such as lung, melanoma and urothelial, when patients have low TMB.
Collapse
Gandara D, et al. Abstract 2503. Presented at: ASCO Annual Meeting; June 2-6, 2023; Chicago.
Click Here to Manage Email Alerts