Tumor mutational burden has been associated with response to immunotherapy in patients with lung cancer, but there have been various criticisms regarding the validity and routine use of this as a biomarker.
Tumor mutational burden (TMB) has been associated with response to immunotherapy in patients with lung cancer, but there have been various criticisms regarding the validity and routine use of this as a biomarker.
“…To the question of whether TMB can be a clinically relevant biomarker…I still think that it is biologically relevant. And I think that it can become a part of our clinical armamentarium, particularly in lung cancer, where we [should be] doing this sort of molecular profiling as routine,” said Matthew D. Hellmann, MD, of Memorial Sloan Kettering Cancer Center in New York, in a virtual presentation during the 15th Annual New York Lung Cancers Symposium.1 “I think that it is quite possible that thinking about the world of lung cancer through a dual biomarker strategy might identify patients with particular benefit, [namely in] the high [PD-L1], high [TMB] patients, and with essentially nil benefit in the low [PD-L1], low [TMB] patients.”
During the presentation, Hellmann sought to clarify the confusion that surrounds the use of TMB as a biomarker of response for treatment with immunotherapy in lung cancer.
Background of Interest in and Issues With TMB
With the rise of immunotherapy as a clinical therapeutic, an effective biomarker to identify patients who were more likely to benefit from immune checkpoint blockade became increasingly necessary as use of these agents spread across tumor types. Looking to the genesis of TMB as a biomarker, Hellmann explained that in the early 2010s, there were preclinical data showing that mutant epitopes could serve as targets for antitumor immunity. Additionally, patients with non–small cell lung cancer (NSCLC) who were heavier smokers were found to benefit more from immunotherapy. These led to questions of the biological factors that led to this effect.
In an analysis in 2015 of 2 cohorts of patients with NSCLC treated with pembrolizumab, improved progression-free survival (PFS) and response rates were observed in patients with higher nonsynonymous mutational burden in their tumors compared with those with lower mutational burden (HR, 0.19; 95% CI, 0.08-0.47; P = .0004).2
Further benefit from immune checkpoint blockade has been observed across tumor types in patients with high TMB beyond just lung cancer, leading to the tumor-agnostic FDA approval this year. “Across tumors, it seems to be the case that the more tumor mutation burden, the more likely [the patient is] to benefit from PD-1 [inhibitor] therapy,” Hellmann said. However, he did note that there are some tumor types where TMB does not seem to be a driver of benefit.
Since TMB first emerged, there have been various questions and criticisms over how it comes into play in the larger biology of the tumor.
For instance, TMB was not found to be synonymous with elements of tumor microenvironment inflammation, including PD-L1 expression and T-cell inflamed gene expression profiles.3 Instead, these biomarkers independently help predict response to PD-1 blockade.
Another such criticism over the use of TMB as a biomarker is that it just acts as a proxy for neoantigen burden. However, Hellmann stressed that this is unproven and current tools for predicting neoantigen burden are not very good. Further, assessments of neoantigen burden and TMB overlap with similarity meaning that they are likely not of distinct biology, he added.
TMB Use in Clinical Trials and Impact
The phase 3 CheckMate 227 study (NCT02477826) sought to identify if TMB could be used to identify patients with NSCLC who were more likely to benefit from frontline treatment with nivolumab (Opdivo) and ipilimumab (Yervoy) over platinum doublet chemotherapy.4,5
PFS was indeed improved in patients with high TMB (≥10 mutations/Megabase) who were treated with the immunotherapy combination versus the chemotherapy regimen (HR, 0.58; 97.5% CI, 0.41-0.81; P = .0002). Comparatively, the hazard ratio for PFS in patients with lower TMB (<10 mutations/Megabase) was 1.07 (95% CI, 0.84-1.35).
Additionally, in patients who also had positive PD-L1 expression, the 1-year PFS rate with immunotherapy was 42% versus 16% with chemotherapy.
“What we expected to see was the opportunity to enrich using this biomarker in a way that was independent of the current biomarkers that we use,” Hellmann commented. “And so at this time…there was reasonable confidence that tumor mutation burden was a relevant biological biomarker. If you had higher TMB, you were more likely to benefit from immunotherapy. But of course, translating that into a clinical biomarker still had different challenges.”
Although there was initial excitement over the potential benefit of TMB as a biomarker, it was dulled by questions regarding harmonization of what characterized tumors as having high TMB expression and how this could be used in clinical practice.
Several months later, Bristol-Myers Squibb released information from CheckMate 227 suggesting that there was not a significant difference in overall survival (OS) benefit between the TMB-high and TMB-low populations; the hazard ratios comparing nivolumab and ipilimumab with chemotherapy were similar in both populations.6 The median OS with nivolumab and ipilimumab was 23.03 months versus 16.72 months with chemotherapy in those with high TMB (HR, 0.77; 95% CI, 0.56-1.06). In those with low TMB, the median OS with immunotherapy was 16.20 months versus 12.42 months with chemotherapy (HR, 0.78; 95% CI, 0.61-1.00).
Hellmann noted that many believed after these results that TMB did not work, but he disagrees. “I’m not sure that it invalidates the relevance of TMB, but I think [it] makes [for] a more nuanced challenge [based on] these results,” Hellmann said.
He advised that there was still a difference in survival between the patients with TMB-high and TMB-low expression in terms of those who received immunotherapy (23.03 vs 16.72 months), which was as expected. However, the complicating factor was that there was also an OS benefit in TMB-high tumors who were treated with chemotherapy versus those that were TMB low (16.72 vs 12.42 months).
“I think [it] represents an idiosyncratic and unusual, or unexpected, conclusion rather than one that should make us throw out the whole TMB baby with the bathwater,” Hellmann commented. “It raised the question of whether TMB is actually prognostic, and perhaps we had missed this. But I would argue, having looked at all the other data in which we have TMB [assessment] for patients who are receiving chemotherapy, we really don’t see this in essentially any other study.”
For example, in the KEYNOTE-189 trial (NCT02578680), which did not show an association with improved efficacy with pembrolizumab added to chemotherapy in frontline metastatic nonsquamous NSCLC based on TMB, the chemotherapy alone arm showed similarly low OS in both the TMB-high and -low populations.7
Hellmann alternatively suggested that the difference seen in the CheckMate 227 study could be due to crossover in the second line. Those who had high TMB and got chemotherapy in the first line but then crossed over to immunotherapy in the second line could have contributed to the increased survival curve.5
“This is entirely speculative on my part. I think it’s very challenging to tease out what’s the impact of crossover in these sorts of studies. But if you look at the 2 studies in which there was a seemingly different outcome in chemotherapy [CheckMate 227 and CheckMate 026], you can see that the curves start to separate somewhat late in the overall outcomes, which could [be associated] with the time in which one might consider a second-line immunotherapy starting to come into play,” he commented.
“But perhaps the most likely, it’s just something idiosyncratic that happened and if we were to have done the study 10 times, this result would have only happened once as it did, particularly in this study where there wasn’t specific stratification for TMB up front.”
Hellmann concluded that he still believes TMB is a relevant biomarker, but perhaps one that requires further improvement.