High TMB Fails to Show Predictive Biomarker Potential for Checkpoint Inhibition Across Cancers

Daniel J. McGrail, PhD

High tumor mutational burden (TMB-H) was useful in predicting clinical responses to checkpoint inhibitors in patients with certain cancer subtypes; however, TMB-H failed to demonstrate utility as a biomarker for treatment with checkpoint inhibitors across all solid cancer types, according to findings from a study published in Annals of Oncology.¹

Patients with TMB-H melanoma, lung, and bladder cancer, in which CD8 T-cell levels positively correlated with neoantigen load, experienced a 39.8% overall response rate (ORR) to checkpoint inhibitors (95% CI, 34.9%-44.8%). The ORR was significantly higher than that observed in TMB–low (TMB-L) tumors (odds ratio [OR], 4.1; 95% CI, 2.9-5.8; P < 2 x 10^-16).

However, patients with TMB-H cancers of the breast and prostate, and glioma, wherein CD8 T-cell levels did not correlate with neoantigen load, failed to achieve a 20% ORR; the ORR was 15.3% (95% CI, 9.2%-23.4%; P = .95). Moreover, the ORR was significantly lower relative to TMB-L tumors (OR, 0.46; 95% CI, 0.24-0.88; P = .02).

“This study represents the most comprehensive analysis to date of TMB as a biomarker for response to immune checkpoint blockade,” said lead study author Daniel J. McGrail, PhD, a postdoctoral fellow in the Department of Systems Biology at The University of Texas MD Anderson Cancer Center.² “Our results do not support applying high TMB status as a universal biomarker for immunotherapy response, suggesting that additional tumor type-specific studies are needed to clarify how best to apply TMB status in cancer types where it does not appear to be associated with outcomes.”

PD-1/PD-L1 inhibitors have reformed the treatment paradigm and improved responses across many cancer subtypes.¹ TMB-H status had been proposed as a potential biomarker of response to checkpoint blockade in patients with cancer under the hypothesis that an increased number of mutant proteins will ultimately elicit a more immunogenetic tumor microenvironment.

In May 2017, the FDA granted an accelerated approval to pembrolizumab (Keytruda) for the treatment of adult and pediatric patients with unresectable or metastatic, microsatellite instability–high (MSI-H) or mismatch repair deficient (dMMR) solid tumors that have progressed following prior treatment and who do not have satisfactory alternative treatment options. The approval also encompassed patients with MSI-H or dMMR colorectal cancer who progressed on treatment with a fluoropyrimidine, oxaliplatin, and irinotecan.³ The approval marked the first tissue or site agnostic indication for pembrolizumab.

In June 2020, the FDA granted another accelerated approval to pembrolizumab for the treatment of adult or pediatric patients with unresectable or metastatic TMB-H solid tumors that have progressed following prior treatment and who have no satisfactory alternative treatment options.⁴

This approval was based on findings from the randomized phase 2 KEYNOTE-158 study, in which pembrolizumab elicited an ORR of 29%, with a 4% complete response rate in 102 patients with TMB-H tumors.⁵ The median duration of response was not reached with pembrolizumab; 57% of patients had response durations lasting 12 months or longer, and 50% had responses lasting 24 months or longer.

The KEYNOTE-158 trial included patients with small cell lung cancer, cervical, endometrial, anal, vulvar, neuroendocrine, salivary, and thyroid cancers, as well as mesothelioma.

“The FDA approval of pembrolizumab for patients with high TMB certainly provides an important option for many patients,” said senior study author Shiaw-Yih Lin, PhD, a professor in the Department of Systems Biology at The University of Texas MD Anderson Cancer Center.² “However, we felt that it was important to look more closely at TMB status in a broader group of cancer types and establish approaches to harmonize TMB across various assays to enable clinicians to best utilize the recent FDA approval.”

The 9 tumor types that were included in the KEYNOTE-158 trial account for an estimated 11% of new cancer cases included in the study.¹ About 25% of new cancer cases comprised subtypes that had a positive correlation between neoantigen load and CD8 T-cell infiltration, whereas about 50% of new cases were not associated with any correlation between neoantigen load and CD8 T-cell infiltration.

As such, evaluating whether TMB-H could serve as a predictive biomarker in the latter subgroup was relevant to a large proportion of patients.

The study evaluated over 10,000 patients across 31 cancer types who had available data in The Cancer Genome Atlas (TCGA). Samples were categorized by whether CD8 T-cell infiltration positively correlated with neoantigen load.

TMB-high status was defined as 10 mut/Mb determined in the context of F1CDx assay genes or when treated as a continuous variable.

Further results revealed that bulk ORRs were not significantly different between the group of cancers that demonstrated correlation between neoantigen load and CD8 T-cell infiltration and the group that did not demonstrate a correlation between the 2 variables.

Overall survival analysis, as well as treating TMB as a continuous variable, yielded similar results.

In addition, the study found that the F1CDx assay overestimated TMB-H cancer-related genes compared with TMB from whole-exome sequencing data in 25 of 31 cancer types analyzed. Moreover, a negative correlation was observed between the median TMB in a given cancer type and the degree of TMB-H overestimation.

The study was limited by its retrospective nature, use of various DNA sequencing approaches, and variations in checkpoint inhibitors and clinical outcomes reported; however, the findings suggest that additional studies should be done to improve cancer type–specific assessment of TMB from targeted sequencing and activity of checkpoint inhibition in TMB-H tumors before TMB becomes a broadly utilized biomarker.

“While TMB status does show value in predicting response to immune checkpoint blockade in several cancer types, this was not generalizable across all cancers. For those cancer types where a high TMB does not appear to increase immunogenicity, additional prospective studies are needed to determine if TMB status can be an effective clinical biomarker and at what threshold,” concluded McGrail.²

References

1. McGrail DJ, Pilié PG, Rashid NU, et al. High tumor mutation burden fails to predict immune checkpoint blockade response across all cancer types. Ann Oncol. Published online: March 15, 2021. doi:10.1016/j.annonc.2021.02.006
2. Study finds high tumor mutation burden predicts immunotherapy response in some, but not all, cancer. News release. The University of Texas MD Anderson Cancer Center. Published March 15, 2021. Accessed March 31, 2021. https://bit.ly/2Pn8FAm.
3. FDA grants accelerated approval to pembrolizumab for first tissue/site agnostic indication. FDA. Published May 23, 2017. Accessed April 1, 2021. https://bit.ly/3dobAkm.
4. FDA approves pembrolizumab for adults and children with TMB-H solid tumors. FDA. Published June 16, 2020. Accessed April 1, 2021. https://bit.ly/39Ax8sN.
5. Marabelle A, Fakih M, Lopez J, et al. Association of tumour mutational burden with outcomes in patients with advanced solid tumours treated with pembrolizumab: prospective biomarker analysis of the multicohort, open-label, phase 2 KEYNOTE-158 study. Lancet. 2020;21(10):1353-1365. doi:10.1016/S1470-2045(20)30445-9