Over the past decade, immune checkpoint inhibitors (ICI) have substantially improved outcomes in a variety of cancers. However, despite these advances, only a subset of patients responds to ICI treatment. There is, therefore, a high unmet need for predictive biomarkers of response in order to identify patients who may benefit from immunotherapy. While the expression of programmed death-ligand 1 (PD-L1) has demonstrated clinical utility for patient selection, as has microsatellite instability or mismatch repair deficiency (MSI/dMMR), PD-L1 is not considered a robust predictive biomarker of response. Tumor mutational burden (TMB) is an emerging biomarker for immunotherapy, which is currently being heavily investigated. TMB is a measure of the somatic mutation rate within a tumor’s genome and varies between patients and across tumor types; the higher the tissue TMB (tTMB), the greater the number of neoantigens expressed by the tumor. These neoantigens can be recognized by T-cells and thereby stimulate an anti-tumor immune response. Consequently, a high tTMB may be associated with a greater likelihood of response to ICI. Indeed, several retrospective analyses have indicated a correlation between a high tTMB and clinical benefit of immune checkpoint blockade.

The association of high tTMB and clinical outcomes with pembrolizumab was prospectively analyzed in ten tumor-specific cohorts from the single arm, open-label, phase 2 KEYNOTE-158 study (N=1066). Tissue TMB was determined by the FoundationOne CDx assay and was defined as high if there was ≥10 mutations per megabase. The analysis included patients with anal, biliary, cervical, endometrial, mesothelioma, neuroendocrine, salivary, small-cell lung (SCLC), thyroid, and vulvar cancers whose disease had progressed following prior treatment, and who had no satisfactory alternative treatment options. Results from this analysis were recently reported in Lancet Oncology. Among the tumor types, the highest proportion of tTMB-H was observed in SCLC (45%) and cervical cancer (21%). Of the 790 patients who were evaluable for tTMB, 102 patients (13%) were tTMB-H and 688 (87%) were non-tTMB-H. The primary endpoint of the study was objective response rate (ORR) per independent central review (ICR), and secondary endpoints included duration of response (DOR), progression-free survival (PFS), overall survival (OS) and safety. The median follow-up was 37.1 months. ORR was 29% in patients who were tTMB-H and 6% in patients with non-tTMB-H; complete responses were seen in 4% and 2%, respectively. Median DOR was not reached in patients with tTMB-H and was 33.1 months in patients with non-tTMB-H. In contrast, the predictive role of the tTMB-H status was not seen in terms of median PFS (2.1 months for both tTMB-H or non-tTMB-H) and median OS (11·7 months in tTMB-H versus 12·8 months for non-tTMB-H). However, the latter part of the PFS and OS Kaplan-Meier curves clearly favored patients with tTMB-H, which suggests a long-term benefit with pembrolizumab in this patient group. Of note, tTMB was not associated with PD-L1 expression and the predictive value of tTMB was more robust and independent of PD-L1 expression.

The authors concluded that these findings indicate that tTMB-H could be a useful biomarker for selecting patients with advanced solid tumors who are most likely to benefit from single agent pembrolizumab. Indeed, based on these results, the FDA granted accelerated approval to pembrolizumab for patients with advanced tTMB-H cancers, irrespective of tumor type, whose disease has progressed after previous treatment and who have no satisfactory alternative treatment options. In a commentary accompanying the study publication, Dr. Melissa Bersanelli (University Hospital of Parma, Italy) highlighted that the study showed the usefulness of tTMB status to predict response to pembrolizumab, even in tumors that are poorly immunogenic such as SCLC, and noted that a combined biomarker score may be helpful to improve the selection of treatment for cancer patients whose treatment is currently driven only by PD-L1 expression.

References

Marabelle A, et al. Lancet Oncol. 2020; September 10. [Epub ahead of print]

Bersanelli M. Lancet Oncol. 2020; September 10. [Epub ahead of print]