BARCELONA – In presentations at the annual meeting of the European Society for Medical Oncology in Barcelona on Friday, researchers reported seemingly contradictory results on the utility of tumor mutational burden (TMB) as a biomarker for treatment of advanced non-small cell lung cancer (NSCLC) with pembrolizumab (Merck's Keytruda).
Researchers have been analyzing the utility of TMB as a biomarker for treatment with immunotherapy in certain cancers, but a study published in Nature Genetics earlier this year found that its use is complicated in pan-cancer applications because the cutoff defining high TMB status differed for patients based on histology.
Although the five studies presented at ESMO focused on advanced NSCLC, the difficulty of using TMB as a biomarker for certain therapeutic applications was also apparent.
Yale School of Medicine researcher Roy Herbst presented findings from the KEYNOTE-010 and KEYNOTE-042 trials, which retrospectively analyzed associations between TMB and efficacy of pembrolizumab monotherapy. He noted that previous studies have shown that TMB may have utility as a biomarker for anti-PD-L1 therapies given as a monotherapy or in combination with anti-CTLA4 therapies in NSCLC.
The researchers determined TMB in 253 patients from KEYNOTE-010 and in 793 patients from KEYNOTE-042 by whole-exome sequencing of tumor and matched normal DNA. They used a pre-specified cutoff point of 175 mutations per exome to assess clinical utility of TMB and looked at the association of the biomarker with overall survival (OS), progression-free survival (PFS), and overall response rate (ORR).
In KEYNOTE-010, the researchers observed general improvements in OS, PFS, and ORR in patients with a TMB of 175 or more who were treated with pembrolizumab, whereas TMB was generally not associated with outcomes for chemotherapy. Herbst said these results showed that a high TMB was useful for predicting the outcome of treatment with pembrolizumab as a single agent.
Median OS for patients taking pembrolizumab was 14.1 months compared to 7.6 months for patients taking chemotherapy. Similarly, median PFS was 4.2 months in the pembrolizumab arm compared to 2.4 months in the chemotherapy arm, and ORR was 23.5 percent for pembrolizumab compared to 9.8 percent for chemotherapy.
In KEYNOTE-042, TMB was again associated with improved outcomes in patients on pembrolizumab. The picture for patients on chemotherapy was a little more mixed in this study. Although there was no improvement in OS or PFS associated with TMB in patients treated with chemotherapy, there was an improvement in ORR.
Median OS in this study was 21.9 months for pembrolizumab compared to 11.6 months for chemotherapy and median PFS was 6.3 months for pembrolizumab compared to 6.5 months for chemotherapy. Median ORR in this study was 34.4 percent in the pembrolizumab arm and 30.9 percent in the chemotherapy arm.
While these results suggested that TMB may provide additional information regarding the clinical benefit of pembrolizumab monotherapy among patients with PD-L1-positive NSCLC, Herbst also said that the analyses and effect estimates were limited to observational subsets of the randomized patient cohorts.
In contrast, Luis Paz-Ares — head of the medical oncology service of the Madrid Hospital 12 de Octubre — presented findings from KEYNOTE-021, KEYNOTE-189, and KEYNOTE-407, which concluded that TMB was not significantly associated with efficacy of pembrolizumab plus platinum-based chemotherapy or of chemotherapy alone as a first-line therapy for metastatic NSCLC, regardless of histology.
Chemotherapy added to pembrolizumab improves its efficacy as a first-line therapy for metastatic squamous and non-squamous NSCLC, regardless of the tumor's proportion of PD-L1, Paz-Ares said.
The researchers determined TMB using whole-exome sequencing of tumor and matched normal DNA, and 175 mutations per exome was again used as the cutoff point.
Here, TMB was not significantly associated with the efficacy of the combination treatment or of chemotherapy alone. The combination of pembrolizumab and chemotherapy improved OS and PFS in KEYNOTE-189 and KEYNOTE-407 both in patients who had TMB of 175 or more and in patients who had TMB of less than 175. In KEYNOTE-021, patients with a TMB of less than 175 treated with the combination treatment had a median ORR of 60.8 percent compared with an ORR of 71.4 percent in patients with high TMB.
Importantly, the combination of pembrolizumab and chemotherapy showed survival benefit in both the TMB-high and TMB-low subgroups, for both squamous and non-squamous NSCLC, Paz-Ares said. Tissue TMB does not seem to differentiate responders from non-responders either for the combination treatment or chemotherapy alone, he added, and may have "limited clinical utility" in this setting.
In discussing these studies, Columbia University Irving Medical Center's Naiyer Rizvi said that the studies showed that TMB "isn't ready for primetime" as a biomarker. But, he also noted that the data has brought up some interesting questions for using pembrolizumab as a treatment for advanced lung cancer.
For example, he said, although doctors generally have no trouble prescribing pembrolizumab in patients with a PD-L1 proportion of 50 percent or more, they have more trouble deciding whether or not to prescribe it to patients with PD-L1 of 1 percent to 49 percent. KEYNOTE-010 and KEYNOTE-042 showed that a high TMB might be usable as a secondary biomarker in this group, he added.
He also noted that TMB-high patients with a PD-L1 proportion of 1 percent or more received a similar benefit from pembrolizumab monotherapy as from the combination of pembrolizumab and chemotherapy. This begs the question of what chemotherapy adds for these patients, and whether pembrolizumab alone might be a better treatment option in order to spare these patients the toxicities from chemotherapy.
He concluded that more data is also needed on treatment in patients who are TMB-high but who have a PD-L1 proportion of less than 1 percent.