KEYNOTE-028 Trial Unveils Biomarkers Linked to Pembrolizumab Efficacy

January 16, 2019

The results of the study add to the growing body of literature regarding potential biomarkers for response to immune checkpoint inhibitors.

Two inflammatory biomarkers and tumor mutational burden (TMB) were associated with improved clinical response to pembrolizumab across several solid tumor types, the KEYNOTE-028 trial showed (ClinicalTrials.gov identifier: NCT02054806). Trial results were recently published in the Journal of Clinical Oncology.

“This interesting study adds to the growing body of literature regarding potential biomarkers for response to immune checkpoint inhibitors,” Daniel Y. Wang, MD, an assistant professor of medicine and hematology/oncology at Baylor College of Medicine and a medical oncologist at the Dan L Duncan Comprehensive Cancer Center at Baylor College of Medicine, told Cancer Network.

The KEYNOTE-028 trial is a single-arm, phase Ib basket trial that evaluated pembrolizumab in 20 cohorts of patients with a range of advanced solid tumors positive for programmed death ligand 1 (PD-L1). Exploratory study endpoints included the association between pembrolizumab efficacy and two inflammatory biomarkers and TMB. The inflammatory biomarkers were T-cell–inflamed gene-expression profile (GEP) and PD-L1 expression.

A total of 471 patients had measureable disease at baseline and received at least one dose of pembrolizumab. Patients had a median age of 59 years (range, 18–87 years), and women made up a slight majority (59.2%). Most patients were white (60.2%) or Asian (20.8%). Nearly all patients had an Eastern Cooperative Oncology Group performance status of 0 or 1 (98.3%). Most patients received at least one line of prior therapy (85.6%).

For the biomarker analysis, not all patients had available biomarker data from their tumor samples. In addition, the biomarker analysis did not include pancreatic cancer because none of those patients responded to pembrolizumab.

An analysis of tumors from 313 patients revealed that T-cell–inflamed GEP scores were greater in patients who achieved response and had better progression-free survival (PFS) across all tumor types evaluated in the cohort. The T-cell–inflamed GEP was made up of 18 genes, and all except one gene were positively associated with response rate.

An analysis of tumors from 198 patients demonstrated that PD-L1 expression was significantly correlated with response rate (P = .018) and longer PFS (P = .005). PD-L1 expression had a moderate, statistically significant association with T-cell–inflamed GEP (r = 0.40; P = .001).

An analysis of tumors from 77 patients showed that higher TMB was significantly correlated with response rate (P = .018) and a longer PFS (P = .51). TMB had a low, statistically significant association with T-cell–inflamed GEP (r = 0.29; P = .007).

Overall, patients with both higher TMB and higher T-cell–inflamed GEP or PD-L1 expression had an even greater likelihood of achieving a response.

“It was interesting to see that the associations between clinical efficacy and biomarkers held true despite a heterogeneous population of different cancer types,” Wang said. “Additionally, there was modest correlation between the individual biomarkers.”

The modest correlation, he said, suggests that the integration of biomarkers for inflammation (ie, T-cell–inflamed GEP or PD-L1 expression) with mutation burden may help to predict responders to anti–PD-1 therapy. “These findings will be informative for future studies in these various cancer types in identifying patients who will respond to anti–PD-1 therapy,” he said.

While the study included a large number of patients, Wang pointed out that there were less than 30 patients for each type of cancer, limiting the generalizability of the findings for each particular cancer.

“At this stage, the biomarkers in these studies are not ready for the clinic in these tumor types as they need to be validated in larger, prospective randomized clinical trials to help identify disease-specific thresholds and cut-offs that effectively predict response to [immune checkpoint inhibitors],” he cautioned. He noted that TMB has been used as a potential biomarker to identify immunotherapy candidates in non–small-cell lung cancer, but no consensus has been made regarding particular assays and cut-off levels. “With more biomarkers on the horizon, the integration of these different assays must be carefully evaluated.”