For some patients, immune checkpoint inhibitors yield remarkable and durable solid-tumor responses. But for most, benefits prove elusive, despite the risk of potentially life-threatening treatment-associated immune-related adverse events (IRAEs). The development of biomarker-based tools that can be used to identify patients who may benefit from immunotherapy, therefore, remains an urgent goal.

An exploratory analysis of data from the nonrandomized phase 1b KEYNOTE-028 study (ClinicalTrials.gov identifier: NCT02054806), sponsored by Merck Sharp & Dohme Corp., offers a “step in the right direction,” said lead study author Patrick Ott, MD, PhD, Clinical Director of the Center for Immuno-Oncology at Dana-Farber Cancer Institute in Boston, Massachusetts.

The study included 475 patients with 20 types of cancer.1 The authors analyzed tumor mutational burden (TMB), a T-cell inflammation gene-expression signature, and PD-L1 expression — 3 candidate predictive biomarkers identified in previous studies.

Related Articles

Unlike targeted therapies, for which treatment outcome is strongly associated with the presence or absence of a particular cancer gene mutation, immunotherapy outcomes hinge importantly on hundreds of factors, some estimate, from tumor biology to the tumor microenvironment. That complicates the search for a predictive test or algorithm to identify which patients could reasonably expect outcomes that justify the risk of IRAEs. So it makes sense to explore associations between several candidate biomarkers and treatment outcomes.

The study authors found associations between the aforementioned markers and treatment outcomes. But the correlations were relatively weak, Dr Ott cautioned.

“The biggest take-home message is probably that patients with both high mutation burden and inflammatory markers separate out as having the highest likelihood of response to pembrolizumab,” he said.

PD-L1 expression level was not available for all patients and was included as a “signal-finding” endeavor, he noted.

“But you can basically lump the inflammation gene-expression profile and PD-L1 expression together as a marker of tumor ‘hotness’,” Dr Ott told Cancer Therapy Advisor.

That held true overall, across tumor types, but the preplanned exploratory analyses lacked the statistical power to examine just how predictive these biomarkers were of patient outcomes for each individual type of cancer, Dr Ott acknowledged.