PD-L1 is an imperfect biomarker, according to formal discussant of this paper, D. Ross Camidge, MD, PhD, of the University of Colorado Comprehensive Cancer Center, Denver.

“The importance of a good predictive assay in this field [of immunotherapy] is so great that it is impossible to ascribe any true differences in efficacy between drugs separate from the degree of enrichment their associated assays produce,” he stated.

Three different companies are developing companion PD-L1 assays for their respective anti–PD-1 compounds. In the case of pembrolizumab, the Merck assay identifies about 23% of lung cancers with a high performance status (PS) score in whom PD-1 inhibition is associated with an overall response rate of 42%. But you cannot say the drug works any better than one with a lower response rate, unless it is being compared in the same enriched population, Dr. Camidge emphasized.

“It is impressive that this assay appears to identify a group with prolonged progression-free survival that is approximately the same size as the responder group. But, as mentioned, it is not a perfect assay. In all of the groups including the high PD-L1 group, approximately 30% to 40% of patients are still rapid progressers on therapy,” he explained.

Looking at overall survival, the biomarker-positive group has impressive survival, he continued. “But as the overall survival tail seems to exceed the size of the groups that are either responding or having prolonged progression-free survival, two hypotheses present themselves: either PD-L1 expression itself is a good prognostic factor, or, alternatively, the biomarker may be predicting for ongoing benefit after the therapy is finished, including potentially modifying outcomes from subsequent lines of therapy,” he continued.

Even though the biomarker is imperfect, it can be useful, according to Dr. Camidge. “When we don’t know in whom a drug works, this does not mean either that we shouldn’t give it to anyone until we do or that we should give it to everybody. Even an imperfect enrichment assay can dramatically alter the ‘bang for the buck’ a drug offers in different settings, addressing the increasingly common hurdle of health economics that exists in health systems around the world.”

In addition, Dr. Camidge concurred with Dr. Topalian that an imperfect enrichment biomarker could be used most pragmatically to prioritize the selection of therapy with the greatest chance of success at each point in the treatment continuum.

For example, in the group with a PS ≥ 50%, pembrolizumab holds the potential to be a better first- and second-line therapy than chemotherapy. In the group that has a PS of between 1% and 49%, first-line therapy would probably be chemotherapy, but pembrolizumab might be prioritized over second-line chemotherapy. In the biomarker-negative group, pembrolizumab might be deprioritized to a salvage therapy.

“All of this would have to be proven prospectively,” he emphasized. ■

Disclosure: Dr. Camdige reported no potential conflicts of interest.