Immune checkpoint inhibitors stole the show at the European Society of Medical Oncology (ESMO) 2017 conference in Madrid, Spain. Although these therapies have been effective in a number of tumor types with long-term survival benefits, proper patient selection is still an active area of research. It is a high priority to better understand the PD-L1 biomarker in order to influence its expression and improve therapeutic responses to immuno-oncology (I-O) treatments through the use of combination regimens and sequencing of therapies.
High tumor expression of PD-L1 has been shown to correlate with positive responses to PD-1/PD-L1 inhibitors, and Keytruda’s approval as a first line monotherapy in non-small cell lung cancer (NSCLC) requires a companion diagnostic to select patients with high expression of the biomarker. While not required in other settings or for other PD-1/PD-L1 inhibitors, testing may still be ordered by oncologists as a complementary diagnostic, in order to inform treatment decisions on which patients are most likely to respond. Expression is still incompletely understood, however, as some patients determined to be PD-L1 low or PD-L1 negative may still respond to therapy, and others who are determined to be high-expressers may not respond.
The complexities of the assay and tissue expression of PD-L1 were underscored in an industry-run session by Bristol Myers Squibb. PD-L1 testing requires a tissue sample and is carried out by immunohistochemistry (IHC), which is offered by various commercial laboratories or determined at in-house pathology labs. There are four different antibodies (clones) used in combination with two separate automated platforms, Ventana Ultrasystems or DAKO Autolink 28. Importantly, antibody clones and platforms are not considered interchangeable. Error may also be introduced due to the somewhat finicky nature of IHC, with sample handling and preparation prior to the assay being a key factor in its performance. Finally, human error is introduced at the final stage of the assay when a pathologist determines the PD-L1 expression and assigns a score to the sample. Inter-pathologist variability is a recognized complication of these assays, especially when tissue expression is borderline and in the low range.
The complexity of the PD-L1 biomarker is underappreciated for several reasons. First, it must be recognized that the expression level is not binary, but is in actuality a spectrum. Nonetheless, expression is segmented into PD-L1 “high”, “low”, and “negative” subsets. Precise cut-off for inclusion into each category is debatable, and may come down to the pathologist’s best assessment. Furthermore, PD-L1 presence in the tumor tissue is varied, it may be expressed by the tumor cells or by different immune cells, is variable in different regions of the tissue, and changes over time or in response to stimuli.
Given the complexities of IHC testing for PD-L1, research into improved measurement techniques continues to provide a better understanding of this important biomarker. It is unreasonable to eliminate all sources of error, however, as the tumor itself is dynamic with variable expression over time and throughout the tissue sample. Nevertheless, researchers are hopeful that biomarker tests in the future will be interchangeable and have fewer sources of error.