Suzanne L.Topalian, MD
The wide variability in clinical outcomes among patients undergoing anti-PD-1/PDL1 immunotherapy has increased interest in finding biomarkers that predict response. Although tumor expression of PD-L1 is classically used as a biomarker, complexities associated with its expression and measurement in the tissue suggest that PD-L1 status should be used in conjunction with other biomarkers to best predict patients who would benefit.
That is the latest thinking of leading experts in the emerging immune checkpoint blockade immunotherapy field, according to several presentations at the 2016 ASCO Annual Meeting.
“We believe these biomarkers are more highly predictive and potent if they can be used in combination,” said Suzanne L.Topalian, MD, of the Sidney Kimmel Comprehensive Cancer Center at Johns Hopkins School of Medicine and a presenter at the conference. “Moving forward, we’re going to examine these areas of intersection.”
Topalian presented data from a 2012 study showing that more than one-third of patients with PD-L1-positive metastatic melanoma, non–small-cell lung cancer (NSCLC), or renal cell cancer had an objective response to PD-1 blockade, whereas none of the patients with PD-L1–negative tumors responded.1
Another study showed that pembrolizumab improved objective response and overall survival (OS) in patients with NSCLC that highly expressed PD-L1 (≥50% of tumor cells),2
and a pooled analysis of seven studies showed that patients with PD-L1-positive NSCLC (PD-L1 staining on ≥1% of cells) had a greater overall response rate to PD-1 inhibitor therapy than those with PD-L1–negative tumors.3
Taken together, data in these studies suggested that tumor expression of PD-L1 is a strong predictor of response to PD-1 blockade, according to Topalian.
However, Topalian identified more recent data that indicated PD-L1 tumor expression does not predict all responders to anti-PD-1/PD-L1 therapy. A 2015 review of all studies published on PD-1/ PD-L1 blockade showed that approximately 15% of patients with PD-L1–negative tumors responded to anti-PD-1 therapy.4
While this was significantly lower than the response rate in patients who had PD-L1–positive tumors (45%), Topalian stated that PD-L1 status alone is insufficient to exclude patients from receiving anti-PD-1/PD-L1 therapy.
Complexities of Measuring PD-L1
Several presenters noted multiple complexities inherent to the measurement and expression of PD-L1. Multiple types of cells in the tumor microenvironment, including infiltrating macrophages and activated lymphocytes, also express PD-L1, and immunohistochemistry (IHC) assays for PDL1 differ in the types of cells (immune-infiltrating cells, tumor cells, or both) they include when scoring PD-L1 expression.
Ravindra Uppaluri, MD, PhD, of Washington University School of Medicine in St. Louis, noted that for patients with recurrent/ metastatic head and neck squamous cell carcinoma in the KEYNOTE-012 trial,5
PD-L1 expression significantly predicted response to pembrolizumab when analyzed as a combined proportion score (expression in tumor and inflammatory cells) but not when analyzed as a tumor composite score (expression in tumor cells only).
Furthermore, he showed that 10 of the 29 patients with PD-L1-positive inflammatory cells and PD-L1-negative tumor cells responded to anti-PD-1 therapy—patients who would have been considered to be PD-L1-negative if only tumor cells had been included.
Additionally, Topalian stated that PD-L1 expression may vary between primary and metastatic lesions in patients with multiple metastases.
She presented a case of a patient with metastatic melanoma who achieved a complete response to anti-PD-1 therapy, yet only had PD-1 expression at one of the three tumor sites. “This patient could have easily been excluded from therapy if [PD-L1 expression] was used as a literal companion diagnostic. We don’t believe that it should be used as a requirement for therapy.”
Topalian further stated that PD-L1 expression is not homogeneous throughout the tumor microenvironment but tends to be concentrated at the boundaries between tumor cells and infiltrating stromal cells. “If you only had a needle biopsy for your diagnostic, you could easily miss the area of PD-L1-positive expression,” she said.
Janice M. Mehnert, MD, of the Rutgers Cancer Institute of New Jersey, also emphasized in her presentation that the highly dynamic nature of the immune system and tumor microenvironment may lead to a change in PD-L1 expression over time, and thus timing of the biopsy relative to initiation of treatment is important.