×

PD-L1 expression varies dramatically within ccRCC tumors

BY WILL PASS
MDedge News

FROM ANNALS OF DIAGNOSTIC PATHOLOGY

PD-L1 expression levels vary dramatically within clear cell renal cell carcinoma (ccRCC) tumors, according to reevaluation of ccRCC samples surgically removed during radical nephrectomy.

Furthermore, PD-L1 expression in the tumor and nonexpression in thrombus-associated samples is linked to aggressive behavior in ccRCC, researchers reported in Annals of Diagnostic Pathology.

Checkpoint inhibitors that target PD-L1, such as nivolumab and atezolizumab, have shown encouraging results for ccRCC patients in recent trials by reducing tumor size and prolonging survival, but responses do not always correlate with PD-L1 expression in biopsy samples, they said.

“Unexpectedly, around 17% of PD-L1 negative patients respond to anti-PD-L1 therapy,” wrote José López, MD, of Cruces University Hospital in Barakaldo, Spain, and his coauthors. They added that “insufficient tumor sampling” could be the cause of these unexpected responses. Clear cell RCC tumors are known to have highly irregular tissue that can also change over time; this study suggests that these irregularities may also apply to PD-L1 expression status.

The retrospective study evaluated 39 cases of advanced ccRCC with renal vein/caval tumor thrombus. Thirty patients were male and nine were female, with a median age of 67 years. Samples were surgically removed during radical nephrectomy at Cruces University Hospital between 2007 and 2016.

Each sample was reevaluated by a pathologist, who reassigned stage and grade, then looked for characteristics of aggressive tumor behavior (nodal status, tumor necrosis, and sarcomatoid change). Age, sex, and presence/absence of metastasis were also recorded.

Researchers assessed PD-L1 expression levels without associated case information. Tissue microarrays were constructed from three locations within the main tumor and two locations where tumor tissue was associated with the thrombus. Anti-PD-L1 (SP-142 ) monoclonal antibody was used to detect PD-L1 expression. A location was considered positive for PD-L1 expression when more than 1% of tumor-infiltrating inflammatory cells stained positive.

In 56.4% of cases, at least one location in the main tumor tested positive for PD-L1, but only 7.6% of cases were positive in all three locations. This points to significant variation within individual ccRCC tumors – if one area of a tumor tests positive for PD-L1, another site could very well test negative.  

In main tumors with at least one positive site, only 10.25% also had at least one positive thrombus site. Just over 46% of cases were consistently negative at the thrombus but positive at the main tumor. These results raise concerns that thrombus-associated samples should be interpreted with caution. Roughly half of them did not accurately represent PD-L1 status.

The authors suggested that the thrombus microenvironment in ccRCC is characterized by a loss of PD-L1 expression. “This could also explain why immune checkpoint blockade is effective in only a fraction of patients with renal cell carcinoma,” they wrote.

Furthermore, PD-L1 positive tumors were more likely to have negative PD-L1 expression at the thrombus site if they were higher grade (P = .01), suggesting that thrombus negativity could also be a sign of aggressive behavior.

PD-L1 as a prognostic marker for ccRCC is already a controversial topic. This dramatic variation of PD-L1 levels within a single tumor could significantly limit predictive value. Further implications may relate to tissue sampling; multiple samples may be necessary to form a complete picture of PD-L1 expression.

In the current study, approximately 90% of tumors were high grade. A tendency was observed toward PD-L1 expression in high grade tumors (P = .05) and tumors that had metastasized at the time of diagnosis (P = .07), but PD-L1 expression was not an independent prognostic indicator.

This study is the first to demonstrate a significant geographic variation of PD-L1 levels within ccRCC tumors, which “reinforces the importance of appropriately selecting the sample to be tested,” the authors advised, “and the need for a more efficient sampling strategy, particularly once atezolizumab [is] approved to treat advanced ccRCC patients.

“Understanding the complexity of the inflammatory tumor microenvironment is mandatory before the clinical value of PD-L1 inhibition can be totally ascertained,” Dr. Lopez and his coauthors concluded.

Tissue microarrays were constructed from three locations within the main tumor and two locations where tumor tissue was associated with the thrombus. A location was considered positive for PD-L1 expression when more than 1% of tumor-infiltrating inflammatory cells stained positive.