Three digital core biopsies are usually enough to determine programmed death ligand 1 (PD-L1) expression in non-small-cell lung cancer (NSCLC), and 4 or more biopsies proffer little additional benefit. These are the findings of a study presented during a poster session at the International Association for the Study of Lung Cancer (IASLC) 19th World Conference on Lung Cancer, held September 23–26 in Toronto.
In NSCLC, assessment of PD-L1 expression on tumor cell membranes via immunohistochemistry helps guide the utilization of checkpoint inhibitors. However, the number of digital biopsies needed to assess PD-L1 expression is obfuscated by the temporal and spatial heterogeneity of PD-L1 expression.
“Immunohistochemistry staining for PD-L1 is an important test for determining patient eligibility for first-line, single-agent immunotherapy, which requires a cutoff of 50% tumor cell staining,” Dara L. Aisner, MD, PhD, associate professor of pathology at the University of Colorado, confirmed in an exclusive interview with Cancer Network.
The study, led by Alex Haragan and colleagues at the Royal Liverpool University Hospital in Liverpool, England, analyzed the expression of PD-L1 using sections from 94 tissue blocks taken from 50 primary pulmonary adenocarcinomas. Initial scoring was done by two trained pathologists. Of the 50 participants, 14 individuals had a PD-L1 score of < 1%; 13 had a score of 1-10; 10 had a score of 11-49; and 13 had a score of 50-100.
Following this conventional assessment, the slides were scanned to create a digital database comprising 1 mm2-area squares. Squares were assigned coordinates and rescored to create several digital core biopsies, which were simulated in a sequential manner to approximate biopsy samples taken by a 17-gauge needle. The researchers then compared expression in these digital core biopsies with the whole tumor and categorized them according to UK prescribing guidelines.
In all 50 cases, increasing the number of digital core biopsies from 3 to 4 did not change tumor categorization in any cases, and boosting the number to 5 resulted in tumor recategorization in only one case. Moreover, at the all-important 1% cutoff, upping the number of digital core biopsies from 3 to 4 or from 4 to 5 resulted in no changes.
The team found the weakest correlation with the whole tumor when PD-L1 expression was low and focal, which is an important concern in light of the 1% cutoff typically used for checkpoint therapy.
“Using this model as a guide, a single good quality biopsy (2x10mm2 area) is sufficient for most tumours scoring 11% or greater PD-L1 expression,” concluded the authors. “However, in the lower range of expression, rebiopsy might be routinely considered if there is doubt about specimen adequacy.”
Aisner reflected on the results of this study. “Haragan et al nicely demonstrated through ‘digital image down-sampling’ that a single high-quality core biopsy is sufficient to confidently identify cases that make [50% tumor cell-staining] cutoff,” she said. “However, it raises clear concern that, for patients with tumors with lower levels of expression, a single biopsy may not properly classify the next cutoff of 1%, which can be used to determine subsequent lines of therapy.”
“This draws important attention to the clinical approach for cases in which the tumor is <1% staining and the potential need for rebiopsy,” Aisner said. However, she noted, “it does not address how the testing of cytopathology specimens [or cell blocks] might also underrepresent PD-L1 status, which will be an important next step.”