VIENNA – Several different tests for PD-L1 levels in tumor cells of patients with metastatic non–small cell lung cancer showed high concordance when run at seven French centers, boosting confidence in the clinical utility of this testing to guide first-line pembrolizumab treatment of patients with this cancer.
Among 27 laboratory-developed tests for PD-L1 (programmed death–ligand 1) levels in tumor cells that used any one of three prespecified testing platforms (made by Dako, Ventana, or Leica), 14 (52%) had “similar” concordance when compared with reference assays, Julien Adam, MD, said at the World Conference on Lung Cancer, sponsored by the International Association for the Study of Lung Cancer. “These results will provide the basis for making national recommendations for PD-L1 testing in patients with non–small cell lung cancer [NSCLC]” in France, added Dr. Adam, a pathologist at Gustave Roussy Cancer Center in Paris. “We expect to produce recommendations by the second half of 2017.”
Several single-center studies had examined harmonization of several different PD-L1 tests, but the new, multicenter study examined several different antibodies and platforms systematically, he noted.Although the results came entirely from French centers, the results will also likely influence U.S. practice, predicted Shirish M. Gadgeel, MD, a thoracic oncologist at the Karmanos Cancer Institute in Detroit. The approval pembrolizumab (Keytruda) received from the Food and Drug Administration in October specified that patients with metastatic NSCLC had to show PD-L1 expression in the tumor using a FDA-approved test to receive pembrolizumab as first-line treatment.
“Before pembrolizumab’s approval there was no incentive to do PD-L1 testing,” but now it is becoming routine, he said. “It has been challenging to U.S. laboratories to decide which platform and antibody to use. Harmonization gives us confidence that if you have a platform and appropriate antibody you should be able to use the results clinically. I think the French results can be extrapolated” to U.S. practice because the results “came from multiple labs using multiple antibodies and platforms,” Dr. Gadgeel said.
The new French study “is unique by being very real-world, in one country across multiple institutions,” commented Julie R. Brahmer, MD, director of thoracic oncology at the Johns Hopkins Kimmel Cancer Center in Baltimore. “It behooves every physician [caring for advanced NSCLC patients] to know what a PD-L1 test means when they get a result.” An ongoing effort to assess performance of PD-L1 testing at U.S. centers began in late 2015 by the National Comprehensive Cancer Network.The French study arranged for PD-L1 testing of NSCLC specimens from 41 patients selected as broadly representative of PD-L1 expression levels. The seven participating centers used a Dako (three centers), Ventana (two centers), or Leica (two centers) testing platform and one of five available antibodies that bind PD-L1. Every center ran tests that depended on different antibodies, and tests occurred on both tumor cells and immune cells. In total the seven testing sites collectively performed 35 stainings on each specimen for a total of 1,435 slides. In tumor cells, the overall, weighted kappa coefficient for concordance was 0.81 for tests that used the SP263 antibody and 0.78 for those using the E1L3N antibody, both high enough to make them potential candidates for routine use, said Dr. Adam. The 28-8 and 22C3 antibodies also showed high concordance levels. In contrast, some antibodies used at certain centers produced unacceptable results with concordance coefficients of less than 0.5. The best performing antibody overall was SP263.
No test measuring PD-L1 level in immune cells had an acceptable concordance rate, Dr. Adam also reported.
On Twitter @mitchelzoler