Interobserver- and interlaboratory-concordance of PD-L1 IHC for NSCLC
Immunohistochemistry (IHC) of the PD-L1 protein has become a mandatory diagnostic test in the treatment of non-small-cell lung cancer (NSCLC). Several research-initiatives have been started to harmonize the five PD-L1 IHC clinical trial assays (CTAs). In Germany, a two-step round robin test was conducted to test interobserver- and interlaboratory-concordance of PD-L1 IHC and to compare four CTAs and laboratory-developed assays.
As first step, interobserver-concordance was tested by central IHC-staining of 2 x 15 NSCLC resection specimens that were scored independently by nine pathologists. Proportions of PD-L1 positive carcinoma cells were scored by a 6-step system that integrates the clinically validated cut-offs. Scoring of the carcinoma cells yielded moderate interobserver-concordance coefficients for the 6-step scoring system (Light’s kappa=0.47-0.50) while included dichotomous cut-offs ≥1% and ≥50% showed substantial concordance coefficients (κ=0.6-0.8). Major challenges in PD-L1 scoring were cases with low staining-intensity and the differentiation of PD-L1 positive cancer and immune cells.
As second step, interlaboratory-concordance was assessed by a centrally prepared tissue-microarray containing 21 NSCLC specimens. Precut tissue-sections were stained at ten sites using CTAs 28-8, 22C3, SP263 and SP142 as well as laboratory-developed assays. Assay-performance was evaluated with a second tissue-microarray containing eleven cell-lines with defined PD-L1 expression. Quality-control was centrally performed by manual and digital analyses.
The CTAs yielded reproducible IHC-stainings at all sites while the results of the laboratory developed assays were mixed: Six protocols showed appropriate IHC quality with staining patterns similar to 22C3 and 28-8 CTAs, five protocols yielded less DAB-deposits and reduced staining intensity. Interlaboratory-concordance of carcinoma cell scoring using the 6-step system was moderate (κ=0.43-0.69) while the included cut-offs ≥1% and ≥50% showed substantial concordance for the CTAs (κ=0.73-0.89) and moderate concordance for the laboratory-developed assays (κ=0.50).
In both tests, no differences in interobserver- and interlaboratory-concordance were found among the CTAs. However, differences in the resulting staining patterns were noticed: While 22C3 and 28-8 showed similar staining patterns, SP263 showed minor differences in some cases and SP142 showed distinct patterns.
Taken together the data show that the PD-L1 CTAs can be reproducibly employed and scored at different sites. Laboratory-developed assays are possible yet have to be carefully calibrated to match the appropriate intensity-range. The choice of assay and the set-up of the IHC-protocol may strongly influence the resulting staining. Standardized IHC-staining and interpretation both are prerequisites for reliable PD-L1 IHC.
Dr. Scheel is resident pathologist at the Institute of Pathology, University Hospital Cologne, Germany, and member of the Lung Cancer Study Group Cologne . His diagnostic focus is biomarker testing in lung cancer, in particular, PD-L1 immunohistochemistry. He contributes to external quality assessment by the QuIP  and is part of the German PD-L1 harmonization study [3,4].
In 2011 Dr. Scheel received his medical degree from the University of Göttingen, where he conducted a medical thesis on microRNAs in the department of molecular oncology lead by Dr. Matthias Dobbelstein. He started his residency at the Institute of Pathology Nordhessen, Kassel, director: Prof. Josef Rüschoff and specialized in biomarker-validation . In 2014 he moved to Cologne and joined the Institute of Prof. Reinhard Büttner. His areas of research are histology-derived biomarkers and image analysis.
- Mod Pathol 2016. 29(10):1165-72
- Pathologe 2016. 37(6):557-567
- Mod Pathol 2015. 28(12):1528-34