23P - PD-L1 IHC validation and comparison of E1L3N & SP142 mAbs in melanoma

Date 20 November 2015
Event ESMO Symposium on Immuno-Oncology 2015
Session Welcome reception and general Poster viewing
Topics Cancer Immunology and Immunotherapy
Melanoma and other Skin Tumours
Presenter Kelly Schats
Citation Annals of Oncology (2015) 26 (suppl_8): 5-14. 10.1093/annonc/mdv514
Authors K. Schats1, E.A.C. van Vré2, S. De Schepper2, B. Neyns3, I. De Meester4, M. Kockx5
  • 1University Of Antwerp, HistoGeneX NV, 2020 Antwerp - Antwerp/BE
  • 2Immunohistochemistry, HistoGeneX NV, 2020 Antwerp - Antwerp/BE
  • 3Medical Oncology, Vrije Universiteit Brussel-Campus Jette, 1090 - Brussels/BE
  • 4Medical Biochemistry, University of Antwerp, 2610 - Antwerp/BE
  • 5Pathology & Imaging, HistoGeneX NV, 2020 Antwerp - Antwerp/BE

Abstract

Aim

The identification of predictive markers for cancer immunotherapy (e.a programmed death (PD)-1 and its ligand PD-L1) is becoming critical. Working under stringent quality requirements PD-L1 immunohistochemistry (IHC) assays were developed. The E1L3N and SP142 clones were selected to independently set up and validate IHC assays. This study aims to compare the staining patterns of both clones in melanoma metastasis as well as their relation to PD-L1 mRNA.

Methods

The IHC assays were optimized on a Ventana Benchmark Ultra (E1L3N) and XT (SP142), using the same detection system. Validation included accuracy testing on characterized control cell pellets and tissue, and precision testing on 5-6 tumor samples. In total 47 melanoma metastases were stained with the IHC assays. Stained slides were digitized for qualitative and quantitative evaluation. A qualified pathologist scored PD-L1 membrane staining of immune (IC) and tumor (TC) cells (blinded). Absolute scorings as well as 7 scoring categories (published cut offs) were compared. A subset of 21 melanoma metastases was evaluated for PD-L1 mRNA by Nanostring technology.

Results

Accuracy was met for both E1L3N and SP142. Both clones passed precision criteria (> 80% visual concordance). Though, SP142 displayed slightly better intra- and inter-run variability. In general, the E1L3N and SP142 IHC showed considerable overlap of staining in melanoma. This was confirmed by pathologist (absolute) scoring: Spearman correlations (IC:0,94;TC:0,93) and inter class correlation coefficient values (IC:0,89;TC:0,84). For 5 out of the 7 applied cut-offs, a concordance of > 85% was obtained. Two resulted in a lower concordance (65-75%). Detailed evaluation demonstrated more TC and/or macrophage staining with E1L3N, as confirmed by pathologist scoring (Bland Altman plots). Correlation of PD-L1 IHC and Nanostring showed Spearman correlations of 0,94 (E1L3N) and 0,86 (SP142).

Conclusions

Validated E1L3N and SP142 IHC assays reveal good overlap in staining patterns and pathologist scores. Both assays show high correlation with mRNA. The clinical and technical relevance of the differences in TC staining between both assays needs further investigation.

Clinical trial identification

Disclosure

K. Schats, E. van Vré and S. De Schepper: Are employees of HistoGeneX NV, Antwerp. M. Kockx: Is CEO of HistoGeneX NV, Antwerp. All other authors have declared no conflicts of interest.