Author of

• 18972

  +

  P2.02 - Biology/Pathology (ID 616)

  • Event: WCLC 2017
  • Type: Poster Session with Presenters Present
  • Track: Biology/Pathology
  • Presentations: 1
  • Moderators:
  • Coordinates: 10/17/2017, 09:30 - 16:00, Exhibit Hall (Hall B + C)

  • 23241

    +

    P2.02-043 - Multicentre Assessment of PD-L1 Immunohistochemistry: Challenges for Establishing the Concordance Between Four Different Antibodies (ID 10275)

    09:30 - 09:30  |  Author(s): H.H.N. Pham
    • Abstract

    Background:
    PD-L1 status in lung cancer may be a predictive biomarker for the use of PD1/PD-L1 inhibitors. Nonetheless, the reproducibility of PD-L1 staining using different antibodies and platforms is still a matter of debate. We investigated the performance of four different antibodies with two different platforms in two different institutions.
    
    Method:
    PD-L1 expression was assessed with IHC (AB clone SP142 and SP263 - Ventana, 22C3 and 28-8 - Dako) on archival FFPE surgical tumour specimens, arrayed on tissue microarrays (TMAs) with duplicated 1 mm cores from two institutions (Karolinska University Hospital and Nagasaki University Hospital). All patients (n = 682) underwent curative surgery between 1987 and 2015. PD-L1 staining was scored as positive if present in ³1% of tumor cells, independently of staining intensity. The slides were scanned and scored by seven experienced pathologists who estimated the expression of PD-L1 in the tumour and inflammatory cells. The statistical analyses were performed to compare the four different antibodies and the scoring of the pathologists on the tumour and inflammatory cells.
    
    Result:
    PD-L1 positivity ³1% of tumor cells using clones 28-8 / 22C3 / SP263 / SP142 was 32%/29%/33%/9%, respectively. All carcinoids were negative for PD-L1. SP142 showed a significantly lower mean score compared with other clones. PD-L1 positivity ³50% of tumour cells using clones 28-8 / 22C3 / SP263 / SP142 was 11%/10%/13%/3%, respectively. The pair comparison analysis in the tumour cells showed that the score from the highest agreement was between 28-8 and 22C3 (Kappa 0.75, CI95% 0.70-0.80) and the lowest concordance was between SP263 and SP142 (Kappa 0.21, CI95% 0.16-0.27). The evaluation of the concordance in the inflammatory cells and among the pathologists is ongoing.
    
    Conclusion:
    The present study shows a comparative analysis using four different antibodies. The clone SP142 shows significantly lower expression of PD-L1 in the tumour cells. The clones 28-8, 22C3 and SP263 showed an excellent concordance in the tumour cells with two different cut offs (>1% an >50%) showing a good reproducibility for the pathology assay. The highest agreement was between the clones 28-8 and 22C3. The lowest concordance was between the clones SP263 and SP142.