InterobServer AgreeMent in Pd-l1 evaLuatIoN on cytoloGical samples-SAMPLING project: A multi-institutional,...

Area of Science:

• Oncology
• Pathology
• Biomarker Analysis

Background:

• Accurate programmed death-ligand 1 (PD-L1) scoring is crucial for non-small cell lung cancer (NSCLC) treatment decisions.
• Cytological specimens offer an alternative to tissue biopsies for PD-L1 assessment.

Purpose of the Study:

• To evaluate interobserver agreement for PD-L1 scoring on NSCLC cytological specimens using a multicenter approach.
• To assess the reliability of PD-L1 evaluation on cell block-derived tissue microarrays (cbTMA).

Main Methods:

• Retrospective collection of 65 NSCLC cell blocks (CBs).
• Preparation of four tissue microarrays (TMAs) from 54 suitable CBs.
• Digitization of H&E and PD-L1 stained slides for review by 31 cytopathologists using tumor proportion score (TPS) cutoffs (<1%, 1%-49%, >50%).

Main Results:

• A total of 1674 evaluations were collected from 31 cytopathologists across 21 institutions.
• Moderate overall interobserver agreement was observed (κ = 0.49).
• Agreement varied by TPS category, with the highest for >50% (κ = 0.57) and lowest for 1%-49% (κ = 0.32).

Conclusions:

• Moderate agreement indicates a need for improved standardization in PD-L1 scoring on NSCLC cytological samples.
• Recommendations include standardized sample preparation, targeted training for cytopathologists, and exploring machine learning tools.
• Enhancing interobserver consistency is vital for reliable PD-L1 biomarker assessment in NSCLC.