A recent multicenter study highlights significant variability in PD-L1 scoring among pathologists analyzing non-small cell lung cancer (NSCLC) cytological specimens. Conducted across nine countries, the research underscores the pressing need for standardized evaluation protocols to ensure consistent and accurate diagnoses.
Study Design and Methodology
Researchers retrospectively gathered 65 cell blocks diagnosed with NSCLC, ultimately utilizing 54 suitable cores to construct four tissue microarrays (TMAs). Hematoxylin-eosin and PD-L1 stained slides underwent digitization before being shared via an open web platform. Thirty-one cytopathologists from 21 institutions participated, assessing PD-L1 expression using the tumor proportion score (TPS) cutoffs: below 1%, between 1% and 49%, and above 50%.
Findings Highlight Variability
The collective analysis revealed a moderate interobserver agreement with a Fleiss’s kappa value of 0.49. The highest concordance occurred in the TPS >50% category (k = 0.57), whereas the middle range (1%-49%) exhibited the lowest agreement (k = 0.32). These discrepancies signify challenges in accurately categorizing PD-L1 expression levels, potentially impacting treatment decisions.
• Standardization of sample preparation processes is crucial.
• Enhanced training programs for pathologists may improve consistency.
• Incorporating machine learning tools could aid in objective PD-L1 assessments.
The study’s outcomes emphasize the necessity for unified guidelines and advanced tools in evaluating PD-L1 on cytological samples. By addressing the identified variability, the medical community can ensure more reliable interpretations, ultimately leading to better-informed therapeutic strategies for lung cancer patients.
Implementing standardized protocols and leveraging technology can bridge the current gaps in PD-L1 scoring. Pathologists should advocate for continuous education and explore machine learning integrations to enhance diagnostic precision. Such measures will not only streamline the evaluation process but also contribute to more personalized and effective cancer treatments.
This article has been prepared with the assistance of AI and reviewed by an editor. For more details, please refer to our Terms and Conditions. We do not accept any responsibility or liability for the accuracy, content, images, videos, licenses, completeness, legality, or reliability of the information contained in this article. If you have any complaints or copyright issues related to this article, kindly contact the author.
