Êtes-vous un étudiant de l'EPFL à la recherche d'un projet de semestre?
Travaillez avec nous sur des projets en science des données et en visualisation, et déployez votre projet sous forme d'application sur Graph Search.
Introduction: Phosphatase and tensin homolog (PTEN) loss is frequently observed in NSCLC and associated with both phosphoinositide 3-kinase activation and tumoral immunosuppression. PTEN immunohistochemistry is a valuable readout, but lacks standardized staining protocol and cutoff value. Methods: After an external quality assessment using SP218, 138G6 and 6H2.1 anti-PTEN antibodies, scored on webbook and tissue microarray, the European Thoracic Oncology Platform cohort samples (n = 2245 NSCLC patients, 8980 tissue microarray cores) were stained with SP218. All cores were H-scored by pathologists and by computerized pixel-based intensity measurements calibrated by pathologists. Results: All three antibodies differentiated six PTEN+ versus six PTEN- cases on external quality assessment. For 138G6 and SP218, high sensitivity and specificity was found for all H-score threshold values including prospectively defined 0, calculated 8 (pathologists), and calculated 5 (computer). High concordance among pathologists in setting computer-based intensities and between pathologists and computer in H-scoring was observed. Because of over-integration of the human eye, pixel-based computer H-scores were overall 54% lower. For all cutoff values, PTEN-was associated with smoking history, squamous cell histology, and higher tumor stage (p < 0.001). In adenocarcinomas, PTEN-was associated with poor survival. Conclusion: Calibration of immunoreactivity intensities by pathologists following computerized H-score measurements has the potential to improve reproducibility and homogeneity of biomarker detection regarding epitope validation in multicenter studies. (C) 2018 International Association for the Study of Lung Cancer. Published by Elsevier Inc. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
Xile Hu, Li Tang, Céline Jasmin Prange