Assessment of the current and emerging criteria for the histopathological classification of lung neuroendocrine tumours in the lungNENomics project
É. Mathian,Y. Drouet,A. Sexton-Oates,M.G. Papotti,G. Pelosi,J.-M. Vignaud,L. Brcic,A. Mansuet-Lupo,F. Damiola,C. Altun,J.-P. Berthet,C.B. Fournier,O.T. Brustugun,G. Centonze,L. Chalabreysse,V.T. de Montpréville,C.M. di Micco,E. Fadel,N. Gadot,P. Graziano,P. Hofman,V. Hofman,S. Lacomme,M. Lund-Iversen,L. Mangiante,M. Milione,L.A. Muscarella,C. Perrin,G. Planchard,H. Popper,N. Rousseau,L. Roz,G. Sabella,S. Tabone-Eglinger,C. Voegele,M. Volante,T. Walter,A.-M. Dingemans,L. Moonen,E.J. Speel,J. Derks,N. Girard,L. Chen,N. Alcala,L. Fernandez-Cuesta,S. Lantuejoul,M. Foll
DOI: https://doi.org/10.1016/j.esmoop.2024.103591
IF: 6.883
2024-06-17
ESMO Open
Abstract:Current LNET classification criteria show imperfect prognostic value and moderate reproducibility in a large cohort. Most misclassifications are due to the poor reproducibility of mitotic counts on haematoxylin and eosin slides. Ki-67 and PHH3 expression could complement the mitotic criteria in the classification system. Automated quantification of these markers using deep learning matches expert pathologists' annotation performance. Six thoracic pathologists reviewed 259 lung neuroendocrine tumours (LNETs) from the lungNENomics project, with 171 of them having associated survival data. This cohort presents a unique opportunity to assess the strengths and limitations of current World Health Organization (WHO) classification criteria and to evaluate the utility of emerging markers. Patients were diagnosed based on the 2021 WHO criteria, with atypical carcinoids (ACs) defined by the presence of focal necrosis and/or 2-10 mitoses per 2 mm 2 . We investigated two markers of tumour proliferation: the Ki-67 index and phospho-histone H3 (PHH3) protein expression, quantified by pathologists and automatically via deep learning. Additionally, an unsupervised deep learning algorithm was trained to uncover previously unnoticed morphological features with diagnostic value. The accuracy in distinguishing typical from ACs is hampered by interobserver variability in mitotic counting and the limitations of morphological criteria in identifying aggressive cases. Our study reveals that different Ki-67 cut-offs can categorise LNETs similarly to current WHO criteria. Counting mitoses in PHH3+ areas does not improve diagnosis, while providing a similar prognostic value to the current criteria. With the advantage of being time efficient, automated assessment of these markers leads to similar conclusions. Lastly, state-of-the-art deep learning modelling does not uncover undisclosed morphological features with diagnostic value. This study suggests that the mitotic criteria can be complemented by manual or automated assessment of Ki-67 or PHH3 protein expression, but these markers do not significantly improve the prognostic value of the current classification, as the AC group remains highly unspecific for aggressive cases. Therefore, we may have exhausted the potential of morphological features in classifying and prognosticating LNETs. Our study suggests that it might be time to shift the research focus towards investigating molecular markers that could contribute to a more clinically relevant morpho-molecular classification.
oncology