Non-small cell lung carcinoma subtyping in conventional cytology: Results of the IASLC Cytology Working Group survey to determine specific cytomorphological criteria for adenocarcinoma and squamous cell carcinoma
Deepali Jain,Aruna Nambirajan,Gang Chen,Kim Geisinger,Kenzo Hiroshima,Lester Layfield,Yuko Minami,Andre L. Moreira,Noriko Motoi,Mauro Papotti,Natasha Rekhtman,Prudence A. Russell,Spasenija Savic Prince,Fernando Schmitt,Yasushi Yatabe,Serenella Eppenberger-Castori,Lukas Bubendorf,Mary Beth Beasley,Sabina Berezowska,Alain Borczuk,Elizabeth Brambilla,Teh-Ying Chou,Jin-Haeng Chung,Wendy Cooper,Sanja Dacic,Yuchen Chan,Fred R. Hirsch,David Hwang,Philippe Joubert,Keith Kerr,Sylvie Lantuejoul,Dongmei Lin,Fernando Lopez-Rios,Daisuke Matsubara,Mari Mino-Kenudson,Andrew Nicholson,Claudia Poleri,Anja Roden,Kurt Schalper,Lynette Sholl,Erik Thunnissen,William D. Travis,Ming Tsao,Ignacio Wistuba,Gang Chen
DOI: https://doi.org/10.1016/j.jtho.2022.02.013
IF: 20.121
2022-03-01
Journal of Thoracic Oncology
Abstract:IntroductionAccurate subtyping of non-small cell lung carcinomas (NSCLC) into lung adenocarcinoma (LUAD) and lung squamous cell carcinoma (LUSC) is the cornerstone of NSCLC diagnosis. Cytology samples show higher rates of classification failures, i.e., subtyping as non-small cell carcinoma, not otherwise specified (NSCC-NOS), as compared to histology. This study aims to identify specific algorithms based on known cytomorphological features that aid accurate and successful subtyping of NSCLC on cytology.MethodsThirteen expert cytopathologists participated anonymously in an online survey to subtype 119 NSCLC cytology cases (gold standard diagnoses being LUAD in 80 and LUSC in 39) enriched for non-keratinising LUSC. They selected from 23 pre-defined cytomorphological features that they used in subtyping. Data were analysed using machine learning algorithms based on Random Forest Method and regression trees.ResultsFrom 1474 responses recorded, concordant cytology typing was achieved in 53.7% (792/1474) responses. NSCC-NOS rates on cytology were similar among gold standard LUAD (36%) and LUSC (38%) cases. Misclassification rates were higher in gold standard LUSC (17.6%) than gold standard LUAD (5.5%; P<0.0001). Keratinisation, when present, recognised LUSC with high accuracy. In its absence, the ML algorithms developed based on experts' choices were unable to reduce cytology NSCC-NOS rates without increasing misclassification rates.ConclusionSuboptimal recognition of LUSC in the absence of keratinisation remains the major hurdle in improving cytology subtyping accuracy with such cases either failing classification (NSCC-NOS) or misclassifying as LUAD. NSCC-NOS appears to be an inevitable morphological diagnosis emphasizing that ancillary IC is necessary to achieve accurate subtyping on cytology.
oncology,respiratory system