Comprehensive molecular and pathological evaluation of transitional mesothelioma assisted by deep learning approach: a multi institutional study of the International Mesothelioma Panel from MESOPATH Reference Center.

Journal of Thoracic Oncology 2020 March 9 [Link]

Salle FG, Le Stang N, Tirode F, Courtiol P, Nicholson AG, Tsao MS, Tazelaar HD, Churg A, Dacic S, Roggli V, Pissaloux D, Maussion C, Moarii M, Beasley MB, Begueret H, Chapel DB, Copin MC, Gibbs AR, Klebe S, Lantuejoul S, Nabeshima K, Vignaud JM, Attanoos R, Brcic L, Capron F, Chirieac LR, Damiola F, Sequeiros R, Cazes A, Damotte D, Foulet A, Giusiano-Courcambeck S, Hiroshima K, Hofman V, Husain AN, Kerr K, Marchevsky A, Paindavoine S, Picquenot JM, Rouquette I, Sagan C, Sauter J, Thivolet F, Brevet M, Rouvier P, Travis WD, Planchard G, Weynand B, Clozel T, Wainrib G, Fernandez-Cuesta L, Pairon JC, Rusch V, Girard N


Histological subtypes of malignant pleural mesothelioma are a major prognostic indicator and decision denominator for all therapeutic strategies. In ambiguous case a rare transitional [TM) pattern may be diagnosed by pathologists either as epithelioid (EM), biphasic (BM) or sarcomatoid (SM) mesothelioma. The aims of this study were to better characterize the TM subtype from a morphological, immunohistochemical, molecular standpoint; deep learning of pathological slides was applied to this cohort. METHODS: A random selection of 49 representative digitalized sections from surgical biopsies of TM were reviewed by 16 panelists. We evaluated BAP1 expression and p16 homozygous deletion [HD]. We conducted a comprehensive integrated transcriptomic analysis. Unsupervised deep learning algorithm was trained to classify tumors. RESULTS: The 16 panelists recorded 784 diagnoses on the 49 cases. Whilst Kappa value of 0.42 is moderate, the presence of a TM component was diagnosed in 51%. In 49%, the reviewers classified the lesion as EM in 53%, SM in 33%, or BM in 14%. Median survival was 6.7 months. Loss of BAP1 observed in 44% was less frequent in TM than in EM and BM. p16 HD was higher in TM 73% followed by BM (63%) and SM (46%). RNA sequencing unsupervised clustering analysis showed that TM grouped together and were closer to SM than to EM. Deep learning analysis achieved a 94% accuracy for TM identification CONCLUSION: These results demonstrated that TM pattern should be classified in non-epithelioid mesothelioma at minimum as a subgroup of SM type.