N. Benanteur*a (Ms), C. Villaa (Dr), D. De Murata (Mr), R. Boëzennecb (Mr), R. Armignaccoa (Dr), A. Jouinota (Dr), F. Letourneura (Mr), K. Perlemoinea (Mrs), S. Allassonnierec (Prof), ML. Raffin-Sansond (Prof), JF. Emilee (Prof), S. Gaillardf (Dr), J. Bertherata (Prof), B. Baussartf (Dr), G. Assiéa (Prof)

a INSERM U1016, Institut Cochin, Paris, FRANCE ; b Ecole des ponts ParisTech, Paris, FRANCE ; c Université Paris Descartes, Paris, FRANCE ; d Department of Endocrinology, Hôpital Ambroise Paré, Paris, FRANCE ; e Department of Pathology, Ambroise Paré, Assistance Publique–Hôpitaux de Paris, Boulogne Billancourt, FRANCE ; f Department of Neurosurgery, Foch Hospital, Suresnes, FRANCE

* nesrine.benanteur@gmail.com

Predicting aggressiveness is a major unmet needs for Pituitary NeuroEndocrine Tumors (PitNETs). The first multi-omics classification of PitNETs demonstrated that lineage (related to pituitary development and hormone secretion type) was by far the first determinant of this classification (1). In contrast, there was no obvious signature of tumor aggressiveness that would be common to all PitNETs types, and instead being potentially specific to each PitNET subtype. Our goal is to bring out these aggressiveness transcriptomic markers. By extending the published cohort (1), transcriptomes of 152 frozen PitNETs of varying aggressiveness were extracted. Different classifiers (linear and radial SVM, decision tree, LASSO and linear regression) were used on transcriptomic data in order to classify PitNETs according to their level of aggressiveness. This first analysis gave rise to aggressive tumors recognition rates close to 30% at best, confirming a signature of common aggressiveness too weak to be decisive in their classification. The failure to identify a global signature lead us to define aggressiveness markers specific to the lineage. The originality and probability of success of this approach rely on not mixing unrelated molecular types for aggressiveness signatures, preventing contamination with the dominant lineage information. Principal component analysis of corticotroph tumors and prolactinomas shows a link between transcriptome and aggressiveness. A measure of the correlation between prolactinomas and corticotroph tumors signatures shows two distinct aggressiveness signatures between these two groups: proliferation signature in prolactinomas, and regulation of protein metabolism signature in corticotroph tumors. This analysis seems to confirm the existence of transcriptomic signatures specific to PitNETs histological types. Specific aggressiveness signatures linked to the remaining histological types have yet to be defined.

1. Neou...Assié. Pangenomic Classification of Pituitary Neuroendocrine Tumors. Cancer Cell 2020.

The author has declared no conflict of interest.