Endocrine Abstracts

Pituitary adenomas, now called pituitary neuroendocrine tumors (PitNETs), vary in histological type, secretion, invasion and growth speed. A World Health Organization (WHO) histo-prognostic classification was released in 2017. Pituitary tumorigenesis is largely unexplained. Rare germline mutations (MENIN, AIP), and common somatic mutations in somatotroph (GNAS) and corticotroph (USP8) are reported. Recently, genomic analyses have been reported. Yet, driver genes and pathways remain to be fully identified, as well as a comprehensive view of omics in the different subtypes of PitNETs.

Aim: To provide a genomic unbiased classification of PitNETs.

Methods: A clinical, histological and genomic characterization of 134 PitNETs of all subtypes was performed, combining exome, RNA and miRNA sequencing, SNP array and methylation array. Unsupervised classifications were generated (non-negative matrix factorization and hierarchical clustering).

Results: Median somatic mutation rate was 95, mainly C>T substitutions belonging to ‘signature 1’ trinucleotide signature. No difference was observed between histo-types. Only GNAS and USP8 presented >5% mutations. Three chromosome alteration profiles were identified: extended losses, ‘quiet’ profiles, and extended gains. Gonadotroph and silent corticotroph were mainly ‘quiet’. Chromosomal alterations were not related to aggressiveness. Epigenomic classifications (miRnome and methylome) were strongly associated with PitNETs histological type and secretion (χ² P<10–18). Especially, PitNETs from POU1F1 (Pit1) lineage (lacto-, thyro- and somatotroph) showed diffuse DNA hypomethylation. Hypomethylation was correlated with genomic instability (correlation coefficient: 0.4). Unsupervised transcriptome classification revealed 6 groups, associated with histotype and secretion (χ² P<10–59), with four noticeable discrepancies compared to WHO2017 classification: « null-cells » were not distinct from gonadotroph; silent corticotroph PitNETs fell apart from overt Cushing corticotroph PitNETs; 5/8 sparsely granulated somatotroph PitNETs were grouped with thyrotroph and PIT1 plurihormonal PitNETs; lactotroph PitNETs were distinct from mixed GH-PRL and somatotroph PitNETs. USP8 and GNAS mutations formed specific homogeneous subgroups. Silent corticotroph and gonadotroph shared a common gonadotroph signature.

Conclusion: This genomic study unravels important new aspects of PitNETs biology. Mainly: - DNA hypomethylation and chromosomal instability of PitNETs of the Pit1 lineage, suggesting that Pit1 differentiation may induce chromatin opening, with subsequent genome instability. - A specific molecular signature of sparsely granulated somatotroph PitNETs, which may help to better understand and predict resistance to somatostatin analogues. - Silent corticotroph PitNETs combine corticotroph and gonadotroph differentiation signatures. This genomic classification of PitNETs supports the importance of pituitary lineage in pituitary tumorigenesis, and proposes a first robust and unbiased classification based on tumor biology.