BioScientifica, Endocrine-Related Cancer, 1(28), p. 79-95, 2021
DOI: 10.1530/erc-20-0128
Full text: Unavailable
Benign adrenal tumors cover a spectrum of lesions with distinct morphology and steroid secretion. Current classification is empirical. Beyond a few driver mutations, pathophysiology is not well understood. Here, a pangenomic characterization of benign adrenocortical tumors is proposed, aiming at unbiased classification and new pathophysiological insights. Benign adrenocortical tumors (n = 146) were analyzed by transcriptome, methylome, miRNome, chromosomal alterations and mutational status, using expression arrays, methylation arrays, miRNA sequencing, SNP arrays, and exome or targeted next-generation sequencing respectively. Pathological and hormonal data were collected for all tumors. Pangenomic analysis identifies four distinct molecular categories: (1) tumors responsible for overt Cushing, gathering distinct tumor types, sharing a common cAMP/PKA pathway activation by distinct mechanisms; (2) adenomas with mild autonomous cortisol excess and non-functioning adenomas, associated with beta-catenin mutations; (3) primary macronodular hyperplasia with ARMC5 mutations, showing an ovarian expression signature; (4) aldosterone-producing adrenocortical adenomas, apart from other benign tumors. Epigenetic alterations and steroidogenesis seem associated, including CpG island hypomethylation in tumors with no or mild cortisol secretion, miRNA patterns defining specific molecular groups, and direct regulation of steroidogenic enzyme expression by methylation. Chromosomal alterations and somatic mutations are subclonal, found in less than 2/3 of cells. New pathophysiological insights, including distinct molecular signatures supporting the difference between mild autonomous cortisol excess and overt Cushing, ARMC5 implication into the adreno-gonadal differentiation faith, and the subclonal nature of driver alterations in benign tumors, will orient future research. This first genomic classification provides a large amount of data as a starting point.