Endocrine Abstracts

JOINT2027

Objective: PIT1-lineage PitNETs are the most common pituitary tumors, encompassing lactotroph, somatotroph and thyrotroph tumor types. At a molecular level, several questions about their biology remain open. Among these, the molecular mechanisms of resistance to medical treatments remain largely unknown. The aim of this study was to better characterize PIT1-lineage PitNETs based on bulk transcriptome analysis, refining their molecular classification and deciphering molecular signatures associated with treatment resistance.

Methods: The transcriptome of 121 patients with PIT1-lineage PitNETs was evaluated. Clinical and histological data were collected, including those about response to dopamine agonists in prolactinomas, and to first-generation somatostatin analogs in acromegaly. Unsupervised transcriptome classifications were generated, both in the overall cohort and within subtypes, and tested for association with clinical data and treatment response.

Results: Unsupervised classification mainly discriminated PIT1-lineage PitNETs according to tumor histopathology (χ₂ P<10_-22 and P<10_-4 for WHO 2017 tumor type and granulations, respectively), secretion (χ₂ P<10_-22), and GNAS mutational status (χ₂ P<10_-6). Compared to previous classifications, a clear separation between sparsely granulated somatotroph PitNETs and thyrotroph/plurihormonal PIT1 PitNETs emerged. Moreover, a majority of aggressive PitNETs cumulated in a distinct transcriptomic subgroup, defined by a signature of mesenchimal differentiation, irrespective of histotype. Among prolactinomas, sensitivity to dopamine agonists was the main driver of transcriptomic classification. Sensitive tumors cumulated in a distinct molecular subgroup. Resistant tumors were classified into three different clusters, with specific expression profiles (dedifferentiated, proliferative, and immune-rich). Among tumors causing acromegaly, three main transcriptomic groups were observed: sparsely granulated somatotrophs, densely granulated somatotrophs with GNAS somatic mutation, and densely granulated somatotrophs with PIT1/SF1 co-expression. Moreover, three smaller clusters were also observed, lacking the hallmark features of the three main ones: one of proliferative tumors with aggressive behavior; one of giant invasive tumors with atypical histology but otherwise indolent growth; and one of smaller tumors with limited extrasellar extension. Sensitivity to first-generation somatostatin analogs was associated with these molecular classes, with resistant tumors grouping within the sparsely granulated, proliferative, and giant invasive tumor clusters.

Conclusions: This transcriptomic study unravels important new aspects of the biology of PIT1-lineage PitNETs and further refines their molecular classification. The transcriptomic classification of prolactinomas recognizes the sensitivity to dopamine agonists as its main driver. The transcriptomic classification of tumors causing acromegaly primarily reflects their histological heterogeneity and identifies distinct molecular clusters associated with treatment response.