STING expression in pituitary neuroendocrine tumour immune microenvironments.

The pituitary neuroendocrine tumour (PitNET) immune microenvironment (IME) has become a recent focus of research interest. Multiplex immunofluorescence combined with digital image analysis is a powerful technique to interrogate the IME. Processes such as DNA damage response and cellular senescence have been implicated in somatotroph tumour pathogenesis, and these have in turn been linked with immunological responses to neoplasia. The objectives of this study were to compare the immune cell infiltrate in PitNETs to normal pituitary, investigate how the PitNET-IME may vary according to treatment with somatostatin receptor ligand, tumour invasiveness, tumour proliferation, and AIP mutation status. Two bespoke multiplex panels were developed and applied to a range of PitNET subtypes before being interrogated with deep learning algorithms and QuPath image analysis software. Tumours had significantly higher total macrophage infiltrate (P < 0.001), STING+ macrophages (P < 0.001) and STING positivity (P < 0.001) compared with normal pituitary tissue. Macrophage infiltration and STING positivity were positively correlated with tumour Ki-67 proliferation index (rs = 0.305, P = 0.003; and rs = 0.263, P = 0.010, respectively). Results suggested that AIP-mutated PitNETs may have higher CD8+ T-lymphocyte infiltrate, macrophage infiltrate, including STING+ macrophages and overall STING positivity compared with AIP-negative tumours. The application of machine learning to multiplex immunofluorescence images identifies macrophages and STING as upregulated in these subgroups of functioning PitNETs. This immunological response may indicate that DNA damage, cellular senescence and autophagy could be features of these tumours.