Digital quantification of somatostatin receptor subtypes 2 and 5 in growth hormone-secreting pituitary tumors.

Abstract

Immunohistochemistry (IHC) of somatostatin receptor subtype 2 can predict response to first-generation somatostatin receptor ligands (fg-SRLs) in acromegaly. Recently, we validated an open-source digital image analysis (DIA) to quantify somatostatin receptor subtype 2 (SSTR2) expression. We aimed to validate the DIA also on somatostatin receptor subtype 5 (SSTR5) in a new cohort of growth hormone (GH)-secreting pituitary tumors, with IHC performed in a different laboratory, and to correlate fg-SRL response with SSTs expression. Somatostatin receptor subtype 2 and SSTR5 were assessed in 42 GH-secreting pituitary tumors, using a semiquantitative immunoreactivity score (IRS) and the DIA by use of the open-access software CellProfiler. The DIA calculates the staining intensity and the percentage of positive cells (%PC). We found a good correlation between IRS and DIA for both SSTR2 and SSTR5 (P < .001), demonstrating the reliability of the DIA in this setting. Response to fg-SRL treatment correlated with SSTR2, but not SSTR5, expression. Somatostatin receptor subtype 2 expression predicted response to fg-SRL. In particular, the identified cut-offs were IRS ≥ 5 (area under the curve [AUC] 0.763; sensitivity 77%; specificity 83%); intensity/area ≥0.106 (AUC 0.833; sensitivity 92%; specificity 83%); and %PC-DIA ≥63.7% (AUC 0.917; sensitivity 92%; specificity 83%). The SSTR2 %PC correlated with treatment response only when evaluated using the DIA, showing a better performance of this method.