Cushing's disease, also known as adrenocorticotropic hormone (ACTH)-secreting pituitary adenomas (PAs) that cause excess cortisol production, accounts for up to 85% of corticotrophin-dependent Cushing's syndrome cases. However, the genetic alterations in this disease are unclear. Here, we performed whole-exome sequencing of DNA derived from 12 ACTH-secreting PAs and matched blood samples, which revealed three types of somatic mutations in a candidate gene, USP8 (encoding ubiquitin-specific protease 8), exclusively in exon 14 in 8 of 12 ACTH-secreting PAs. We further evaluated somatic USP8 mutations in additional 258 PAs by Sanger sequencing. Targeted sequencing further identified a total of 17 types of USP8 variants in 67 of 108 ACTH-secreting PAs (62.04%). However, none of these mutations was detected in other types of PAs (n = 150). These mutations aggregate within the 14-3-3 binding motif of USP8 and disrupt the interaction between USP8 and 14-3-3 protein, resulting in an elevated capacity to protect EGFR from lysosomal degradation. Accordingly, PAs with mutated USP8 display a higher incidence of EGFR expression, elevated EGFR protein abundance and mRNA expression levels of POMC, which encodes the precursor of ACTH. PAs with mutated USP8 are significantly smaller in size and have higher ACTH production than wild-type PAs. In surgically resected primary USP8-mutated tumor cells, USP8 knockdown or blocking EGFR effectively attenuates ACTH secretion. Taken together, somatic gain-of-function USP8 mutations are common and contribute to ACTH overproduction in Cushing's disease. Inhibition of USP8 or EGFR is promising for treating USP8-mutated corticotrophin adenoma. Our study highlights the potentially functional mutated gene in Cushing's disease and provides insights into the therapeutics of this disease.

Context: The diagnosis of Cushing syndrome (CS) can be challenging. It remains to be determined which diagnostic tests are the most accurate.Objective: To summarize the accuracy of diagnostic tests for CS using contemporary meta-analytic techniques (hierarchical models).Data Sources: PubMed, Embase, Scopus, Web of Science, and the Cochrane Database of Systemic Reviews (inception until August 3, 2018).Study Selection: Studies performed in adults that determined the accuracy of one or more diagnostic tests: overnight 1-mg dexamethasone suppression test (DST), 2-day low-dose DST (2d DST), 24-hour urinary free cortisol (UFC), late-night salivary cortisol (LNSC), midnight serum cortisol (MSC), and the dexamethasone-suppressed CRH (dex-CRH) and desmopressin (dex-DDAVP) tests.Data Extraction: Two authors independently extracted data and performed methodological assessments.Data Synthesis: One hundred thirty-nine studies (14 140 participants) were included in the analysis. The respective sensitivity, specificity, positive likelihood ratio, and negative likelihood ratio (95% confidence interval [CI]) estimates include the following: DST 98.6% (96.9%-99.4%), 90.6% (86.4%-93.6%), 10.5 (7.2-15.3), and 0.016 (0.007-0.035); 2d DST 95.3% (91.3%-97.5%), 92.8% (85.7%-96.5%), 13.2 (6.47-27.1), and 0.051 (0.027-0.095); UFC 94.0% (91.6%-95.7%), 93.0% (89.0%-95.5%), 13.3 (8.47-21.0), and 0.065 (0.046-0.092); LNSC 95.8% (93.%-97.2%), 93.4% (90.7%-95.4%), 14.6 (10.3-20.7), and 0.045 (0.030-0.066); MSC 96.1% (93.5%-97.6%), 93.2% (88.1%-96.3%), 14.2 (7.96-25.2), and 0.042 (0.026-0.069); and dex-CRH 98.6% (90.4%-99.8%), 85.9% (67.6%-94.7%), 7.0 (2.80-17.6), and 0.016 (0.002-0.118). A single study evaluated dex-DDAVP. Meta-regression and a novel network meta-analytic approach suggest that DST is the most sensitive while UFC is the least sensitive.Conclusions: All of the included diagnostic tests for CS are highly sensitive and specific. It appears that the DST is the most sensitive while the UFC is less sensitive. The specificity of all first-line tests appears comparable.

Introduction: Machine learning methods in sellar region diseases present a particular challenge because of the complexity and the necessity for reproducibility. This systematic review aims to compile the current literature on sellar region diseases that utilized machine learning methods and to propose a quality assessment tool and reporting checklist for future studies. Methods: PubMed and Web of Science were searched to identify relevant studies. The quality assessment included five categories: unmet needs, reproducibility, robustness, generalizability and clinical significance. Results: Seventeen studies were included with the diagnosis of general pituitary neoplasms, acromegaly, Cushing’s disease, craniopharyngioma and growth hormone deficiency. 87.5% of the studies arbitrarily chose one or two machine learning models. One study chose ensemble models, and one study compared several models. 43.8% of studies did not provide the platform for model training, and roughly half did not offer parameters or hyperparameters. 62.5% of the studies provided a valid method to avoid over-fitting, but only five reported variations in the validation statistics. Only one study validated the algorithm in a different external database. Four studies reported how to interpret the predictors, and most studies (68.8%) suggested possible clinical applications of the developed algorithm. The workflow of a machine-learning study and the recommended reporting items were also provided based on the results. Conclusions: Machine learning methods were used to predict diagnosis and posttreatment outcomes in sellar region diseases. Though most studies had substantial unmet need and proposed possible clinical application, replicability, robustness and generalizability were major limits in current studies.