Development and validation of clinical screening systems for Cushing disease in the United States

Purpose: Clinical screening systems (CSSs) for Cushing syndrome (CS) validated in Europe have not been evaluated for CS or Cushing disease (CD) in the United States (US). We aimed to evaluate existing CSSs in US patients and develop two new symptom-based CSSs to identify patients with high pre-test probability of disease warranting referral for definitive biochemical workup—one for broad CS screening and one specifically for CD. Methods: Data were obtained from the Registry of Adenomas of the Pituitary and Related Disorders (RAPID)—comprising 615 patients who underwent transnasal transsphenoidal resection for CD at one of 11 US pituitary centers, the Centers for Disease Control and Prevention’s 2019 National Health Interview Survey (NHIS)—comprising 31,997 US respondents, and a single institution CD-NFA cohort—comprising 468 US patients diagnosed with either CD (n = 385) or nonfunctioning adenoma (n = 83). The RAPID Community Cushing CSS was derived from differences between the RAPID and NHIS cohorts, and the RAPID CD CSS from differences between patients with CD versus NFA. Results: In external validation using US-based cohorts, the RAPID Community CSS achieved an AUC of 0.707, compared to the Spanish (AUC = 0.691) and Italian (AUC = 0.685) models, and the RAPID CD CSS demonstrated greater external sensitivity (0.836, threshold = 0.5) at the Youden-optimized threshold than the Spanish (0.605, threshold = 4) and Italian (0.735, threshold = 6) CSSs. Conclusions: In US patient populations, the RAPID Community Cushing CSS demonstrated superior discriminative ability for CD compared to the Italian and Spanish CSSs, and the RAPID CD CSS achieved the highest sensitivity for CD among all CSSs evaluated.