Risk of recurrence after successful surgery for Cushing's disease and association with USP8 genotype and tumour size: an international, retrospective, longitudinal cohort study

BACKGROUND: Recurrence after successful pituitary surgery remains a challenge in the management of patients with Cushing's disease, with no reliable predictors of long-term outcome. Pathogenic somatic USP8 variants are found in one third of cases and their association with recurrence is unclear. The aim of this study was to determine the association between USP8 status and postoperative outcome. METHODS: This international, retrospective, longitudinal study was done in eight tertiary centres in east Asia, Europe, and North America. We reviewed clinical records and genetic information of patients with clinically diagnosed and pathologically confirmed Cushing's disease who underwent first pituitary surgery in any of the participating centres between Jan 1, 1989, and April 1, 2024. Inclusion criteria were postoperative follow-up of at least 3 months after first pituitary surgery, known postsurgical outcome, and tissue availability or known USP8 status. The primary outcomes were recurrence and time to recurrence after first surgery. We assessed recurrence-free intervals and recurrence risk using survival analysis, multivariate logistic regression, and Cox proportional hazards models. FINDINGS: We retrospectively retrieved and examined clinical records from 558 patients, 123 of whom were excluded. 360 (83%) of 435 patients were female and 75 (17%) were male. We detected USP8 variants in 195 (45%) cases. Recurrence was recorded in 66 (18%) of 371 patients in immediate remission. Risk of recurrence depended on USP8 status and tumour size. Patients with USP8-variant microadenomas and USP8-variant macroadenomas had similar cumulative hazards for recurrence, so they were considered as a single risk group. 10-year recurrence rate was higher for USP8-variant tumours (36·8%, 95% CI 23·7-47·7) than for wildtype microadenomas (15·0%, 4·9-24·0; adjusted p=0·016) but was lower than for wildtype macroadenomas (44·5%, 26·2-58·2; adjusted p=0·025). Patients with USP8-variant tumours (hazard ratio 2·41, 95% CI 1·19-4·87; p=0·014) or wildtype macroadenomas (4·48, 2·11-9·52; p