Objective: Central venous sampling (CVS) is used frequently in the evaluation of ACTH-dependent Cushing's syndrome. However, several controversies exist including the diagnostic accuracy, the sampling site of choice (cavernous sinus vs. inferior petrosal sinus) and the use of lateralization data in tumour localization. We have analysed our experience with CVS to address these controversies.
Design: We retrospectively reviewed CVS data in patients with ACTH-dependent Cushing's syndrome, in whom cavernous sinus sampling (CSS), inferior petrosal sinus sampling (IPSS) and IPSS after administration of ovine corticotrophin releasing hormone (oCRH) were performed.
Patients: Data on 95 patients were analysed, including 79 patients with suspected Cushing's disease (CD) and 16 patients with suspected ectopic ACTH syndrome (EAS).
Results: For the differential diagnosis of ACTH-dependent Cushing's syndrome, the diagnostic accuracy of IPSS after oCRH stimulation was 97% compared to 86% for CSS. While no single sampling site was perfect in diagnostic accuracy, sampling both CS and IPS achieved a combined diagnostic accuracy of 100%. Lateralization data predicted tumour location in 62-68% of the patients with various central venous drainage patterns and in 77-80% of the patients with symmetrical drainage. CSS was not significantly superior to IPSS in tumour lateralization. In patients with suspected CD based on CVS and in whom an adenoma was not found on magnetic resonance imaging (MRI) and not located by the surgeon intraoperatively, hemihypophysectomy based on lateralization data was successful in only 10 of the 18 patients (56%) with various central venous drainage patterns and in 5 of 10 patients with symmetrical drainage.
Conclusion: CVS is a powerful method for differentiating CD from the EAS. CSS without oCRH was not superior to IPSS after oCRH stimulation; however, we achieved a 100% diagnostic accuracy if at least two sites were sampled. Tumour localization by CVS did not accurately predict the tumour site at surgery and should not be used to guide surgical resection.