Low immediate postoperative serum-cortisol nadir predicts the short-term, but not long-term, remission after pituitary surgery for Cushing’s disease

The initial endocrine evaluation was performed at a specialised endocrinology centre at OUS. Patients with clinical features and a baseline laboratory test indicative of Cushing’s syndrome were evaluated with 24-h urinary free cortisol, late night salivary cortisol, 48-h, low dose dexamethasone suppression test (DST) (2 mg/d for 48 h) and corticotrophin-releasing hormone (CRH) test. Once the diagnosis of ACTH-dependent Cushing’s syndrome was established [17], a multidisciplinary team, consisting of an endocrinologist, radiologist, oncologist and neurosurgeon, discussed the patients and established the indication for surgery. All patients underwent 1.5 tesla MRI scans after a dedicated pituitary protocol, and those with initial normal scans were investigated with dynamic sequences. When indicated, inferior petrosal sinus sampling (IPSS) was performed to establish the diagnosis of Cushing’s disease. When IPSS was performed ACTH was measured both before and after intravenous CRH administration. A central-to-peripheral ratio of more than two before CRH administration, or more than three after, was interpreted as positive for Cushing’s disease [18].

All patients underwent operation while under general anaesthesia, with total intravenous anaesthesia in most cases. The standard microsurgical endonasal transseptal approach was used prior to 2006. Through the right nostril, an incision was made in the anterior part of the septum, and the mucosa was dissected from the septal cartilage and bone. The anterior wall of the sphenoid sinus was opened bilaterally and any septum in the sphenoid sinus was removed as necessary. The floor of the sella was opened with a high-speed drill. The tumour was removed with standard instruments, such as curettes, suction and micro forceps. After tumour removal, the floor of the sella was reconstructed with two layers of Neuropatch® (BBraun, Tutlingen, Germany) and the septal bone to reconstruct the floor of the sella, which was sealed with fibrin glue and gelfoam. In a few patients with small nostrils (paediatric cases), a sublabial approach was performed. After 2006, the endoscopic endonasal transsphenoidal approach became standard procedure. One or both nostrils were entered according to the available space and need for exposure during the surgery. We used standard Storz endoscopes (Karl Storz GmbH & Co, Tuttlingen, Germany) (180/4 mm), with 0⁰, 30⁰ and 45⁰ angulations coupled to cameras, in the later period of the study to HD cameras. In the first cases, we used a fixed endoscope support, whereas in the later cases, the endoscope was handled freehand to increase flexibility during surgery. Image-guided navigation (BrainLab AG, Feldkirchen, Germany) was used in some procedures. Tumour resection was performed with standard surgical instruments, such as dissectors, curettes, suction and micro forceps, dependent on the tumour size and firmness. In tumours with parasellar extension, we routinely used an ultrasonic Doppler probe (Mitzhuo Medical Inc, Tokyo, Japan) to localise the internal carotid artery. The surgical corridor was closed with two layers of Duragen (Integra LifeScience Corporation, New Jersey, USA) and a Porex plate (Porex Corporation, Atlanta, Georgia, USA) for rigid reconstruction of the sellar floor. One dose of cefalotin was administered intravenously before the operation. No glucocorticoids were given pre- or perioperatively.

The following complications were recorded: mortality, CSF-leakage, vascular injury, infection, new neurological deficits and new hormone deficiencies, including ADH deficiency.

No exogenous glucocorticoids were given postoperatively, and the serum levels of cortisol were measured every 6 h at fixed hours (12, 18, 24, 06, 12, and so forth), starting immediately after the end of surgery. If the patient showed clinical symptoms or signs of hypocortisolaemia (headache, nausea, fatigue, low serum sodium, fever, and hypotension) or had S-cortisol levels below 100 nmol/l, exogenous glucocorticoids were administered, initially per oral cortisone acetate 100 mg × 2, and no further measurements of S-cortisol was done. Patients without clinical symptoms of hypocortisolaemia who did not reach a nadir of 100 nmol/dl were prescribed exogenous glucocorticoids at discharge from the hospital for use as required. The S-cortisol levels were measured for 1 to 4 days after surgery. We used a S-cortisol nadir of <50 nmol/l, <100 nmol/l and <200 nmol/l to predict remission at the different endpoints, and we calculated the sensitivity and specificity at each time point.

Remission was defined as relief from clinical symptoms, urinary free cortisol below threshold, late night salivary cortisol below threshold, and suppression to <50 nmol/l of S-cortisol on 1 mg overnight or 2 mg/D, 48 h DST. The definition of recurrence is difficult, and it must be stressed that this is an integrated clinical and biochemical assessment. Relapse of clinical symptoms, such as cutaneous symptoms (red striae, easy bruising and thin skin), central obesity and proximal muscle weakness, are strong indicators of disease recurrence, but these symptoms often appear late and long after biochemical tests indicate relapse from the disease. We used elevated morning S-cortisol, elevated late night salivary cortisol, elevated 24 h urinary free cortisol, elevated morning serum cortisol and inadequate suppression of cortisol after DST as the criteria for recurrence.

Patients were seen in the outpatient ward of the section of specialised endocrinology at 6 weeks, 3 months and 6 months after surgery and thereafter at least once a year.

The data are summarised with counts, percentages, means and medians as appropriate. T-tests robust for unequal variances are used. Due to the correlational structure of the repeated intra-individual cortisol measurements, the confidence intervals of Figs. 2 and 3 were calculated according to Morey [19]. The time trend of cortisol was investigated using logistic regression and the area under the curve (AUC) of the cortisol values over time, nadir values and slope until nadir. The nadir value was further investigated as a classifier for remission using receiver-operating-characteristics (ROC) curves. The optimal cut-off was calculated by finding the maximum Youden index over all nadir values. The confidence interval for the AUC in the ROC was calculated using bootstrapping techniques. A confidence interval for the ROC AUC excluding 0.5 was considered significant.

Excel 2010 (Microsoft) and R version 3.0.3 [20] were used for all statistical analysis. P-values <0.05 were considered significant.