Late-night salivary cortisol (LNSC) is a measure of nadir

Transcription

1 ORIGINAL ARTICLE Accuracy of Late-Night Salivary Cortisol in Evaluating Postoperative Remission and in Cushing s Disease Fatemeh G. Amlashi, Brooke Swearingen, Alexander T. Faje, Lisa B. Nachtigall, Karen K. Miller, Anne Klibanski, Beverly M.K. Biller, and Nicholas A. Tritos Neuroendocrine Unit (F.G.A., A.T.F., L.B.N., K.K.M., A.K., B.M.K.B., N.A.T.) and Department of Neurosurgery (B.S.), Massachusetts General Hospital, Boston, MA 02114, and Harvard Medical School (F.G.A., B.S., A.T.F., L.B.N., K.K.M., A.K., B.M.K.B., N.A.T.), Boston, MA Context: Late-night salivary cortisol (LNSC) is well-validated in the diagnosis of Cushing s disease (CD). The accuracy of LNSC during follow-up of patients undergoing transsphenoidal surgery (TSS) has not been fully characterized. Objectives: We examined the accuracy of LNSC in establishing remission and identifying recurrence in postoperative patients with CD. Design: This is a retrospective study. Patients: Records of patients with CD who underwent TSS by a single neurosurgeon in our tertiary center ( ) were analyzed (N 224). Patients were selected for further investigation (n 165) if there was at least one available LNSC test obtained after TSS (either within 3 months or during long-term follow-up). Extracted data included demographic and clinical characteristics, magnetic resonance imaging and laboratory data (morning serum cortisol, 24-hour urine free cortisol [UFC], LNSC). Main Outcomes and Measures: Remission was defined as nadir morning serum cortisol less than 5 mcg/dl and nadir 24-hour UFC less than 23 mcg. was considered definite if confirmed surgically or prompted radiotherapy. Results: Surgical remission occurred in 89% of 89 patients with available LNSC data. LNSC, obtained within 3 months of TSS, established remission with 94% sensitivity and 80% specificity at a cutpoint of 1.9 nmol/l (area under the curve [AUC] 0.90). At a median follow-up of 53.5 months, LNSC established recurrence (75% sensitivity and 95% specificity) at a cutpoint of 7.4 nmol/l (AUC 0.87), and 24-hour UFC established recurrence (68% sensitivity and 100% specificity) at a cutpoint of 1.6-fold above normal (AUC 0.82). Conclusions: LNSC may accurately establish remission after TSS and identify recurrence more accurately than 24-hour UFC during long-term follow-up. (J Clin Endocrinol Metab 100: , 2015) ISSN Print X ISSN Online Printed in USA Copyright 2015 by the Endocrine Society Received April 27, Accepted July 17, First Published Online July 21, 2015 Late-night salivary cortisol (LNSC) is a measure of nadir cortisol levels and is a validated test in the diagnosis of patients with suspected Cushing s disease (CD) (1). First-line treatment in most cases is transsphenoidal surgery (TSS), with reported success rates varying widely, possibly reflecting differences in surgical expertise and cortisol cutpoints used to define remission, which is commonly established on the basis of low morning serum cor- Abbreviations: AUC, area under the curve; CD, Cushing s disease; DM, diabetes mellitus; DST, dexamethasone suppression test; IPSS, inferior petrosal sinus sampling; LNSC, latenight salivary cortisol; MRI, magnetic resonance imaging; NPV, negative predictive value; PPV, positive predictive value; ROC, receiver operating characteristic; TSS, transsphenoidal surgery; UFC, urine free cortisol; ULN, upper limit of normal press.endocrine.org/journal/jcem J Clin Endocrinol Metab, October 2015, 100(10): doi: /jc

2 doi: /jc press.endocrine.org/journal/jcem 3771 tisol and 24-hour urine free cortisol (UFC) levels. However, the accuracy of LNSC in establishing remission has not been elucidated (2). may develop in the years following TSS. Therefore, these patients require lifelong clinical and laboratory monitoring. In addition to clinical evaluation, biochemical tests have an important role in determining both disease remission and recurrence. There are conflicting data as to whether it is possible to identify patients at greater risk for recurrence of CD. A thorough review and meta-analysis of possible predictive factors of recurrence found that age, gender, tumor size, and macroscopic tumor invasion were not associated with CD recurrence, whereas low levels of morning serum cortisol immediately after surgery appeared to be a predictor of long-term remission (2, 3). In addition to its accuracy in establishing the diagnosis of CD, recent studies with limited population size or short follow-up have suggested that LNSC may also be helpful as a sensitive test for detecting recurrence (4, 5). However, there are no data regarding the role of LNSC in predicting lasting remission in patients with CD undergoing TSS. In the present study, we investigated the accuracy of LNSC in establishing remission and predicting and identifying recurrent hypercortisolism in patients with CD after TSS compared with morning serum cortisol and 24-hour UFC levels. Materials and Methods Electronic and paper medical records of patients with CD who underwent TSS by a single neurosurgeon (B.S.) in our institution between 2005 and 2014 were studied (N 224). Patients with at least one available LNSC test, obtained before/after TSS or during follow-up, were selected for further investigation (n 165). Extracted data included demographic variables, medical history (including history of smoking, diabetes mellitus [DM], depression, hypertension), and magnetic resonance imaging (MRI) findings. Laboratory data, including morning serum cortisol, LNSC, and 24-hour UFC, were recorded. The number of LNSC tests done in a given interval following TSS was registered as a sequence. Because patients were generally discharged from the hospital at 24 to 36 hours after TSS, they were prescribed 0.5 to 1 mg dexamethasone daily (selected as it is not detected in cortisol assays), until their remission status was established. Of note, 89% of postoperative patients received low-dose dexamethasone for less than 14 days before cortisol data were obtained. After remission was established, shorter-acting glucocorticoid replacement was substituted. All patients withheld glucocorticoid replacement for 24 hours before cortisol testing. The study was approved by the Partners Healthcare Institutional Review Board (Boston, MA). Salivary samples were collected by patients at home between 11 PM and midnight. All patients had been advised to avoid smoking, brushing their teeth, or using glucocorticoid-containing creams and cosmetics before specimen collection. All salivary samples were sent to a single laboratory (ACL, West Allis, WI) where cortisol was measured by enzyme immunoassay. The lower detection limit of the enzyme immunoassay was 0.3 nmol/l. The intraassay coefficient of variation was 5.2% at 3.1 (SD 0.2) nmol/l (n 10) and 2.6% at 10.4 (0.3) nmol/l (n 10). Interassay (total) coefficient of variation was 11% at 2.8 (0.3) nmol/l (n 10), 11% at 10.1 (1.1) nmol/l (n 10), and 6.9% at 25.0 (1.7) nmol/l (n 10) (6). Because the present study did not recruit healthy subjects, the normative range for LNSC was not derived. The normal reference limit for LNSC provided by ACL was used in the present study, including data from 73 healthy controls, 39 patients with proven Cushing s syndrome, and 39 patients with rule out Cushing s in whom the diagnosis was considered unlikely on further testing (7). This was less than 4.3 nmol/l for specimens collected between 11 PM and midnight. Of note, 24-hour UFC was assayed by various methods (liquid chromatography-tandem mass spectrometry and radioimmunoassay) in different laboratories. Therefore, to allow for a meaningful analysis of 24-hour UFC data, maximal 24-hour UFC values are reported as fold elevation above the upper limit of the corresponding normal range (ULN). Morning serum cortisol was measured by immunoassays in different laboratories. The diagnosis of CD was established based on the presence of hypercortisolism (including 24-hour UFC, overnight 1 mg dexamethasone suppression test [DST] and/or LNSC), nonsuppressed plasma adrenocorticotropic hormone (corticotropin) levels, and imaging findings on pituitary-protocol MRI examination. Of note, in patients without a clear-cut ( 10 mm) adenoma on MRI, bilateral inferior petrosal sinus sampling (IPSS) was performed to establish the diagnosis of CD. Remission (within 3 months of TSS) was defined as nadir morning serum cortisol less than 5 mcg/dl (obtained within 2 weeks of TSS) and nadir 24-hour UFC less than 23 mcg (obtained within 3 months of TSS) (8). Patients whose remission was established beyond 2 weeks postoperatively were considered to have a delayed remission. Of note, LNSC levels were not included in the definition of remission. was established clinically based on the treating physicians impression of patients clinical status (symptoms and signs suggestive of recurrence) and data from endocrine reassessment (including any reproducible abnormalities in 24- hour UFC, DST, and/or LNSC), and was considered definite if confirmed surgically (in patients who underwent repeat TSS), or prompted the administration of radiation therapy to the sella with interim medical therapy; otherwise, it was considered as probable. Patients with definite recurrence of CD were not offered medical therapy alone as definitive therapy. Statistical analyses were performed by using JMP Pro (SAS institute Inc., Cary, NC, USA). Variable distributions were analyzed to examine if they were normal by using the Shapiro- Wilk test. Because variables were found not to be normally distributed, the Wilcoxon test was used for further analysis. Receiver operating characteristic (ROC) curves were plotted to assess the diagnostic role of LNSC in establishing remission and predicting or identifying recurrence after TSS. Sensitivity, specificity, and negative and positive predictive values (NPV and PPV, respectively) were calculated at different levels of LNSC and optimal cutpoints were identified. Results are reported as median (range) or percentages as appropriate. Statistical significance was accepted as P.05 (two-tailed analyses).

3 3772 Amlashi et al Late-Night Salivary Cortisol in Cushing s Disease J Clin Endocrinol Metab, October 2015, 100(10): Table 1. Baseline Characteristics of the Study Population Patients Evaluated for Remission Following TSS (n 89) Patients Evaluated for Following TSS (n 68) Patients Evaluated for Long-Term During Follow-Up (n 34) Age at TSS (years) 42.4 ( ) 40.3 ( ) 41.3 ( ) Gender F 74 (83%) 57 (84%) 30 (88%) M 15 (17%) 11 (16%) 4 (12%) BMI (kg/m 2 ) (63%) 43 (63%) 23 (68%) (37%) 25 (37%) 11 (32%) Current smoker Yes 10 (11%) 7 (10%) 0 (0%) No 79 (89%) 61 (90%) 34 (100%) Hypertension Yes 59 (66%) 44 (65%) 17 (50%) No 30 (34%) 24 (35%) 17 (50%) Diabetes mellitus Yes 27 (30%) 23 (34%) 10 (29%) No 62 (70%) 45 (66%) 24 (71%) Depression Yes 27 (30%) 21 (31%) 13 (39%) No 62 (70%) 47 (69%) 21 (61%) Tumor size Microadenoma 79 (89%) 58 (85%) 23 (68%) Macroadenoma 10 (11%) 10 (15%) 11 (32%) Maximal 24-hour UFC before TSS 3.24 ( ) 3.24 ( ) 3.1 ( ) (fold above normal) Maximal LNSC a before TSS (nmol/l) 15.9 ( ) 15.7 ( ) 9.6 ( ) Abbreviation: BMI, body mass index; F, female; M, male. Data are shown as median (range) or percentages (as appropriate). a Normal level for LNSC: 4.3 nmol/l. Results Among 165 patients with at least one available LNSC level, there were 89 patients with all cortisol data (obtained within 3 months) available after TSS, including LNSC, 24-hour UFC, and morning serum cortisol. Demographic, clinical, and laboratory findings of the study population are shown in Table 1. The ratio between female to male patients was 4.93; 84% of patients (n 75) were between 21 and 65 years of age; 63% of patients had body mass index of 30 kg/m 2 or higher. Age distribution and body mass index were not significantly different between women and men. Among medical comorbidities at presentation, only the diagnosis of DM was associated with higher median levels of nadir LNSC (0.8 and 0.5 nmol/l in patients with and without DM, respectively; P.01). Of note, 79 patients (89%) presented with a microadenoma (tumor size 10 mm in greatest diameter). Of 80 histologic specimens with evidence of adenoma on pathology, 71 showed positive immunostaining for adrenocorticotropic hormone. There were nine patients without evidence of adenoma on pathology. However, five of them showed biochemical remission after initial TSS. Remission Cortisol data after TSS are shown in Table 2. LNSC was measured at a median sequence of 4 tests (range: 1 14) within 3 months after TSS, whereas 24-hour UFC was tested at a median sequence of three specimens (range: 1 9). Nadir LNSC level was 0.6 nmol/l ( ). Remission occurred in 79 patients (89%), including 87% of 79 patients with microadenomas and 100% of 10 patients with macroadenomas. There was no significant difference in the median duration of follow-up between patients with microadenomas and those with macroadenomas (data not shown). Of all patients in remission, 10 patients were considered to be in delayed remission based on low nadir 24-hour UFC ( 23 mcg), despite having nadir morning serum cortisol exceeding 5 mcg/dl within 2 weeks after TSS. Among 10 patients who were not in remission after TSS, five had a visible sellar hypodensity on MRI (presumed microadenoma) at presentation. Among the remaining five patients, IPSS was confirmative of a central source in three and nondiagnostic in two because of technical difficulties encountered during IPSS. One of these two patients underwent whole body imaging, but no ec-

4 doi: /jc press.endocrine.org/journal/jcem 3773 Table 2. Characteristics of Patients After TSS Patients Evaluated for Remission Following TSS (n 89) Patients Evaluated for Following TSS (n 68) Patients Evaluated for Long-Term During Follow-Up (n 34) Remission (n 79) No Remission (n 10) P Value (n 19) No (n 49) P Value (n 16) No (n 18) P Value Nadir morning serum cortisol within ( ) 11.8 (0.5 21) ( ) 0.7 ( ) (0.5 20) 1 ( ).0466 months after TSS (mcg/dl) Nadir UFC within 3 months after 1 ( ) 20.3 ( ) < (1 22.4) 1 ( ) (1 22.4) 1 ( ).0110 TSS (mcg/24 hours) Nadir LNSC a within 3 months after 0.6 ( ) 3.4 ( ) < ( ) 0.6 ( ) NS 1.25 ( ) 0.70 ( ) NS TSS (nmol/l) Pathology (TSS) ACTH-adenoma 67 (85%) 4 (40%) (79%) 40 (81%) NS 9 (60%) 15 (83%) NS Duration of glucocorticoid replacement 5 ( ) 1 ( ) < (0.5 12) 8.3 ( ) (0 12) 9.6 (0.7 60).0009 after TSS (months) Sequences of LNSC within 3 months 4 (1 14) 3.5 (1 7) NS 4 (2 14) 3 (1 10).006 N/A N/A N/A after TSS Maximal 24-hour UFC at recurrence N/A N/A N/A 2.12 ( ) N/A N/A N/A N/A N/A (fold above normal) Maximal LNSC at recurrence (nmol/l) N/A N/A N/A 10 ( ) N/A N/A N/A N/A N/A Maximal 24-hour UFC at last follow-up N/A N/A N/A N/A N/A N/A 1.8 ( ) 0.5 ( ).001 (fold above normal) Maximal LNSC at last follow-up (nmol/l) N/A N/A N/A N/A N/A N/A 8.5 ( ) 2.8 ( ).0002 Sequences of LNSC within 12 months of last visit N/A N/A N/A N/A N/A N/A 6 (1 12) 3 (1 14) NS Abbreviations: ACTH, adrenocorticotropic hormone; NS, nonsignificant; N/A, nonapplicable. Data are shown as median (range) or percentages as appropriate. Bold values are statistically significant (P.05). a Normal level for LNSC: 4.3 nmol/l. topic source was found; the second patient had a pituitary adenoma resected on repeat TSS. In patients who were in remission after TSS, nadir LNSC ranged between 0.3 and 3.5 nmol/l, with median values being significantly lower in patients in remission (0.6 nmol/l) in comparison with those not in remission (3.4 nmol/l, P.0001). In patients in remission, nadir LNSC was significantly higher among patients (n 26) with DM (which was present both at the time of diagnosis of CD and at the time of LNSC testing) than nondiabetic (n 53) patients (0.75 nmol/l vs. 0.5 nmol/l, respectively; P.0007). On ROC analysis, sensitivity and specificity of nadir LNSC in establishing remission were calculated at different cutpoints. A nadir LNSC cutpoint of 1.9 nmol/l established remission with 94% sensitivity and 80% specificity (area under the curve [AUC] 0.90, Figure 1). At this cutpoint, PPV and NPV were 97% and 61%, respectively. Of note, nadir 24-hour UFC and nadir morning serum cortisol individually showed 95% sensitivity and 80% specificity at the cutpoint of 6.4 mcg/24 hours and 8.2 mcg/dl, respectively (AUC 0.87, PPV 97%, NPV 66%). Sensitivity analyses were performed after excluding patients (n 10) with delayed remission after their initial TSS. The same optimal cutpoint for LNSC (1.9 nmol/l) was found to establish remission with higher sensitivity (97%) and NPV (80%), but the same specificity and PPV (AUC 0.91). The optimal cutpoint for nadir 24-hour UFC was 5 mcg/24 hours with 98% sensitivity and 80% specificity (AUC 0.88, PPV 97%, NPV 88%); the optimal cutpoint for morning serum cortisol was 4.2 Figure 1. ROC analysis for nadir LNSC (A), 24-hour UFC (B), and morning serum cortisol (C) in establishing remission (obtained within 3 months after TSS).

5 3774 Amlashi et al Late-Night Salivary Cortisol in Cushing s Disease J Clin Endocrinol Metab, October 2015, 100(10): Figure 2. ROC analysis for maximal LNSC (A) and 24-hour UFC (fold above ULN) (B) in establishing recurrence (obtained within 12 months of last follow-up). mcg/dl with 100% sensitivity and 80% specificity (AUC 0.90, PPV 97%). In addition, the accuracy of LNSC in establishing remission was examined using a somewhat lower cutpoint for nadir morning serum cortisol (1.8 mcg/dl). Using this more stringent cutpoint, the overall remission rate did not change. However, there were now 23 patients who would be classified as being in delayed remission after TSS. A second sensitivity analysis was performed after excluding these 23 patients in delayed remission. A slightly lower LNSC cutpoint (1 nmol/l) was identified, which established remission with 89% sensitivity and 90% specificity (AUC 0.93). The corresponding nadir morning serum cortisol and 24-hour UFC established remission with sensitivity of 100% (for both of them) and specificity of 90 and 80%, respectively (AUC 0.93 and 0.90, respectively). developed in 19 (28%) of 68 patients in remission with available follow-up LNSC after TSS, and was considered definite in 14 patients (21%) and probable in five (7%). occurred in 24% of 58 cases with microadenomas and 50% of 10 cases with macroadenomas. Demographic, clinical, and laboratory characteristics of this patient subgroup are shown in Table 1. occurred at a median interval of 21.7 months ( ) after initial TSS with no significant difference in time to recurrence between patients with microadenoma and macroadenoma (data not shown). In patients who were in remission after TSS and subsequently recurred (n 19), median nadir and maximal levels of LNSC obtained within 3 months after TSS were 0.7 nmol/l ( ) and 5.9 nmol/l ( ), respectively (Table 2). The corresponding levels in patients without recurrence (n 49) were 0.6 nmol/l ( ) and 1.2 nmol/l ( ), respectively. Neither nadir nor maximal early postoperative LNSC predicted recurrence. In contrast, nadir morning serum cortisol above a cutpoint of 2.4 mcg/dl predicted recurrence with sensitivity of 63% and specificity of 86% (AUC 0.79) and 24-hour UFC (median sequence: 2.5 specimens) predicted recurrence above a cutpoint of 1.5 mcg/24 hours with a poor sensitivity (42%) but higher specificity of 88% (AUC 0.66). Sensitivity analysis was performed after excluding patients (n 5) with probable recurrence after their initial TSS. Similarly, no association was seen between nadir or maximal levels of LNSC (obtained within 3 months from initial TSS) and recurrence. Thirty-four patients had available data on LNSC and 24-hour UFC, obtained within 12 months from the date of their most recent follow-up visit. Demographic, clinical, and laboratory characteristics of this patient subgroup are shown in Table 1. The follow-up interval between TSS and most recent visit was 53.5 months ( ). In this subgroup, there were 16 patients with recurrences (13 definite and three probable recurrences) at their last visit, including 43% of 23 patients with microadenoma and 54% of 11 patients with macroadenoma. Maximal LNSC obtained within 12 months of last visit was 8.5 nmol/l ( ) in patients with recurrence, which was significantly higher than maximal LNSC value in patients without recurrence (2.8 nmol/l [ ]), P.0002). Similarly, maximal 24-hour UFC in the former group (1.8-fold ULN [ ]) was significantly higher than the latter group (0.5-fold ULN [ ]; P.001). Using ROC analysis, sensitivity and specificity of LNSC in identifying recurrence were calculated at different cutpoints. At an optimal cutpoint of 7.4 nmol/l, LNSC established recurrence with 75% sensitivity and 95% specificity (AUC 0.87, Figure 2). PPV and NPV at this level were 92% and 80%, respectively. Of note, a cutpoint of 4.7 nmol/l, which is just above the upper end of the normal range provided by the laboratory, established recurrence with higher sensitivity (87%) and NPV (87%), but lower specificity (73%) and PPV (74%). The corresponding maximal 24-hour UFC (median sequence: two specimens) established recurrence with 68% sensitivity and 100% specificity at the optimal cutpoint of 1.6 fold ULN (AUC 0.82, NPV 78%). Using a cutpoint of 1.01-fold ULN for 24 UFC, recurrence was established with the same sensitivity but lower specificity (89%) and NPV (76%).

6 doi: /jc press.endocrine.org/journal/jcem 3775 Sensitivity analyses were performed after excluding patients (N 3) with probable recurrence at last follow-up visit. Maximal LNSC could still establish recurrence at the same cutpoint with the same specificity and PPV. However, higher sensitivity and NPV were found (77% and 85%, respectively). Similar trends were seen in 24-hour UFC (data not shown). Among 13 patients with definite recurrence within 12 months of last visit, there were six with abnormal LNSC that occurred earlier than 24-hour UFC (in three: months earlier and in the remaining 3 cases: all available UFCs were normal). There were four patients with abnormal 24-hour UFC earlier than LNSC (in three: 1 5 months earlier and, in one, no abnormal LNSC) and finally, in three patients, there were abnormal tests for both UFC and LNSC at the same time. Discussion In the present study, we investigated the diagnostic accuracy of LNSC during follow-up of patients with CD undergoing TSS. Our overall remission rate (89%) is consistent with previous data from our institution and other tertiary pituitary centers (8 10). At a cutpoint of 1.9 nmol/l, LNSC established remission within 3 months after TSS with 94% sensitivity and 80% specificity. Moreover, in comparison with corresponding 24-hour UFC and morning serum cortisol levels, this test had similar sensitivity and specificity in establishing remission, which is noteworthy, considering its simplicity and good reproducibility (11). Our data suggest that LNSC can be of value and may be included in the postoperative assessment for remission of patients undergoing TSS. Nadir postoperative morning serum cortisol has been suggested as a good predictor of sustained remission (12 18), as was found in our study. However, even undetectable postoperative early morning serum cortisol does not exclude the possibility of subsequent recurrence. Of note, the accuracy of 24-hour UFC as a measure of the activity of the hypothalamic pituitary adrenal axis may be limited (19). Castinetti et al reported that a combined dexamethasone suppressed-desmopressin stimulation test may accurately predict the recurrence of hypercortisolism in patients with immediate postsurgical hypoadrenalism (20). Bou Khalil et al suggested that elevated 24-hour UFC levels occur late during recurrence and corticotropin-releasing hormone or vasopressin stimulation tests, as well as midnight cortisol, might be more sensitive than 24-hour UFC in detecting patients needing closer follow-up or more extensive investigations (12). In another study, Estrada et al suggested that the complete normalization of adrenal function, including a normal cortisol circadian rhythm (based on plasma cortisol levels obtained in the morning and at midnight during long-term follow-up) predicted low recurrence risk (18). In contrast, early postoperative LNSC (obtained within 3 months) did not predict recurrence risk in our study population. During follow-up of our patients, definite recurrence of hypercortisolism occurred in 21% of patients. rates ranging between 7 and 25% have been reported in previous studies (8, 10, 21). In our patient population, LNSC (obtained within 12 months from most recent follow-up) identified recurrence with a diagnostic accuracy exceeding that of 24-hour UFC. At a cutpoint of 7.4 nmol/l, LNSC established recurrence with higher diagnostic accuracy compared to the corresponding 24-hour UFC. Of note, 7% of our patients (n 10) were considered to be in delayed remission based on nadir 24-hour UFC ( 23 g/24 hours), despite having morning serum cortisol exceeding 5 mcg/dl within 2 weeks after TSS. These data are in agreement with those from a previous study from our institution, which found that 5.6% of patients with CD had delayed remission after TSS (9). Sensitivity analysis in the present study showed no change in the accuracy of LNSC after excluding patients in delayed remission. The presence of delayed remission in some patients with CD suggests that early hormonal assessment may not always reflect ultimate outcomes (9, 14, 22). Few other studies have investigated the performance of LNSC after surgery (4, 5, 23, 24), and only one study compared it with other cortisol tests (4). In a study of 68 patients with CD by Carrasco et al, significant differences in LNSC were found between patients in remission and those with recurrent disease. They identified that an LNSC cutpoint of 2 ng/ml (5.52 nmol/l), measured by radioimmunoassay, within 6 to 12 months after TSS had a sensitivity of 100% and specificity of 98% in detecting persistent CD. The diagnostic performance of LNSC was higher than that of 24-hour UFC or early postsurgical morning serum cortisol (4). In that study, remission and recurrence were not investigated separately. Ceccato et al reported higher LNSC in newly diagnosed patients with CD (n 41) than in those with CD in remission (n 27). They also found that patients with CD in remission had LNSC less than 5.24 ng/ml and patients with recurrence (n 11) had higher LNSC levels than those in remission. They stated that LNSC (which was measured once for each patient during the 12 months after surgery) may represent a simple and accurate test to follow patients with CD (23). In another study, Nunes et al applied a higher threshold of 8 nmol/l for LNSC in an outpatient setting and found that this

7 3776 Amlashi et al Late-Night Salivary Cortisol in Cushing s Disease J Clin Endocrinol Metab, October 2015, 100(10): test had 90% sensitivity and 91.8% specificity in identifying recurrence after TSS. They also found that the diagnostic performance of LNSC was not statistically different from that of midnight plasma cortisol (24). In a recent study by Danet-Lamasou et al, the accuracy of LNSC was examined in 36 patients with CD who were in remission within 3 months after surgery or in early recurrence at the time of saliva sampling. Using the maximal level of four LNSC tests, with a 6.2 nmol/l cutpoint, LNSC could predict recurrence with 90% sensitivity and 90.9% specificity. Similar results were obtained when the mean of successive measurements was used (5). LNSC can be easily collected by patients on multiple occasions. In contrast, 24-hour UFC is rather inconvenient for patients to perform, requires meticulous attention to assure a complete collection, and can be compromised by renal dysfunction and high urine volumes (25 27). In our study, the median number of 24-hour UFC specimens (sequences) was at least two. Several 24-hour UFC collections are needed, in light of the substantial variability between successive 24-hour UFC levels in patients with CD, as was recently reported in a study by Petersenn et al (28). The median of LNSC sequences was four. Patients in recurrence had been tested significantly more than patients without recurrence. Although some previous reports on the diagnostic accuracy of LNSC are based on a single measurement, Carrasco et al suggested that the performance of this test is improved by obtaining two samples (29). Similar to our study, Danet-Lamasou et al performed LNSC up to four times and pointed out that four measurements of LNSC, rather than fewer collections, yield optimal specificity with 90% sensitivity (5). In two other studies, using different assays for LNSC measurement, more than 10% of patients with overt CS had had at least one normal LNSC among several samples (5, 29, 30). It has also been suggested that variability in cortisol secretion may be increased during hypercortisolism and during recurrence compared to patients with newly diagnosed, overt CD (5, 28, 29). There are some challenges in using LNSC in clinical practice, including the lack of consensus on diagnostic cutpoints during follow-up of patients with CD. The present study yielded novel insights in the diagnostic accuracy of LNSC after TSS, both within 3 months and during longterm follow-up, and identified diagnostic cutpoints for remission and recurrence of CD employing a widely used (in the United States) LNSC immunoassay. Potential limitations of LNSC in the assessment of hypercortisolism include the effect of stressors, differences in lifestyle (third shift work), abnormal sleep patterns, and prior addiction, all of which could confound test results (10, 31, 32). Liu et al found that elderly male veterans with type 2 DM had an increased frequency of false-positive salivary cortisol results (33). In our study, diabetic (n 26) patients had slightly higher levels of nadir LNSC (obtained within 3 months of TSS) than nondiabetic (n 53) patients (0.8 vs. 0.5 nmol/l, respectively). Because the present study was carried out at a tertiary center for patients with CD, our study was larger than several previous reports with regard to population size or follow-up duration (4, 5, 23, 24). In addition, all TSS were performed by a single experienced neurosurgeon and all LNSC levels were measured in the same laboratory. However, its retrospective nature necessitates the exclusion of patients without available data. Because DST was not routinely performed in our patients postoperatively, we could not compare the accuracy of LNSC with that of DST. Of the 165 patients reviewed, only 68 of those initially in remission returned to the clinic and had adequate longterm follow-up data including LNSC tests. This may represent a selection bias because patients who had experienced symptoms suggestive of recurrence might be more likely (than those who were asymptomatic) to return to the clinic and undergo follow-up endocrine testing, thus potentially overestimating recurrence rates. In addition, prospectively collected, long-term LNSC data from this asymptomatic group, if or when available, may help us better establish recurrence rates and further define optimal LNSC levels diagnostic of recurrence. In conclusion, LNSC is not only accurate in the diagnosis of CD (as previously reported), but also in establishing remission after TSS and identifying recurrence of hypercortisolism during follow-up. During long-term follow-up, this test is more accurate than 24-hour UFC in detecting recurrence. Prospective studies in large groups of patients with CD would be helpful to confirm the diagnostic accuracy of LNSC in establishing remission and predicting recurrence of hypercortisolism after TSS. Acknowledgments Address all correspondence and requests for reprints to: Nicholas A. Tritos, MD, DSc, Massachusetts General Hospital, Neuroendocrine Clinical Center, Zero Emerson Place #112, Boston, MA, ntritos@mgh.harvard.edu. Disclosure Summary: N.A.T. has received institution-directed research funding from Pfizer, Ipsen, and Novo Nordisk. B.M.K.B. is the principal investigator of research grants to Massachusetts General Hospital from Cortendo and Novartis and is an occasional consultant to Cortendo, HRA Pharma, and Novartis. F.G.A., B.S., A.T.F., L.B.N., K.K.M., and A.K. have no conflicts of interest to disclose.

8 doi: /jc press.endocrine.org/journal/jcem 3777 References 1. Nieman LK, Biller BM, Findling JW, et al. The diagnosis of Cushing s syndrome: an Endocrine Society Clinical Practice Guideline. J Clin Endocrinol Metab. 2008;93: Ayala A, Manzano AJ. Detection of recurrent Cushing s disease: proposal for standardized patient monitoring following transsphenoidal surgery. J Neurooncol. 2014;119: Roelfsema F, Biermasz NR, Pereira AM. Clinical factors involved in the recurrence of pituitary adenomas after surgical remission: a structured review and meta-analysis. Pituitary. 2012;15: Carrasco CA, Coste J, Guignat L, et al. Midnight salivary cortisol determination for assessing the outcome of transsphenoidal surgery in Cushing s disease. J Clin Endocrinol Metab. 2008;93: Danet-Lamasou M, Asselineau J, Perez P, et al. Accuracy of repeated measurements of late-night salivary cortisol to screen for early-stage recurrence of Cushing s disease following pituitary surgery. Clin Endocrinol (Oxf). 2015;82(2): Raff H, Homar PJ, Skoner DP. New enzyme immunoassay for salivary cortisol. Clin Chem. 2003;49: Raff H, Raff JL, Findling JW. Late-night salivary cortisol as a screening test for Cushing s syndrome. J Clin Endocrinol Metab. 1998; 83: Swearingen B, Biller BM, Barker FG, et al. Long-term mortality after transsphenoidal surgery for Cushing disease. Ann Intern Med. 1999; 130: Valassi E, Biller BM, Swearingen B, et al. Delayed remission after transsphenoidal surgery in patients with Cushing s disease. J Clin Endocrinol Metab. 2010;95: Tritos NA, Biller BM. Cushing s disease. Handb Clin Neurol. 2014; 124: Bukan AP, Dere HB, Jadhav SS, et al. The performance and reproducibility of late-night salivary cortisol estimation by enzyme immunoassay for screening cushing disease. Endocr Pract. 2015;21: Bou Khalil R, Baudry C, Guignat L, et al. Sequential hormonal changes in 21 patients with recurrent Cushing s disease after successful pituitary surgery. Eur J Endocrinol. 2011;165: Yap LB, Turner HE, Adams CB, Wass JA. Undetectable postoperative cortisol does not always predict long-term remission in Cushing s disease: a single centre audit. Clin Endocrinol (Oxf). 2002;56: Chee GH, Mathias DB, James RA, Kendall-Taylor P. Transsphenoidal pituitary surgery in Cushing s disease: can we predict outcome? Clin Endocrinol (Oxf). 2001;54: Pereira AM, van Aken MO, van Dulken H, et al. Long-term predictive value of postsurgical cortisol concentrations for cure and risk of recurrence in Cushing s disease. J Clin Endocrinol Metab. 2003; 88: Trainer PJ, Lawrie HS, Verhelst J, et al. Transsphenoidal resection in Cushing s disease: undetectable serum cortisol as the definition of successful treatment. Clin Endocrinol (Oxf). 1993;38: Esposito F, Dusick JR, Cohan P, et al. Clinical review: Early morning cortisol levels as a predictor of remission after transsphenoidal surgery for Cushing s disease. J Clin Endocrinol Metab. 2006;91: Estrada J, García-Uría J, Lamas C, et al. The complete normalization of the adrenocortical function as the criterion of cure after transsphenoidal surgery for Cushing s disease. J Clin Endocrinol Metab. 2001;86: Alexandraki KI, Grossman AB. Is urinary free cortisol of value in the diagnosis of Cushing s syndrome? Curr Opin Endocrinol Diabetes Obes. 2011;18: Castinetti F, Martinie M, Morange I, et al. A combined dexamethasone desmopressin test as an early marker of postsurgical recurrence in Cushing s disease. J Clin Endocrinol Metab. 2009;94: Utz AL, Swearingen B, Biller BM. Pituitary surgery and postoperative management in Cushing s disease. Endocrinol Metab Clin North Am. 2005;34: , xi 22. Toms GC, McCarthy MI, Niven MJ, Orteu CH, King TT, Monson JP. Predicting relapse after transsphenoidal surgery for Cushing s disease. J Clin Endocrinol Metab. 1993;76: Ceccato F, Barbot M, Zilio M, et al. Performance of salivary cortisol in the diagnosis of Cushing s syndrome, adrenal incidentaloma, and adrenal insufficiency. Eur J Endocrinol. 2013;169: Nunes ML, Vattaut S, Corcuff JB, et al. Late-night salivary cortisol for diagnosis of overt and subclinical Cushing s syndrome in hospitalized and ambulatory patients. J Clin Endocrinol Metab. 2009; 94: Doi SA, Clark J, Russell AW. Concordance of the late night salivary cortisol in patients with Cushing s syndrome and elevated urine-free cortisol. Endocrine. 2013;43: Inder WJ, Dimeski G, Russell A. Measurement of salivary cortisol in laboratory techniques and clinical indications. Clin Endocrinol (Oxf). 2012;77: Mericq MV, Cutler GB. High fluid intake increases urine free cortisol excretion in normal subjects. J Clin Endocrinol Metab. 1998; 83: Petersenn S, Newell-Price J, Findling JW, et al. High variability in baseline urinary free cortisol values in patients with Cushing s disease. Clin Endocrinol (Oxf). 2014;80: Carrasco CA, García M, Goycoolea M, et al. Reproducibility and performance of one or two samples of salivary cortisol in the diagnosis of Cushing s syndrome using an automated immunoassay system. Endocrine. 2012;41: Zerikly RK, Amiri L, Faiman C, et al. Diagnostic characteristics of late-night salivary cortisol using liquid chromatography-tandem mass spectrometry. J Clin Endocrinol Metab. 2010;95: Raff H. Utility of salivary cortisol measurements in Cushing s syndrome and adrenal insufficiency. J Clin Endocrinol Metab. 2009; 94: Garde AH, Persson R, Hansen AM, et al. Effects of lifestyle factors on concentrations of salivary cortisol in healthy individuals. Scand J Clin Lab Invest. 2009;69: Liu H, Bravata DM, Cabaccan J, Raff H, Ryzen E. Elevated latenight salivary cortisol levels in elderly male type 2 diabetic veterans. Clin Endocrinol (Oxf). 2005;63: