Primary empty sella syndrome (ESS) is an anatomical. Effect of primary empty sella syndrome on pituitary surgery for Cushing s disease

Transcription

1 J Neurosurg 121: , 2014 AANS, 2014 Effect of primary empty sella syndrome on pituitary surgery for Cushing s disease Clinical article Gautam U. Mehta, M.D., 1,2 Kamran D. Bakhtian, M.S., 1 and Edward H. Oldfield, M.D. 1,2 1 Surgical Neurology Branch, National Institutes of Neurological Disorders and Stroke, National Institutes of Health, Bethesda, Maryland; and 2 Department of Neurosurgery, University of Virginia Health Sciences Center, University of Virginia, Charlottesville, Virginia Object. Primary empty sella syndrome (ESS) results from herniation of arachnoid mater into the pituitary fossa. It has been suggested to have a negative effect on pituitary surgery; however, outcomes in this cohort have not been defined. This study was performed to determine the effect of ESS on immediate and long-term biochemical outcome after pituitary surgery for Cushing s disease (CD). Methods. Using a matched cohort study design, the authors followed patients treated with pituitary surgery for CD with and without ESS. Complete ESS was defined as pituitary gland height 2 mm, whereas partial ESS was defined as pituitary gland height > 2 mm but less than three-quarters of the total sellar depth. The primary end points were immediate and long-term biochemical outcome. Cerebrospinal fluid leaks were recorded as a secondary end point. Results. Seventy-eight patients with CD and primary ESS were identified and matched with 78 patients with CD without ESS. After surgical management, immediate biochemical remission was achieved in 69 patients (88%) with ESS and 75 controls (96%, p = 0.10). Long-term remission was achieved in most patients in both groups (5-year cure: 85% vs 92%, p = 0.10). Among patients with ESS, the presence of complete ESS predicted a worse long-term outcome (p = 0.04). Intraoperative CSF leaks were significantly more frequent with ESS (54% vs 24%, p < 0.001), and despite sellar floor repair, the rate of postoperative CSF leaks was also increased (6% vs 3%, p = 0.27). Conclusions. Biochemical outcome after pituitary surgery for CD was worse in patients with complete ESS, and the risk of a CSF leak was increased with both partial and complete ESS. However, as outcome remains superior to those following alternative therapies and the biology of these tumors is unchanged in the setting of ESS, pituitary surgery should remain the initial treatment of choice. ( Key Words Cushing s disease empty sella syndrome pituitary adenoma transsphenoidal surgery pituitary surgery Primary empty sella syndrome (ESS) is an anatomical variation caused by herniation of suprasellar arachnoid mater and CSF into the pituitary fossa, which occurs in 5.5% 28% of the general population. 5,9,14,22 Empty sella syndrome can be partial or complete depending on the degree to which CSF occupies the intrasellar space. 14,22 Primary ESS is typically an incidental finding and infrequently exhibits clinical manifestations on its own; however, primary ESS and pituitary tumors can occur concurrently. 12,15 As large tumors can fill the sella and eliminate any arachnoid herniation, ESS is more likely to present with microadenomas and hormonesecreting tumors. Abbreviations used in this paper: ACTH = adrenocorticotropin; BMI = body mass index; CD = Cushing s disease; CRH = corticotropin-releasing hormone; DI = diabetes insipidus; ESS = empty sella syndrome; GH = growth hormone; SIADH = syndrome of inappropriate antidiuretic hormone; UFC = urinary free cortisol. To date, treatment outcomes for pituitary tumors in the setting of primary ESS remain undefined, described only in case reports and small series. 1,2,4,7,12,15,19,20,26,27 Pituitary surgery, which is first-line therapy for adrenocorticotropin (ACTH)- and growth hormone (GH)-secreting tumors, is particularly challenging in the setting of ESS due to diminished pituitary gland volume, as well as the frequent presence of an arachnoid pouch in the surgical field. Given these obstacles it has been suggested that pituitary surgery may be less effective in the setting of ESS, and that medical or radiation therapy should be considered in some cases, particularly with complete ESS. 20,26 To determine the outcome and utility of pituitary surgery in the setting of ESS, we analyzed a cohort of patients treated by a single surgeon for Cushing s disease (CD). Biochemical outcomes of surgery for patients with ESS were compared with those for a contemporaneous cohort of patients without ESS. Overall, ESS is more common in 518 J Neurosurg / Volume 121 / September 2014

2 Cushing s disease and empty sella syndrome older patients and thought to be more frequent with smaller tumors. 11,12,16 As the distributions of these variables are likely to be different between cohorts of patients with and without ESS, controls were chosen by matching on these two variables. Methods Patients Patients who underwent pituitary surgery for ACTHsecreting pituitary adenomas by a single surgeon (E.H.O.) at the National Institutes of Health (protocols 79-N-0089 and 03-N-0164) and the University of Virginia (study #13737) were identified. Patients were accrued and data were collected prospectively between 1984 and Informed consent to participate in clinical research studies was obtained from all patients. Patients with ESS and evaluable imaging were identified and included in the present study. Patients who had undergone prior pituitary surgery or irradiation were excluded to eliminate secondary causes of ESS. Controls without ESS or prior treatment directed at the pituitary were selected from the same cohort of patients with CD. To minimize variation in tumor biology, cases with atypical histology (defined by numerous mitotic figures or an MIB-1 proliferation index > 3%) or cases associated with tumor predisposition syndromes (such as MEN-1) were excluded. 35 Diagnosis of CD Cushing s disease was diagnosed based on biochemical evidence of hypercortisolism (elevated 24-hour urinary free cortisol [UFC], absence of diurnal serum cortisol variation, as well as corticotropin-releasing hormone [CRH] stimulation tests) as well as the presence of a pituitary lesion on MRI. 3 Inferior petrosal sinus sampling was used to confirm a pituitary source of ACTH in patients with negative or equivocal imaging, based on a basal central-to-peripheral gradient > 2.0 or a gradient > 3.0 after CRH administration. 30 Imaging Patients underwent pre- and postcontrast T1-weighted MRI and spin echo sequences of the pituitary, as well as spoiled gradient recalled (SPGR) sequences after Initial radiologist impressions, including tumor identification, size, and cavernous sinus involvement, were recorded. Categorization of ESS according to pituitary gland thickness was based on preoperative T1-weighted MRI. Complete ESS was defined as pituitary gland height 2 mm, and partial ESS was defined as pituitary gland height > 2 mm but less than three-quarters of the total sellar depth. 14 Measurements were made in the midsagittal plane and at the deepest portion of the gland for consistency. J Neurosurg / Volume 121 / September 2014 Surgical Technique All patients underwent pituitary adenoma resection via a sublabial, endonasal transsphenoidal approach using the operating microscope. When possible, tumors were resected using the histological pseudocapsule as a surgical capsule, as previously described. 31 Fluoroscopy was used to confirm the trajectory to the sella, and intraoperative ultrasound was typically used for tumor localization. The dura mater often was opened inferiorly to minimize exposure of the arachnoid pouch. Intraoperative factors, such as tumor size, cavernous sinus or dural invasion, and the presence or absence of an intraoperative CSF leak, were carefully recorded during surgery. In cases with cavernous sinus involvement, the medial wall of the cavernous sinus was removed and, when safe, the sinus was explored for free tumor. When no discrete adenoma was found, a partial hypophysectomy was performed. Biochemical Outcome Postoperative evaluation of biochemical remission included serum cortisol levels and 24-hour UFCs 3 5 days after surgery. 3,23,25 Postoperative hypocortisolism was defined as a morning serum cortisol nadir 5 μg/dl and/ or 24-hour UFC less than the contemporary reference range. 23,25 Eucortisolism was defined as morning serum cor tisol and 24-hour UFC within the reference range. Long-term outcome was determined through clinic visits with serological testing, patient phone calls, and written sur veys. Date of the last follow-up varied as available but was no later than July Statistical Analysis Statistical analyses were done using the software R (version 3.0.0, The R Foundation for Statistical Computing). Patients with ESS were matched at a 1:1 ratio with controls without ESS by propensity-score matching using a nearest-neighbor algorithm. 18 To avoid overmatching, cases were matched only on variables thought to be related to both the incidence of ESS and outcome. 8 Cases in which no tumor was demonstrated on MRI were matched with MRI-negative patients based on age. Cases with a tumor noted on MRI were matched on both age and tumor size. Balance was visualized by quantile-quantile plots and defined by a standardized mean difference less than Conditional logistic regression was used to compare predictors with immediate outcome while controlling for potential confounders (cavernous sinus and dural invasion) and accounting for matched data. Survival analyses using Cox proportional hazards accounting for matched data were used to measure recurrence-free survival. Events were recorded at Time 0 if early surgical management (initial surgery plus early repeat surgery) did not result in eucortisolism or hypocortisolism or at the time that a recurrence of CD was diagnosed. Patients were otherwise censored at the time of the last follow-up. The relationship between pre- and intraoperative factors and immediate outcome was defined using Fisher s exact test and logistic regression for binary and continuous predictors, respectively. The effect on long-term outcome was assessed using a Cox proportional hazards model. Degree of ESS was analyzed as a continuous variable and grouped as either partial or complete based on diagnostic convention. Statistical significance was determined by a 2-sided p value less than

3 G. U. Mehta, K. D. Bakhtian, and E. H. Oldfield Patients Results Four patients with ESS whose preoperative imaging was no longer available at the time of analysis were excluded. Seventy-eight patients with ESS who had undergone initial pituitary surgery for CD and had evaluable imaging were identified and included in the study. Propensity-score matching based on age and tumor size was used to match CD patients with ESS to controls from a cohort of 469 patients with evaluable imaging. Seventyeight CD patients without ESS who had undergone pituitary surgery performed by the same surgeon were identified. Matched baseline characteristics between patients with ESS and controls were similar (Fig. 1). Unmatched baseline characteristics, including sex, body mass index (BMI), and symptoms at presentation, were also similar between patients with ESS and matched controls (Table 1). Relative surgical experience during the 30-year study period, defined as time from the start of the study (1984) to the date of surgery, was similar between the two groups. Imaging Characteristics Of the patients with ESS, 68 (87%) had partially empty sellas and 10 (13%) had completely empty sellas (Fig. 2). Fewer tumors appeared to involve the cavernous sinus on imaging in cases with ESS (24%) than without ESS (35%). Twelve patients with ESS and negative MRI demonstrated a clear lateral difference in gland height (Fig. 3). At the time of surgery, an increased lateral gland height was not found to be specific for tumor location, as differential gland height correctly localized the side of the tumor in 6 patients (50%) and incorrectly localized the side of the tumor in 3 patients (25%). No tumor was found at surgery in 3 patients (25%). Surgical Characteristics Surgical findings, including the identification of a discrete adenoma, tumor size, and histopathological results, were similar between patients with ESS and controls (Table 2). Findings of prior hemorrhage, suggesting tumor apoplexy, were infrequent but more common in patients with ESS (4% vs 1%). Cavernous sinus invasion and noncavernous sinus dural invasion were slightly more frequent in patients with ESS (19% and 13%, respectively) than in matched controls (17% and 9%, respectively). One patient with ESS was found to have an extra-pituitary parasellar microadenoma contained completely within the medial cavernous sinus wall and the cavernous sinus itself. No connection to the pituitary gland was identified. Perceived gross-total resection was more frequent in the control group, whereas hemihypophysectomy, performed when no tumor was identified at surgery, was performed more often in patients with ESS. Intraoperative CSF leaks were more frequent in patients with ESS than in controls (54% vs 24%; HR 3.3, 95% CI , p < 0.001). Intraoperative CSF leaks were more frequent in patients with either complete (3 [33%] of 10 cases) or partial (39 [57%] of 68 cases) ESS, compared with controls. This led to an increased rate of sellar floor repair, including the use an abdominal fat graft (37% vs 22%), and an increased rate of lumbar drain placement (18% vs 5%). A CSF leak caused by direct arachnoid dissection for tumor resection, gland exploration, or hypophysectomy was more common in patients with ESS (27 cases) than controls (13 cases). Intraoperative leaks occurred on dural opening in 5 patients with ESS (6%), as compared with none in the control group. Biochemical Outcome Immediate Biochemical Outcome. Sixty-nine (88%) of 78 controls achieved immediate biochemical remission after initial pituitary surgery (Table 3). Eight of 9 patients with persistent hypercortisolism underwent early reoperation, which was successful in 6 patients. In the 3 remaining patients, hypercortisolism was managed with radiation therapy (2 cases) and/or medical therapy (2 cases). Sixty-three (81%) of 78 patients with ESS achieved immediate biochemical remission after initial pituitary surgery. Eleven of 15 patients with persistent hypercortisolism underwent early reoperation, which was successful in 6 patients. Hypercortisolism was managed in 8 of the 9 remaining patients using radiation therapy (6 cases), medical therapy (9 cases), and/or bilateral adrenalectomy (1 case). One patient died of medical complications before definitive control of her hypercortisolism was achieved. Immediate biochemical remission was more frequent in controls than in patients with ESS both after initial surgery (HR 1.99, p = 0.20) and after overall early surgical management (initial surgery plus early repeat surgery in select cases, HR 6.00, p = 0.10). Fifty-seven (84%) of 68 patients with partial ESS achieved immediate biochemical remission. Eight of 11 patients with persistent hypercortisolism underwent early reoperation, which was successful in 5 patients. Six (60%) of 10 patients with complete ESS achieved immediate biochemical remission. Three of 4 patients with persistent hypercortisolism underwent early reoperation, which was successful in one patient. Long-Term Biochemical Outcome. Follow-up data (me dian 4.1 years, range 3 months 20 years) were available in 141 patients (90%), including 72 patients with ESS (92%) and 69 controls (88%). To determine the overall utility of surgery in these cohorts, treatment was recorded as a failure if hypercortisolism persisted after initial surgical management (including early repeat surgery) or recurred during follow-up. Treatment failures were noted in 12 (17%) of 72 patients with ESS compared with 6 (9%) of 69 controls. Persistent hypercortisolism and late recurrence most frequently occurred because a discrete adenoma could not be identified at surgery or on histopathological analysis (10 of 18 cases overall) or because of extensive parasellar disease in patients with ESS (4 of 12 cases with ESS; Table 4). Two patients with ESS and no adenoma identified at surgery were later found to have extra-pituitary parasellar microadenomas. One patient developed Nelson s syndrome after bilateral adrenalectomy with tumor growth completely within the cavernous sinus, and another patient was found to have an ACTH-positive microadenoma within the inferior dura of the sella at au- 520 J Neurosurg / Volume 121 / September 2014

4 Cushing s disease and empty sella syndrome Fig. 1. Quantile-quantile plots of matching parameters before and after propensity-score nearest-neighbor matching for 34 patients with a pituitary adenoma found on MRI (A) and 44 patients with negative imaging (B). Patients with ESS are represented on the y-axis and controls are represented on the x-axis. Although a lack of balance between patients with ESS and controls is present before matching (left), after matching (right) the means of both age (standardized difference = 0.063) and tumor size (standardized difference = 0.036) were similar. topsy. Survival analysis demonstrated diminished recurrence-free survival in patients with ESS (2, 5, and 10-year recurrence-free survival: 88%, 85%, and 71%) compared with controls (2, 5, and 10-year recurrence-free survival: 92%, 92%, and 92%; p = 0.10; Fig. 4). At the last follow-up, recurrence-free survival was observed in 56 (89%) of 63 of patients with partial ESS and 5 (56%) of 9 of patients with complete ESS. Factors Affecting Outcome in Patients With ESS. Preoperative and surgical factors were assessed with univariate analyses for their effect on outcome in patients with ESS (Table 5). A multivariate model was not used because J Neurosurg / Volume 121 / September 2014 of the limited number of events. No factors significantly predicted immediate outcomes in this cohort, but the presence of complete ESS predicted a significantly greater likelihood of surgical failure or recurrence (HR 3.68, 95% CI , p = 0.04). Complications Persistent postoperative CSF leaks were more common in patients with ESS (5 cases [6.4%]) than in controls (2 cases [2.6%]), although this difference was not significant (HR 2.5, 95% CI , p = 0.27). Postoperative CSF leaks were more frequent in patients with either 521

5 G. U. Mehta, K. D. Bakhtian, and E. H. Oldfield TABLE 1: Summary of baseline characteristics in 156 patients with or without ESS* Variable No. (%) Patients w/ ESS Controls total clinical characteristics age at surgery in yrs mean 45.5 ± ± 14.7 range pediatric patients (age 18 yrs) 5 (6) 5 (6) females 63 (81) 62 (79) mean BMI in kg/m ± ± 11.0 presented w/ headache 10 (13) 7 (9) presented w/ endocrine dysfunction 19 (24) 26 (33) mean yrs from start of study to surgery 14.8 ± ± 6.8 imaging characteristics degree of ESS complete 10 (13) partial 68 (87) mean pituitary gland thickness in mm 4.4 ± 1.6 negative findings on MRI 44 (56) 44 (56) mean greatest tumor diameter in mm 5.0 ± ± 2.8 cavernous sinus invasion 8 (24) 12 (35) mean follow-up in yrs 4.1 ± ± 5.6 * BMI = body mass index; = not applicable. Means followed by ± SD. Includes only the 73 adults in each group. Percent of MRI-positive tumors only. Includes 72 patients with ESS and 69 controls with at least 3 months of follow-up. complete (1 [10%] of 10 cases) or partial (4 [6%] of 68 cases) ESS, compared with controls. Postoperative CSF leaks were treated with lumbar drainage (5 cases) or reoperation (2 cases). Transient diabetes insipidus (DI) was more common in patients with ESS (18 cases [23%]) than without (13 cases [17%]). Permanent DI developed in 2 patients with ESS. Syndrome of inappropriate antidiuretic hormone (SIADH) was also more common after surgery in patients with ESS (22 cases [28%]) than without (15 cases [19%]). As pituitary function other than cortisol secretion was not tested uniformly throughout the study period, they were not assessed. Among the two cohorts, other complications included urinary tract infection (4 cases [3%]), myocardial infarction (1 case), pulmonary embolism (1 case), epistaxis requiring transfusion (1 case), aspiration pneumonia (1 case), maxillary bone fracture (1 case), and severe pancreatitis and sepsis leading to death in one patient. The rate of nonendocrine complications was not significantly different in patients with ESS (10 cases [13%]) than without (6 cases [8%]; p = 0.43). Discussion Empty sella syndrome was first described by Busch in 1951 and has been detected in 5.5 to 28% of subjects on imaging and at autopsy. 5,9,14,22 The etiology of primary ESS is unknown but likely requires a congenitally incompetent diaphragma sellae and may be related to increased intracranial pressure or changes in pituitary gland volume associated with pregnancy. 11 Empty sella syndrome is an incidental finding in many patients but may be associated with symptoms of headache and endocrine disturbance. 16 It is more frequent in adults, females, and patients who are obese. 11,16,21 Given the prevalence of ESS in the general population, some pituitary adenomas are likely to present concurrently with ESS. Macroadenoma apoplexy and subsequent volume loss have been implicated as an etiology of ESS in some coincident cases. 24 Alternatively, Wilson and colleagues have suggested that ESS may be less frequent with pituitary adenomas, which fill the sella with soft tissue and may essentially reverse the arachnoid herniation. 12 Consistent with reported cases, pituitary adenomas reported in conjunction with ESS are more likely to be microadenomas. 12,15 Clinically relevant microadenomas are most frequently ACTH- or GH-secreting tumors found in CD and acromegaly, respectively. Study Design The present study is the largest reported series of CD in the setting of ESS and is among the largest series on Fig. 2. Midsagittal MR images demonstrating a normal pituitary gland (A), partial ESS (B), and complete ESS (C). Partial ESS was defined as pituitary gland height > 2 mm; complete ESS, as pituitary gland height 2 mm. 522 J Neurosurg / Volume 121 / September 2014

6 Cushing s disease and empty sella syndrome Fig. 3. Coronal MR images demonstrating lateral differences in pituitary gland height with ESS. These height differences did not predict the tumor location at surgery in patients with CD and otherwise negative imaging. Twelve patients with negative imaging had lateral differences in gland height. In 6 patients an increased lateral gland height correctly predicted tumor location (A, tumor on right), in 3 patients it incorrectly predicted tumor location (B, tumor on left), and in 3 patients no tumor was found at surgery (C). TABLE 2: Surgical findings and tumor characteristics in patients with or without ESS* Variable J Neurosurg / Volume 121 / September 2014 No. (%) Patients w/ ESS Controls tumor identified 66 (85) 68 (87) posterior lobe tumor 2 (3) 1 (1) mean tumor size in mm ± SD 6.2 ± ± 3.8 prior hemorrhage 3 (4) 1 (1) GTR achieved 61 (78) 67 (86) hemihypophysectomy performed 13 (17) 8 (10) ACTH-positive adenoma by pathology 63 (81) 61 (78) cavernous sinus invasion 15 (19) 13 (17) other dural invasion 10 (13) 7 (9) intraop CSF leak 42 (54) 19 (24) abdominal fat graft used 29 (37) 17 (22) lumbar drain used 14 (18) 4 (5) * GTR = gross-total resection. Includes patients with a tumor identified at surgery only. ESS alone. 11,16 Patients with ESS in this study were prospectively identified from a cohort of patients undergoing pituitary surgery. Data analysis was performed retrospectively and was guided by an improved understanding of the biology of CD over the 30-year study period. To limit bias and allow for the generalizability of results to other centers, a matched cohort design was used to compare the results of surgery performed by a single surgeon in patients with and without ESS. Although patients were accrued over a 30-year study period, the cohorts were contemporaneous and surgical experience was similar (Table 1). Patients were matched by age at surgery and tumor size to maintain balanced distributions of these covariates between the two groups. Patients with ESS generally present at a later mean age (52 years) than those with CD (34 years). 11,36 Recently, in a pediatric cohort with CD, an older age at presentation was associated with a worse outcome. 25 Therefore, an imbalance in the distribution of this covariate could influence outcome. Similarly, as tumor tissue may fill an empty sella, the distribution of tumor size is likely to be skewed toward smaller tumors in patients with ESS. As a smaller tumor size may lead to greater numbers of MRI-negative tumors, this factor may also affect outcome. As patients with CD and ESS both tend to be female and obese, these factors were similar between the groups. Imaging Findings As preoperative imaging was not available for every patient who underwent initial surgery for CD at the time of study completion, the overall coincidence of CD and ESS was not determined. Among our cohort of patients with CD and ESS (78), 68 patients (87%) had partial ESS and 10 patients (13%) had complete ESS. Prior studies have reported that 35% 44% of patients with ESS have complete ESS. 11,14 The reduced incidence of complete ESS in this cohort may be attributable to more stringent criteria to distinguish complete and partial ESS. Our study used a strict cutoff of a pituitary gland height 2 mm at the deepest portion of the sella in the midsagittal plane to differentiate groups. Prior studies have variably used a 2-mm-height cutoff as well as a pituitary gland volume of less than 50% of the sella. 11 This definition leaves room for differences in interpretation as well as increased interobserver variation with measurement. Alternatively, the relatively low rate of complete ESS may be specific to CD, as tumor tissue in the sella can prevent complete flattening of the gland. Microadenomas in CD can be difficult to visualize and are frequently not identifiable on MRI. 3 Certain nontumor features, such as the direction of pituitary stalk deviation, can aid in the localization of pituitary adenomas. We postulated that lateral differences in gland height in the setting of ESS could be used to identify some MRInegative tumors in CD. These differences in height only predicted tumor location (confirmed by surgical identification of an ACTH-positive pituitary adenoma) in 50% of cases (Fig. 3). The lack of correlation may be explained by the small size of the pathologic lesions in CD, which can lead to negligible changes in gland height. 523

7 G. U. Mehta, K. D. Bakhtian, and E. H. Oldfield TABLE 3: Immediate biochemical outcome End Point Total No. of Patients per Group No. (%) Patients w/ ESS Matched Controls Crude OR (95% CI) Adjusted OR* (95% CI) p Value initial surgical cure (81) 69 (88) 1.83 ( ) 1.99 ( ) 0.20 partial ESS (84) 61 (90) complete ESS 10 6 (60) 8 (80) overall early surgical cure (88) 75 (96) 7.00 ( ) 6.00 ( ) 0.10 partial ESS (91) 66 (97) complete ESS 10 7 (70) 9 (90) * Adjusted for cavernous sinus invasion and other dural invasion noted at the time of surgery. Results of initial surgery plus early repeat surgery in select cases. Management of Patients With CD and ESS Although pituitary surgery is first-line therapy for CD, recent reports have suggested that alternative therapies can be used initially for pituitary adenomas in the setting of ESS. 17,20,26 Manavela and colleagues reported surgical outcomes in 6 patients with CD and ESS but also described 5 other patients who were treated initially with medical therapy. 26 The presence of complete ESS was cited as a reason for alternative therapy. Other authors have used focused radiation or somatostatin analogs as initial therapy for patients with GH-secreting adenomas in the setting of ESS, although pituitary surgery is also the initial treatment of choice for acromegaly. 17,20,29 In the present study, we compared the outcome of pituitary surgery for CD in patients with ESS with the outcome in controls without ESS. After initial surgery, immediate biochemical remission was achieved in 81% of patients with ESS and 88% of controls (p = 0.20; Table 3). Early repeat surgery can provide benefit for some patients with CD when initial surgery does not result in cure. 6,34 To determine the overall utility of surgical management in ESS, the outcome of early repeat surgery was combined with the results of initial surgery. After initial surgery and early repeat surgery in select cases, immediate biochemical remission was achieved in 88% of patients with ESS and 96% of controls (p = 0.10). Survival analyses were performed based on this overall surgical outcome and demonstrated 5-year recurrence-free survival in 85% of patients with ESS and 92% of controls (p = 0.10; Fig. 4). Overall, although there were trends toward differences in outcome between patients with ESS and controls in our cohort, these differences were not statistically significant. This finding may result from a lack of any true difference in biochemical outcome between the groups or may be attributable to Type II error of an insufficiently powered study. Among patients with ESS, the presence of complete ESS predicted a worse long-term outcome (p = 0.04), although surgical remission was observed in 5 (56%) of 9 patients with complete ESS at the last follow-up (Table 5). Although significant, this finding is tempered by the fact that it is based on a small number of events (n = 4). Future multicenter studies may allow for the measurement of a greater number of events, but may not be able to similarly control for surgical experience and technique. Complications after pituitary surgery were uncommon in both patients with ESS and controls. Most frequently, pa- TABLE 4: Causes of persistent hypercortisolism and long-term recurrence Parameter No. Patients w/ ESS Controls no. of patients identified cause no tumor found at surgery 3 3 presumed tumor not ACTH positive 2 2 extensive parasellar disease 4 0 unknown 2 1 total 12 6 Fig. 4. Survival analysis of long-term biochemical outcome after surgical management of CD with and without ESS. Recurrence-free survival for patients with ESS at 2, 5, and 10 years was 88%, 85%, and 71%, respectively, and was 92% for controls at 2, 5, and 10 years (p = 0.10). 524 J Neurosurg / Volume 121 / September 2014

8 Cushing s disease and empty sella syndrome TABLE 5: Univariate analysis of factors affecting immediate and long-term outcome among patients with ESS* Covariate Initial Surgical Outcome Overall Early Surgical Outcome J Neurosurg / Volume 121 / September 2014 Long-Term Outcome no. of patients clinical characteristic age sex imaging characteristic complete ESS gland thickness MRI-negative tumor size cavernous sinus invasion surgical characteristic cavernous sinus invasion cavernous sinus or dural invasion * Bold value indicates significance. Results of initial surgery plus early repeat surgery in select cases. Includes 34 patients with adenomas identified on MRI only. tients experienced transient DI or SIADH. Postoperative CSF leak was the most frequent nonendocrine complication and occurred in 6% of patients with ESS and 3% of patients without ESS. Second-line therapy for CD includes repeat surgery, radiotherapy, and bilateral adrenalectomy. 6 Radiotherapy, including stereotactic radiosurgery, can control hypercortisolism in up to 50% 60% of patients but frequently has a long latency period, averaging more than a year to result in eucortisolism. 6,13,32 Furthermore, radiation often results in hypopituitarism (32% of patients) and can result in optic or cranial nerve dysfunction (4% of patients). 10,32,37 Medical therapy can also be used for select cases of persistent hypercortisolism after surgery. Steroidogenesis inhibitors, such as ketoconazole and metyrapone, can control hypercortisolism in some patients but cannot achieve a biochemical cure. 6 Although pituitary surgery for CD was not as successful and CSF leaks were more common in the patients with ESS as compared with controls, surgery was generally effective and safe. Even in this challenging cohort, surgical outcomes appear to provide significant improvement over medical and radiation-based alternatives. Pituitary surgery in the setting of ESS presents certain difficulties, particularly the presence of an arachnoid pouch and the likelihood of an intraoperative CSF leak. Certain surgical techniques may be helpful to diminish the risk of leak. Identification of the arachnoid pouch before dural opening will prevent some CSF leaks, as the dural opening can be positioned more inferiorly along the sella. The use of intraoperative lumbar CSF drainage has been shown to reduce the incidence of intraoperative CSF leaks with pituitary macroadenomas. 28 This reduction may be attributable to diminished tension on the arachnoid during dissection. As the majority (64%) of intraoperative CSF leaks in patients with CD and ESS were due to direct arachnoid dissection, the use of intraoperative CSF drainage may also reduce the incidence of CSF leaks in this cohort. Another challenge among patients with ESS and CD is the frequency of extra-pituitary parasellar microadenomas (3 [4%] of 78 cases). An empty sella was present in 3 of 6 patients in a series of extra-pituitary parasellar microadenomas in CD (2 patients are also included in the present study). 33 These lesions may be difficult to identify on imaging and at surgery and may underlie persistent hypercortisolism in this cohort. Conclusions Adrenocorticotropin-secreting pituitary adenomas in patients with ESS represent challenging surgical lesions. Although the biology of these tumors in CD patients with and without ESS was the same, the risk of CSF leak was increased in patients with ESS. Among patients with ESS, the presence of complete ESS predicted a worse outcome. Despite these differences, pituitary surgery in the setting of ESS is effective, durable, and safe, such that surgery in this cohort offers significant benefit over existing therapeutic alternatives. Therefore, pituitary surgery should remain the treatment of choice for CD regardless of the presence of an empty sella. Disclosure The authors report no conflict of interest concerning the materials or methods used in this study or the findings specified in this paper. This research was supported by the Intramural Research Program of the National Institute of Neurological Disorders and Stroke at the National Institutes of Health. Author contributions to the study and manuscript preparation include the following. Conception and design: Oldfield, Mehta. Acquisition of data: Mehta, Bakhtian. Analysis and interpretation of data: Mehta, Bakhtian. Drafting the article: Mehta. Critically revising the article: all authors. Reviewed submitted version of manuscript: all authors. Statistical analysis: Mehta. Study supervision: Oldfield. References 1. Abbassioun K, Fatourehchi V, Amirjamshidi A, Meibodi NA: Familial acromegaly with pituitary adenoma. Report of three affected siblings. J Neurosurg 64: , Agrawal NK, Jain P, Garg S: Primary empty sella with isolated ACTH deficiency and microprolactinoma. Gynecol Endocrinol 28: , Arnaldi G, Angeli A, Atkinson AB, Bertagna X, Cavagnini F, Chrousos GP, et al: Diagnosis and complications of Cushing s syndrome: a consensus statement. J Clin Endocrinol Metab 88: , Benbow EW: Pituitary corticotroph adenoma in a primary enlarged empty sella turcica. Histopathology 32: , Bergland RM, Ray BS, Torack RM: Anatomical variations in the pituitary gland and adjacent structures in 225 human autopsy cases. J Neurosurg 28:93 99, Biller BMK, Grossman AB, Stewart PM, Melmed S, Bertagna X, Bertherat J, et al: Treatment of adrenocorticotropin-dependent Cushing s syndrome: a consensus statement. J Clin Endocrinol Metab 93: , Bjerre P, Lindholm J, Videbaek H: The spontaneous course of pituitary adenomas and occurrence of an empty sella in un- 525

9 G. U. Mehta, K. D. Bakhtian, and E. H. Oldfield treated acromegaly. J Clin Endocrinol Metab 63: , Breslow NE, Day NE: Statistical Methods in Cancer Research. Volume 1 The Analysis of Case-Control Studies. Lyon: IARC Press, 1980, pp Busch W: [Morphology of sella turcica and its relation to the pituitary gland.] Virchows Arch 320: , 1951 (Ger) 10. Cifarelli CP, Schlesinger DJ, Sheehan JP: Cranial nerve dysfunction following Gamma Knife surgery for pituitary adenomas: long-term incidence and risk factors. Clinical article. J Neurosurg 116: , De Marinis L, Bonadonna S, Bianchi A, Maira G, Giustina A: Primary empty sella. J Clin Endocrinol Metab 90: , Domingue JN, Wing SD, Wilson CB: Coexisting pituitary adenomas and partially empty sellas. J Neurosurg 48:23 28, Estrada J, Boronat M, Mielgo M, Magallón R, Millan I, Díez S, et al: The long-term outcome of pituitary irradiation after unsuccessful transsphenoidal surgery in Cushing s disease. N Engl J Med 336: , Foresti M, Guidali A, Susanna P: [Primary empty sella. Incidence in 500 asymptomatic subjects examined with magnetic resonance.] Radiol Med (Torino) 81: , 1991 (Ital) 15. Gharib H, Frey HM, Laws ER Jr, Randall RV, Scheithauer BW: Coexistent primary empty sella syndrome and hyperprolactinemia. Report of 11 cases. Arch Intern Med 143: , Guitelman M, Garcia Basavilbaso N, Vitale M, Chervin A, Katz D, Miragaya K, et al: Primary empty sella (PES): a review of 175 cases. Pituitary 16: , Hekimsoy Z, Yünten N, Sivrioglu S: Coexisting acromegaly and primary empty sella syndrome. Neuroendocrinol Lett 25: , Ho DE, Imai K, King G, Stuart EA: Matching as nonparametric preprocessing for reducing model dependence in parametric causal inference. Polit Anal 15: , Ho KL: Coexistence of primary empty sella and silent corticotrophic adenoma. Mod Pathol 9: , Iwai Y, Yamanaka K, Yoshioka K, Kanai M: Gamma knife radiosurgery for GH-secreting microadenoma with empty sella. J Clin Neurosci 11: , Jordan RM, Kendall JW, Kerber CW: The primary empty sella syndrome: analysis of the clinical characteristics, radiographic features, pituitary function and cerebrospinal fluid adenohypophysial hormone concentrations. Am J Med 62: , Kaufman B, Chamberlin WB Jr: The ubiquitous empty sella turcica. Acta Radiol Diagn (Stockh) 13: , Lindsay JR, Oldfield EH, Stratakis CA, Nieman LK: The postoperative basal cortisol and CRH tests for prediction of longterm remission from Cushing s disease after transsphenoidal surgery. J Clin Endocrinol Metab 96: , Login I, Santen RJ: Empty sella syndrome. Sequela of the spontaneous remission of acromegaly. Arch Intern Med 135: , Lonser RR, Wind JJ, Nieman LK, Weil RJ, DeVroom HL, Oldfield EH: Outcome of surgical treatment of 200 children with Cushing s disease. J Clin Endocrinol Metab 98: , Manavela MP, Goodall CM, Katz SB, Moncet D, Bruno OD: The association of Cushing s disease and primary empty sella turcica. Pituitary 4: , Mancini A, Calabrò F, Lagonigro G, Saporosi A, Colosimo C Jr, Anile C, et al: Cushing s syndrome: new variants and association with empty sella syndrome. J Nucl Med Allied Sci 34:59 66, Mehta GU, Oldfield EH: Prevention of intraoperative cerebrospinal fluid leaks by lumbar cerebrospinal fluid drainage during surgery for pituitary macroadenomas. Clinical article. J Neurosurg 116: , Melmed S, Colao A, Barkan A, Molitch M, Grossman AB, Kleinberg D, et al: Guidelines for acromegaly management: an update. J Clin Endocrinol Metab 94: , Oldfield EH, Doppman JL, Nieman LK, Chrousos GP, Miller DL, Katz DA, et al: Petrosal sinus sampling with and without corticotropin-releasing hormone for the differential diagnosis of Cushing s syndrome. N Engl J Med 325: , Oldfield EH, Vortmeyer AO: Development of a histological pseudocapsule and its use as a surgical capsule in the excision of pituitary tumors. J Neurosurg 104:7 19, Petit JH, Biller BMK, Yock TI, Swearingen B, Coen JJ, Chapman P, et al: Proton stereotactic radiotherapy for persistent adrenocorticotropin-producing adenomas. J Clin Endocrinol Metab 93: , Pluta RM, Nieman L, Doppman JL, Watson JC, Tresser N, Katz DA, et al: Extrapituitary parasellar microadenoma in Cushing s disease. J Clin Endocrinol Metab 84: , Ram Z, Nieman LK, Cutler GB Jr, Chrousos GP, Doppman JL, Oldfield EH: Early repeat surgery for persistent Cushing s disease. J Neurosurg 80:37 45, Thapar K, Kovacs K, Scheithauer BW, Stefaneanu L, Horvath E, Pernicone PJ, et al: Proliferative activity and invasiveness among pituitary adenomas and carcinomas: an analysis using the MIB-1 antibody. Neurosurgery 38:99 107, Wind JJ, Lonser RR, Nieman LK, DeVroom HL, Chang R, Oldfield EH: The lateralization accuracy of inferior petrosal sinus sampling in 501 patients with Cushing s disease. J Clin Endocrinol Metab 98: , Xu Z, Vance ML, Schlesinger D, Sheehan JP: Hypopituitarism after stereotactic radiosurgery for pituitary adenomas. Neurosurgery 72: , 2013 Manuscript submitted September 16, Accepted March 26, Please include this information when citing this paper: published online May 23, 2014; DOI: / JNS Address correspondence to: Edward H. Oldfield, M.D., Department of Neurosurgery, University of Virginia Health Sciences Center, University of Virginia, Box , Charlottesville, Virginia. eho4u@hscmail.mcc.virginia.edu. 526 J Neurosurg / Volume 121 / September 2014