Sven Berkmann, M.D., 1 Sven Schlaffer, M.D., 1 Christopher Nimsky, M.D., Ph.D., 1,2 1

Transcription

1 J Neurosurg 121: , 2014 AANS, 2014 Intraoperative high-field MRI for transsphenoidal reoperations of nonfunctioning pituitary adenoma Clinical article Sven Berkmann, M.D., 1 Sven Schlaffer, M.D., 1 Christopher Nimsky, M.D., Ph.D., 1,2 Rudolf Fahlbusch, M.D., Ph.D., 1,3 and Michael Buchfelder, M.D., Ph.D. 1 1 Department of Neurosurgery, University Hospital Erlangen, Erlangen; 2 Department of Neurosurgery, University of Marburg, Marburg; and 3 International Neuroscience Institute, Hannover, Germany Object. The loss of anatomical landmarks, frequently invasive tumor growth, and tissue changes make transsphenoidal reoperation of nonfunctioning pituitary adenomas (NFAs) challenging. The use of intraoperative MRI (imri) may lead to improved results. The goal of this retrospective study was to evaluate the impact of imri on transsphenoidal reoperations for NFA. Methods. Between September 2002 and July 2012, 109 patients underwent reoperations in which 111 transsphenoidal procedures were performed and are represented in this study. A 1.5-T Magnetom Sonata Maestro Class scanner (Siemens) was used for imri. Follow-up imri scans were acquired if gross-total resection (GTR) was suspected or if no further removal seemed possible. Results. Surgery was performed for tumor persistence and regrowth in 26 (23%) and 85 (77%) patients, respectively. On the initial imri scans, GTR was confirmed in 19 (17%) patients. Remnants were located as follows: 65 in the cavernous sinus (71%), 35 in the suprasellar space (38%), 9 in the retrosellar space (10%). Additional resection was possible in 62 (67%) patients, resulting in a significant volume reduction and increased GTR rate (49%). The GTR rates of invasive tumors on initial imri and postoperative MRI (pomri) were 7% and 25%, respectively. Additional remnant resection was possible in 64% of the patients. Noninvasive tumors were shown to be totally resected on the initial imri in 31% of cases. After additional resection for 69% of the procedures, the GTR rate on pomri was 75%. Transcranial surgery to resect tumor remnants was indicated in 5 (5%), and radiotherapy was performed in 29 (27%) patients. After GTR, no recurrence was detected during a mean follow-up of 2.2 ± 2.1 years. Conclusions. The use of imri in transsphenoidal reoperations for NFA leads to significantly higher GTR rates. It thus prevents additional operations and reduces the number of tumor remnants. The complication rates do not exceed the incidences reported in the literature for primary transsphenoidal surgery. If complete tumor resection is not possible, imri guidance can facilitate tumor volume reduction. ( Key Words pituitary adenoma intraoperative MRI reoperation tumor recurrence transsphenoidal surgery pituitary surgery The goals of surgery for nonfunctioning pituitary adenoma (NFA) are recovery from neurological deficits, including visual field (VF) impairment resulting from compression of the optic chiasm, restoration of pituitary function in the case of hypopituitarism, and long-term tumor control. These goals may be achieved by a transsphenoidal approach with low perioperative morbidity in the majority of cases. Nevertheless, due to the restricted access to the sellar confines, the rate of unsuspected tumor remnants may be as high as 50%. 10 Therefore, various tools have been introduced, starting with Hardy and Wigser s use of radiofluoroscopy in 1965, 32 to increase the efficiency of this Abbreviations used in this paper: DI = diabetes insipidus; FSH = follicle-stimulating hormone; GKS = Gamma Knife surgery; GTR = gross-total resection; imri = intraoperative MRI; LH = luteinizing hormone; NFA = nonfunctioning pituitary adenoma; pomri = postoperative MRI; VA = visual acuity; VF = visual field. procedure. Intraoperative MRI (imri) has been used in pituitary surgery for more than a decade. 45,71 It allows evaluation of progress during transsphenoidal tumor resection, an update of images for intraoperative navigation tools, and the exclusion of imminent complications, such as hemorrhage, before the site is closed. 17 Five decades of transsphenoidal pituitary surgery have left a legacy of thousands of patients with remnants of benign, mostly slow-growing NFAs, which can only be monitored by serial imaging and may eventually cause symptoms. Because of varying operative anatomy, frequently invasive tumor growth, loss of anatomical landmarks, and the presence of scarring, NFA reoperations remain a challenge for pituitary surgeons. The literature has addressed different issues, such as the feasibility and general outcome of imri-guided transsphenoidal surgery, 2,3,5,10,14,25,27,28,31,34,40,45,46,49 51,54,57 59,65,66,71,77 and special issues such as its impact on endocrinological 9 and ophthalmological 10 outcomes or its use for acromega J Neurosurg / Volume 121 / November 2014

2 Intraoperative MRI for transsphenoidal reoperations ly. 29 Nevertheless, the benefit of imri for the growing number of patients in need of repeat surgery remains unclear. The aim of this retrospective study was to evaluate the impact of imri on the extent of transsphenoidal reoperations in patients presenting with recurrent or persisting NFA. J Neurosurg / Volume 121 / November 2014 Methods Patient Population and Data Assessment Between September 2002 and July 2012, 925 patients with NFA were operated at the Department of Neurosurgery, University Hospital of Erlangen in Erlangen, Germany. A total of 207 (22%) procedures were indicated for recurrent or persisting tumors, and 168 (81%) of these operations were performed via a transsphenoidal approach. While repeat transsphenoidal surgery is the primary indication for imri use in our department, it was not used for all of these patients due to limited availability. The following additional criteria were used to select patients who might benefit from repeat surgery using imri: invasive tumor growth pattern (for example, into the cavernous sinus), spacious suprasellar or retrosellar tumors, and tumors with a high probability of incomplete resection and/or persisting remnants that eventually would require a transcranial procedure. All 109 patients with a history of surgery for NFA in whom surgery was assisted by high-field imri guidance (1.5-T magnet field strength; Magnetom Sonata Maestro Class scanner, Siemens) between September 2002 and July 2012 were included in this retrospective study. A total of 111 reoperations were performed in these patients. The mean age was 57 ± 13 years (range years), and in 77 (69%) procedures, the patients were male. A comparison with the remaining 49 patients with recurrent or persisting tumors who underwent surgery without imri guidance during the same time interval was not possible because the predictors for clinical outcome (that is, tumor size, growth pattern, invasiveness) were significantly different due to the indications for imri use. Endocrine dynamic tests, as well as ophthalmological examinations according to a standard protocol were done preoperatively and 7 days and 3 months after surgery. Depending on these results, further follow-up visits were scheduled at least once each year thereafter. The minimal follow-up time for study inclusion was set at 6 months. The follow-up time was defined as the time between the repeat surgical procedure and the last follow-up visit at our department. The dynamic endocrine testing protocol included basal serum measurement of triiodothyronine (T3), free thyroxine (T4), thyroid-stimulating hormone (TSH), prolactin, luteinizing hormone (LH), follicle-stimulating hormone (FSH), growth hormone (GH), insulin-like growth factor-i (IGF-I), and testosterone or estradiol. A short adrenocorticotropic hormone (ACTH) stimulation test with determination of serum cortisol levels was used to assess the pituitary-adrenal axis. The details of the endocrine testing routinely performed at the Department of Neurosurgery, University Hospital of Erlangen, Erlangen, Germany were previously published. 11,28 The ophthalmological examination included testing of visual acuity (VA) and color vision, VF testing with the Goldmann perimeter, and fundus examination. The neuropathological examination included analysis by the pituitary tumor registry of the German Endocrine Society. Surgical complications, such as CSF fistulas or meningitis were documented. Clinical records and digitally recorded postoperative MRI (pomri) scans were reviewed for follow-up information. MRI was performed before surgery, during surgery (imri), and 3 months after surgery. The maximal tumor diameters were digitally measured, and volumetry was performed using the formula proposed by Lundin and Pedersen. 44 Giant adenomas were defined as tumors with a maximum diameter larger than 40 mm. Lesion extension was classified by the modified version of the Hardy grading system 32,69 as follows: Grade 1 (craniocaudal diameter of the lesion < 10 mm, no lesions were Grade 1), Grade 2 (craniocaudal diameter mm and suprasellar extension within 10 mm of the sphenoidal plane), Grade 3 (craniocaudal diameter mm and suprasellar extension of up to 30 mm), and Grade 4 (craniocaudal diameter > 40 mm and extension far beyond the sellar space). Invasion into the cavernous sinus space was described by the Knosp classification. 35 The resulting data were analyzed using statistical software from GraphPad Software, Inc. The results were evaluated using Fisher exact tests, Mann-Whitney U- tests, and paired t-tests. Results with p values < 0.05 were regarded as statistically significant. The use of imri and the assessment of the resulting data were approved by the ethical committee of the University Hospital of Erlangen, Germany. Signed informed consent was obtained from each patient. imri and Surgical Technique The characteristics of the operating room and the operative technique were previously published. 11,53 The first imri sequences were acquired after positioning of the patient on the operating table. Thereafter, timing of imri controls was decided by the neurosurgeon in case that gross-total resection (GTR) was supposed or if no further tumor removal was possible, but incomplete resection was suspected. If imri revealed an accessible remnant, the surgery was continued. Before the site was closed, a final imri scan was acquired as baseline for future followup imaging. A GTR was defined as no tumor visible on pomri at 3 months. All the patients were operated on by or in the presence of the 2 senior authors (M.B. and R.F.). The transsphenoidal procedure was accomplished as a transnasal, either paraseptal-submucosal, or in the case of considerable scaring a direct transsphenoidal microscopic approach with the option of endoscopic assistance. Results Patient Characteristics and Preoperative Assessment An overview of the characteristics of the 109 patients included in this study is shown in Table 1. The patients underwent a total of 111 transsphenoidal reoperations. On admission, tumor persistence was present in 26 (23%) pa- 1167

3 S. Berkmann et al. tients at a mean of 0.9 ± 0.6 years after surgery (range years). In the remaining 85 (77%) patients, tumor regrowth was detected during a mean follow-up time of 9.9 ± 7.5 years (range ). Thirty-three (30%) patients were initially operated on at our department; the remaining patients were referred to us by other institutions. Seventy-two (66%) patients underwent a single surgery (transsphenoidal approach, n = 66 [92%]; transcranial approach, n = 6 [8%]). Twenty-nine (27%) patients had 2 procedures, and 4 (4%) patients required 3 or more tumor resections. Combinations of transsphenoidal and transcranial approaches were used for 26 (79%) patients who underwent more than one surgery. Four (4%) patients had received adjuvant radiotherapy. Although dopamine agonists were used in 1 patient, he still experienced rapid tumor progress. Histopathological analyses demonstrated the following diagnoses for the 111 lesions: silent gonadotropic adenoma (n = 68 [61%]; immunohistochemical staining positive for LH and FSH, n = 40; positive for LH, n = 14; positive for FSH, n = 14); null cell adenoma (n = 31 [28%]); silent corticotrope adenoma (n = 8 [7%]); and single cases of atypical silent prolactinoma, atypical null cell adenoma, oncocytic adenoma, silent prolactin- and thyrotropin-secreting adenoma, and silent thyrotropic adenoma. Silent corticotrope adenomas recurred significantly earlier (4.5 ± 1.0 years, p = 0.009) than the other pathological entities. Before reoperation, hypopituitarism was detected in 67 (60%) of the 111 procedures. The prevalence rates of pituitary axis insufficiency were as follows: corticotroph axis, 39% (n = 43); thyrotroph axis, 40% (n = 44 ); gonadotroph axis, 39% (n = 43); and somatotroph axis, 29% (n = 32). Panhypopituitarism was detected in 19 (17%) patients. Permanent diabetes insipidus (DI) was seen in 4 (4%) patients. On ophthalmological examination, 36 (33%) patients exhibited VF deficits. Thirty-one (86%) of these patients suffered from bitemporal hemianopsia, and bitemporal quadrantanopsia was detected in 5 (14%) cases; VA was impaired due to optic nerve compression in 39 (36) patients. An overview of tumor dimensions is given in Table 1. The mean tumor diameters were as follows: lateral, 2.4 ± 0.9 cm (range cm); anterioposterior, 2.2 ± 0.8 cm (range cm); and apicobasal, 2.6 ± 1.0 cm (range ). The mean tumor volume was 9.6 ± 9.2 cm 3 (range cm 3 ). In 91 (82%) procedures, the patients suffered from macroadenomas (mean volume 6.3 ± 5.0 cm 3 ). The remaining 20 (18%) cases presented with giant adenomas (mean volume 24.8 ± 8.5 cm 3 ). For 69 procedures (62%), patients had tumors larger than 5 cm 3. According to the modified Hardy grading system 32,69 the following distribution of craniocaudal extensions was seen: Grade 1, n = 3 (3%); Grade 2, n = 48 (43%); Grade 3, n = 48 (43%); and Grade 4, n = 12 (11%). The mean Hardy grade was 2.6 ± 0.7. Invasive tumor growth was seen in 59 (53%) procedures on preoperative MRI (cavernous sinus, n = 54 [49%]; clivus, n = 8 [7%]). The mean preoperative volume of noninvasive tumors (6.8 ± 6.5 cm 3 ) did significantly differ from the mean volume of invasive tumors (12.1 ± 10.5 cm 3 ; p = 0.002). According to the Knosp grading system, 35 TABLE 1: Patient characteristics and preoperative workup results* Variable Value (% or range) no. of patients 109 mean age, yrs (range) 57 ± 13 (23 80) sex male 77 (69%) female 34 (31%) procedures 111 due to tumor persistence 26 (23%) due to tumor recurrence 85 (77%) mean latency until recurrence, yrs (range ± SD) 9.9 ± 7.5 ( ) histology null cell 31 (28%) gonadotroph 68 (61%) silent corticotroph 8 (7%) others 5 (4%) radiological workup tumor size classification microadenoma (diameter <10 mm) 0 macroadenoma (diameter >10 mm) 91 (82%) giant adenoma (diameter >40 mm) 20 (18%) mean modified Hardy s grade 2.6 ± 0.7 mean tumor dimensions anteroposterior diameter (mm) 22 ± 8 (8 54) lateral diameter (mm) 24 ± 9 (10 56) apicobasal diameter (mm) 26 ± 10 (9 55) vol (cm³) 9.6 ± 9.2 ( ) invasive tumor growth 59 (53%) into the cavernous sinus 54 (49%) into the retrosellar space 8 (7%) surgical procedures surgery 72 (66%) transsphenoidal 66 (92%) transcranial 6 (8%) 2 surgeries 29 (27%) 3 surgeries 9 (8%) transcranial and transsphenoidal approach 26 (67%) nonsurgical therapies radiotherapy 4 (4%) dopamine agonist 1 (1%) preop endocrinological state hypopituitarism 67 (60%) gonadotroph axis 43 (39%) somatotroph axis 32 (29%) corticotroph axis 43 (39%) thyrotroph axis 44 (40%) preop ophthalmological state VF impairment 36 (32%) bitemporal quadrantanopsia 6 (5%) bitemporal hemianopsia 31 (28%) diminished VA 39 (35%) * Values are number of procedures (% or range) unless otherwise noted; means are presented ± SD J Neurosurg / Volume 121 / November 2014

4 Intraoperative MRI for transsphenoidal reoperations the distribution of cavernous sinus invasion was as follows: Grade 1, n = 4 (7%); Grade 2, n = 17 (31%); Grade 3, n = 22 (41%); and Grade 4, n = 11 (20%). J Neurosurg / Volume 121 / November 2014 TABLE 2: Surgical results* Variable Value (% or range) GTR, all tumors 111 on 1st control imri 19 (17%) additional resection possible 62 (67%) on pomri 54 (49%) GTR rate % change caused by imri 32% (p < ) imri results mean remnant vol on 1st control imri (cm³) 1.6 ± 2.7 ( ) vol reduction on 1st control imri (%) 88% (22 100%) mean remnant vol on last imri control (cm³) 0.8 ± 1.7 ( ) mean vol reduced by imri-guidance (cm³) 1.3 ± 2.0 ( ; p < ) % change of tumor remnant vol by imri (%) 50% (0 100) pomri results mean residual tumor volume (cm³) 0.6 ± 1.3 (0 9.0) mean difference of remnants vs imri (cm³) 0.3 ± 0.3 ( ) GTR, noninvasive tumors on 1st imri control 16 (31%) additional resection possible 25 (69%) on pomri 39 (75%) GTR rate % change caused by imri 44% (p < ) GTR, invasive tumors on 1st imri control 4 (7%) additional resection possible 35 (64%) on pomri 15 (25%) GTR rate % change caused by imri 18% (p = 0.005) * Values are number of procedures (% or range) unless otherwise noted; means are presented ± SD. Intraoperative Findings and Surgical Results An overview of the surgical results is shown in Table 2. Including the scan acquired before resection, 2 scans were obtained in 51 (46%) procedures, 3 scans in 56 (50%), 4 scans in 3 (3%), and 5 scans in one operation. The mean duration of follow-up scanning, including the draping, was 16 minutes (range 8 22 minutes). Whereas the length of transsphenoidal reoperations (time from incision to closure) without imri between 2002 and 2011 was 63.0 ± 18.7 minutes (range minutes), the use of imri increased the duration to ± 32.2 minutes (range minutes) per procedure. This length does not include preoperative imaging, segmentation of the tumor and/or internal carotid arteries, or neuronavigation system registration, but it never exceeded an additional total preparation time of 30 minutes. During all operations exceeding a length of 120 minutes, at least a second or even third intraoperative imaging procedure was performed. Gross-total resection was detected on the first imri control scan in 19 (17%) of the patients. The residual tumor volume in the remaining cases was 1.6 ± 2.7 cm 3 (range cm 3 ). The remnant volume exceeded 5 cm 3 in 9 (10%) tumors. The location of tumor remnants in these 92 patients was as follows: cavernous sinus/middle cerebral fossa, n = 65 (71%); suprasellar space, n = 35 (38%); sella (view obstructed by sellar diaphragm bulging), n = 8 (9%); and retrosellar confines/intraclival, n = 9 (10%). Further resection based on the updated imri sequences was possible in 62 (67%) patients. While intraoperative navigation was often crucial for tailoring the approach for optimal sellar exposure, the data sets were only updated in 7 (11%) procedures after the imri control scans to guide further tumor resection. In the 30 (33%) patients in whom no further tumor resection was possible, remnants were located in the cavernous sinus in 23 (77%) cases. The mean tumor volume of remnants on final imri scans was 0.8 ± 1.7 cm 3 (range cm 3 ). The volume reduction achieved with imri guidance (1.3 ± 2.0 cm 3, range cm 3 ; 50% ± 40% of the residual tumor volume on the first control imri scan) was significant (p < ). A GTR was possible in 2 patients with only intrasellar remnants. Six patients had intrasellar remnants that invaded the cavernous sinus, and GTR was only possible in 3 (50%) of these cases. Fifty-seven of all suprasellar remnants seen on the first imri control scan were totally resected, but we could only remove 29% and 22% of the lateral and retrosellar remnants, respectively. In cases of noninvasive tumor, GTR was confirmed by the first control imri scan in 16 (31%) patients. The mean volume of noninvasive tumors that initially could not be totally resected (8.7 ± 6.9 cm 3 ) was significantly different from that of those were resectable (2.6 ± 2.1 cm 3 ; p < ). In the 111 procedures assessed in this study, the remnants on the first control imri scan were located as follows: lateral sellar wall, adjacent to the cavernous sinus, n = 22 (61%); in the suprasellar space, n = 15 (42%); in the sella and the descended sellar diaphragm, n = 2 (6%); posterior to the sella, n = 2 (6%). Additional imri-guided tumor resections in 25 (69%) patients led to the significantly higher final GTR rate of 75% in this subgroup (p < ). It was evident that invasive tumors were less likely to be totally resected (p < ; relative risk 2.8; 95% CI ). Initial GTR of invasive tumors was possible in 4 (7%) patients (lateral invasion, n = 3; invasion into the clivus, n = 1). Compared with noninvasive tumors, GTR without further imri guidance was significantly less likely (p = 0.001; OR 3.0, 95% CI ). The remnants on the first control imri scan were located as follows: cavernous sinus, n = 43 (78%); suprasellar space, n = 21 (38%); clivus, n = 7 (13%); and in both the sella and bulging sellar diaphragm, n = 6 (11%). An additional resection was possible in 35 (64%) of these patients. The final GTR rate of this subgroup on pomri was 25%, which was significantly higher (p = 0.005), than on the first imri scan. With respect to cavernous sinus invasion, GTR was possible in all 4 patients with Knosp Grade 1, in 10 (59%) patients with Knosp Grade 2, and 0 patients with Knosp Grade 3 or 4. In 10 (9%) patients, endoscopic assistance was useful for further tumor resection. Before the first control imri scan in this subgroup, GTR was assumed in 4 (40%) patients, partial resection was expected in 4 (40%) patients, and the status was not clear in the remaining 2 patients. 1169

5 S. Berkmann et al. All of the patients showed tumor remnants on imri, and additional imri-guided resection was possible in 8 (80%) patients. The localization on the imri scans of the remnants not seen by endoscopic view was as follows: lateral in the cavernous sinus, n = 6 (60%); suprasellar, n = 5 (50%); sellar, n = 1 (10%). Postoperative Findings and Follow-Up An overview of the surgical results is given in Table 3. The 3-month pomri data were available for all patients. A GTR was seen in 54 (49%) patients. The difference between the GTR rate for the first imri control scan and the pomri sequences was significant (p < , OR 4.6, 95% CI ). Nonresectable tumor remnants found on the last control imri before closure showed spontaneous volume decreases after 31 (37%) procedures. The mean volume of remnants on pomri was 1.1 ± 1.7 cm 3 (range cm 3 ). Only 1 (2%) of these patients suffered from a remnant exceeding 5 cm 3. The locations of tumor remnants on pomri in 57 patients were as follows: cavernous sinus/middle cerebral fossa, n = 46 (85%); suprasellar space, n = 15 (28%); retrosellar confines/intraclival, n = 7 (12%); intrasellar space with bulging of the sellar diaphragm, n = 4 (7%). Smaller tumor volume (< 10 cm 3 ) and histological diagnosis did not correlate with a higher GTR rate or the extent of tumor volume reduction. The GTR rates did not differ significantly between recurring (53%) and persisting (35%) tumors. On ophthalmological follow-up within 1 month after surgery, overall VF deficit improvement was seen in 29 (81%) patients. The VFs in 22 (61%) patients were already normal at that time point. In 5 (14%) patients, VF deficits persisted; however, suprasellar remnants with persistent optic nerve compression were only seen on pomri scans in 2 of these cases, and the remnants were further resected via transcranial procedures. No patients experienced VF worsening. Normal or improved VA after surgery was documented in 31 (79%) cases. In 15 (41%) patients, VA had normalized, while there were significant improvements in 16 (41%) patients. VA deficits persisted in 5 (13%) patients, and 3 (8%) patients experienced postoperative VA decreases. These patients had tumors with extensive growth into the suprasellar space (median craniocaudal diameter 37 mm), and while GTR was achieved in 1 patient, the tumors were further resected by transcranial approaches in the other 2 cases. Endocrinological testing after a mean follow-up time of 2.2 ± 2.1 years (range years) showed hypopituitarism in 66 (59%) cases. The prevalence rates of deficiencies in the different axes were as follows: corticotroph axis, 42 (38%); thyrotroph axis, 46 (41%); gonadotroph axis, 49 (44%); and somatotroph axis, 33 (30%). The mere prevalence of hypopituitarism did not change significantly after surgery; however, 10 (15%) patients showed recovery of hormonal axes, while 11 (12%) patients suffered from postoperative loss of pituitary axes. Additional imri-guided resections or GTRs were not associated with postoperative recovery or loss of pituitary axes. In noninvasive tumors, recovery and loss of axes were seen in 5 (16%) and 6 (13%) patients, respectively. Recovery and loss of axes in invasive tumors were seen TABLE 3: Follow-up and clinical outcome* Variable Value (% or range) mean follow-up time, yrs (range) 2.2 ± 2.1 ( ) postop endocrinological state hypopituitarism 66 (59%) gonadotroph axis 49 (44%) somatotroph axis 33 (30%) corticotroph axis 42 (38%) thyrotroph axis 46 (41%) patients w/ recovery of pituitary axes 10 (15%) new onset of axis insufficiency 11 (12%) DI 5 (5%) recovery of preop DI 2 (50%) new permanent DI 3 (3%) postop ophthalmological state postop VF deficits normalization 22 (61%) improvement 7 (20%) unchanged 5 (14%) worsening 0 (0%) postop VA deficits normalization 15 (38%) improvement 16 (41%) unchanged 5 (13%) worsening 3 (8%) postop complications recurrent rhinoliquorrhea 5 (4%) meningitis 2 (2%) hemorrhage into tumor remnant 1 (1%) further therapies revision surgery due to tumor 7 (6%) transsphenoidal revision due to recurrence 2 (2%) transcranial revision due to persistence 5 (5%) radiotherapy 29 (27%) * Values are number of procedures (% or range) unless otherwise noted; means are given ± SD. in 11 (31%) and 10 (29%) patients, respectively. Invasive growth was a risk for new onset of hypopituitarism after surgery (p = 0.04). New onset of permanent DI was diagnosed in 3 (3%) patients. Two (50%) patients with preoperative permanent DI recovered postoperatively. Although thorough closures of the sellar floor and, in the case of intraoperative CSF loss, prolonged intrathecal pressure decrease by lumbar drains were standard in all procedures, postoperative rhinoliquorrhea persisted in 5 (4%) patients. Two of these patients developed postoperative meningitis (overall incidence 2%). One patient suffered from postoperative hemorrhage into a suprasellar remnant 1 day after surgery and required revision surgery with a transcranial approach. No patient had an injury to the internal carotid artery, and there was no surgery-related mortality J Neurosurg / Volume 121 / November 2014

6 Intraoperative MRI for transsphenoidal reoperations Seven (6%) patients needed an additional tumor resection during the mean follow-up time of 2.2 ± 2.1 years (range years). Transcranial reoperations were indicated in 5 (5%) cases due to symptomatic persisting remnants. Two (2%) patients with initially small parasellar remnants of a null cell adenoma and a gonadotropic adenoma underwent reoperations after 5.6 and 6.4 years by imri-guided transsphenoidal surgery because of recurrent intra- and suprasellar growth, respectively. In 29 (27%) patients, radiotherapy was indicated because of tumor remnants. None of the 54 patients with GTR on pomri scans suffered from tumor recurrence during a mean follow-up of 1.9 ± 1.7 years ( years). Illustrative Case A 57-year-old male patient was referred to the Department of Neurosurgery, University Hospital of Erlangen. He had undergone transsphenoidal NFA resection 6 years earlier but experienced new VF deficits. Histological analysis after subtotal tumor resection showed a pituitary adenoma with positive staining for LH and FSH without any criteria for atypia or hormonal activity. Besides persistent hypogonadotropic hypogonadism, the immediate postoperative follow-up was uneventful. Upon presentation to our department, MRI showed a sellar mass with supra- and parasellar extension (modified Hardy Grade 3; Knosp Grade 2; mm; volume 11.8 cm 3 ; Fig. 1A). A transsphenoidal reoperation guided by imri and intraoperative navigation was indicated. The tumor was resected until the descent of the sellar diaphragm blocked the view to the right part of the sella. The control imri scan showed an intra- and suprasellar remnant adjacent to the cavernous sinus on the right side and a nonresectable small remnant adjacent to the internal carotid artery on the left side (Fig. 1B). Further resection of the remnant on the right side was possible. On the final imri scan, a small remnant adjacent to the pituitary gland was suspected (Fig. 1C), but this was smaller on the 3-month pomri study (Fig. 1D). The histological characteristics of the tumor did not differ from those of the initial specimen removed 6 years earlier. The ophthalmological follow-up after resection confirmed VF widening, but his hypopituitarism persisted. There was no evidence of tumor regrowth at the follow-up visit 3.5 years after repeat surgery. Discussion This study demonstrates the impact of imri findings on the resection rates of imri-guided reoperations for NFA, as well as the safety of repeat transsphenoidal surgery with imri. While transsphenoidal surgery is the primary treatment option in most patients with NFA, the surgical outcomes may significantly depend on tumor size, 55 the presence of invasive growth into adjacent structures, 16 histological characteristics, 15,20,42,55,61 and the experience of the surgeon. 21 As virtually all symptomatic NFA are macroadenomas with frequent tumor growth exceeding the narrow confines of the sella, a considerable number of J Neurosurg / Volume 121 / November 2014 Fig. 1. Illustrative pre-, intra-, and postoperative MR images obtained in a 57-year-old patient with recurrent NFA. A: Coronal preoperative 1.5-T T1-weighted contrast-enhanced MR image showing a sellar mass with supra- and parasellar extension. B: Coronal intraoperative 1.5-T T2-weighted MR image acquired as a control after no further resection seemed possible. The image shows intra-, para-, and suprasellar tumor remnants (arrowheads) adjacent to the cavernous sinus dislocating the pituitary gland. C: Coronal intraoperative 1.5-T T2-weighted MR image acquired after further resection of the intra- and suprasellar remnants showing bilateral small tumor remnants (arrowheads) adjacent to the internal carotid artery, which could not be further resected. D: Coronal postoperative 1.5-T T1-weighted contrast-enhanced MR image showing the small parasellar remnants (arrowheads) correlating with the results of the last imri scan. these tumors may not be completely resectable. Roelfsema et al. 62 reported a remission rate of 44% in their meta-analysis of 5022 patients with NFA. The recurrence rate was patients/year during a follow-up of 5 ± 0.2 years; however, a considerable percentage of these tumors may recur after longer time lapses. Notably, the mean latency from surgery until reoperation because of recurrence in the present study was 9.9 ± 7.5 years (range 2 45 years). Treatment options for recurrent or asymptomatic NFA include observation, radiotherapy, stereotactic radiosurgery, and transsphenoidal reoperation or craniotomy. Transsphenoidal reoperations for recurrent or persisting tumors may be more difficult to perform than the initial surgery. Obliteration of surgical landmarks, scar tissue formation, and the effects of preoperative radiotherapy may exacerbate the challenges. In a more recent study that included a total of 96 consecutive patients with hormonally active tumors as well as NFA, the authors reported a GTR rate of 26% after transsphenoidal reoperations. 6 Visual deficit improvement was noted in 56% of these patients. Major complications (a major epistaxis) occurred in 1% of cases. Minor complications (such as DI, sinusitis, fat graft harvest site infection) occurred in 30% of the procedures. New onset of hypopituitarism was the most frequent incident. Eight percent of the patients 1171

7 S. Berkmann et al. had to undergo reoperation, and 9% required postoperative radiotherapy. In the present study, imri was performed if GTR was suspected or if no further tumor removal seemed safely possible. The initial GTR rate on these images was 17%, which correlates well with the rates reported in the literature. 6 Further resection based on the updated imri sequences was possible in 67% of the patients, resulting in a final GTR rate of 49%. In patients with noninvasive tumors, the initial GTR rate was 31%. The major finding on control imri scans was a tumor remnant adjacent to the cavernous sinus. Even in consideration of scarring and anatomical landmark loss, further tumor resection toward the lateral border of the sella was possible due to imri visualization in the majority of these patients. This resulted in a final GTR rate of 75% for noninvasive tumors. For tumors with cavernous sinus invasion, all Knosp Grade 1 lesions and 59% of Grade 2 tumors could be totally resected. The finding that imri-guided surgery led to significantly higher resection rates corresponds with previous reports in the literature regarding transsphenoidal procedures for NFA. 9,14,17,54 The GTR rate of 49% in the present study is comparable to the rates described for first-time transsphenoidal surgery of NFA without imri in large meta-analyses. 62 Nevertheless, because all of these procedures were reoperations, the rate is low compared with data from imri studies, which included both firsttime procedures and reoperations, and reported GTR rates of about 80% with a 30% increase in GTR due to imri. 8,54 One of the drawbacks of transsphenoidal surgery using imri is the substantial increase in operating time. Half of the patients in the present study only had 1 intraoperative control scan, the majority of the remaining procedures included 2 intraoperative scans, and only a few patients had more than 2. The mean duration of imri scans, including the draping, was about 15 minutes. Nevertheless, the operating time was also prolonged by the consecutive additional resections, which obviously would not have been possible without the use of imri. In a recent study focusing on the utility of imri for transsphenoidal surgery despite longer operating time, the authors concluded that even though the surgical time was longer, it did not increase the complication rate. Moreover, the authors recommended the routine use of high-field imri as they detected a significantly higher rate of GTR and a distinctly higher progression-free survival time. 22 In this context, other techniques for tumor remnant visualization must be mentioned. Several reports of endoscopically assisted procedures as an adjunct to lowand high-field imri-guidance for resection control have been published. 48,66,73 In the present study, only a small subgroup (9%) underwent further tumor resection based on the endoscopic view; however, because of its possible benefit, the endoscope was always available during the procedures. A drawback of using the endoscope without imri for transsphenoidal reoperations may be that GTR was assumed in 40% of these patients based on the endoscopic assessment, and all of them showed tumor remnants on imri. Additional imri-guided resection was possible in 80% of these cases. These false-negative findings and the additional increase of the GTR rates made possible by imri is consistent with the literature. 48,66,73 Theodosopoulos et al. 73 retrospectively analyzed the incidence of unexpected remnants detected by high-field imri after fully endoscopic transsphenoidal surgery. Their rate of 15% in 27 consecutive patients was small compared with published rates of residual tumor following microscopic resections, which led the authors to conclude that intrasellar endoscopy could replace imri with only a modest sacrifice in the extent of tumor resection. A randomized prospective or retrospective case-control trial would be helpful to clarify this issue. Unfortunately the authors did not indicate how many procedures in these 27 patients were reoperations. The varying anatomy and the presence of altered tissue and scarring in reoperations may have influenced their 100% rate of wrongly declared GTR based on the endoscopic view. 73 As 3D image-guided navigation systems are increasingly used in transsphenoidal surgery, others have already demonstrated how these techniques may reduce the morbidity of repeat transsphenoidal procedures, especially in cases of massive scarring and loss of normal anatomical structures after the initial surgery. 26,33,74 In our present experience, while it was often crucial for tailoring the approach to optimally expose the sella, the navigation could not reliably estimate the extent of resection. Intraoperative navigation data sets may be updated by imri sequences to guide further tumor resection; however, an update of the navigation was only used in a minority of our procedures with additional resection due to the optimal visualization of residual tumor with high-field imri. It was suggested that complications may occur more often after reoperations and that recovery rates of ophthalmological and endocrinological symptoms may be lower than those for primary procedures. 37,38,72 There was no surgery-related mortality in the present study, and none of the patients suffered from injury to the internal carotid artery. One patient experienced postoperative hemorrhage into the remaining tumor. According to the literature, the incidence of this major complication is about 0.5%. 12,21 The rate of postoperative meningitis and CSF fistula formation in the present study were well within the reported incidences. 12,21 Improvement of VF deficits was seen in 81% of the patients in the present study. No patients experienced VF worsening. Increased and decreased VA was postoperatively documented in 79% and 8%, respectively. The patients with worsened VA suffered from large tumors with extensive suprasellar growth, and the majority recovered VA after transcranial resection of remnants. The impact of imri guidance on the prognosis of ophthalmological deficits in patients who underwent transsphenoidal surgery has been previously reported. 10 The finding of a decompressed optic chiasm on imri scans is a significant factor for predicting early recovery. Recovery rates of ophthalmological symptoms of imri-guided surgery in mixed populations range from 66% to 100%. 5,10,14,28,31,54,66,77 Decreased VF was reported in a total of 3 patients out of 296 who underwent imri-guided transsphenoidal surgery. 14,52,58 For VA decrease, a complication rate of 3% was reported for 101 imri-guided procedures in a mixed group of patients with sellar tumors. 63 In studies without imri use, the incidence 1172 J Neurosurg / Volume 121 / November 2014

8 Intraoperative MRI for transsphenoidal reoperations J Neurosurg / Volume 121 / November 2014 of postoperative ophthalmological symptoms worsening was reported to range from 0% to 4%. 7,21,39,47 Nevertheless, due to patient and tumor characteristic variations, a direct comparison is prone to be biased. In the present study, endocrinological follow-up testing showed postoperative loss and recovery of hormonal axes in 12% and 15% of patients, respectively. Additional imri-guided resection or GTR was not associated with improved postoperative recovery or loss of pituitary axes. The rate of postoperative hypopituitarism in series on transsphenoidal tumor resections is commonly below 20%; 1,4,18,21,24,30,56 however, higher rates may be seen in the case of larger tumors or NFA. 23,30,56,76 GTR was reported to be a significant factor for postoperative recovery from hypopituitarism. 76 It may be argued, that the comparably low recovery rate in the present study was influenced by the relatively low GTR rate. This is in accordance with the results of a recent controlled cohort study assessing the impact of intraoperative imaging on endocrinological outcomes in imri-guided transsphenoidal surgery for NFA. 9 The study group concluded that the higher GTR rate for imri does not provoke a higher incidence of postoperative hypopituitarism or a lower postoperative recovery rate of pituitary axes. Because of the difficulties associated with transsphenoidal reoperations discussed above, radiotherapy may be an option in patients with residual or recurrent pituitary adenoma. Long-term tumor control rates of 90% have been reported for fractionated radiotherapy, 13,78 as well as for stereotactic radiosurgery such as Gamma Knife surgery (GKS). 43,60,67 The peak of tumor size reduction is seen about 3 years after radiotherapy, and volume reductions of 50% have been described within 5 years. 36,64 It must be mentioned that delayed hypopituitarism rates up to 100% have been reported for fractionated radiotherapy; 41 however, lower rates have been described following GKS. 43,70 Additionally, GKS may be an option for local control in patients with recurrent pituitary adenomas who have already undergone radiotherapy. 75 Tumor volume is an important issue for indicating GKS. 19 In a recent study 70 of 140 patients with NFA undergoing GKS, the only factor influencing the incidence of tumor regrowth or neurological deficits after radiosurgery was a tumor volume greater than 5 cm 3. In concordance with these results, Sheehan et al. 68 performed a retrospective multicenter study including 512 patients and reported that smaller tumor volume and a lack of suprasellar extension were associated with a significant improvement of local tumor control during a median follow-up of 36 months. In the present study, two-thirds of the patients had tumors exceeding 5 cm 3, and imri still showed remnants exceeding 5 cm 3 in 10% of patients after the initial resection. Additional imriguided volume reduction achieved remnant volumes below 5 cm 3 in all but 1 patient. Therefore, patients with accessible tumor masses, especially those with mass lesion symptoms, such as deteriorating vision due to chiasmal compression, may be candidates for repeat surgery, as the use of imri leads to a significant tumor volume reduction. Even if postoperative radiotherapy may be reasonable because of residual tumor masses, as it was in 27% of the patients in the present series, it is possible that remnant volume reduction by imri-guided surgery enhances the efficiency and reduces the morbidity of radiotherapy. Although the present study represents the largest series of imri-guided reoperations for NFA, there are weaknesses that should be noted. This was a single-center analysis and therefore reflects our institution s treatment biases and referral patterns. As with previous studies assessing the use of imri in transsphenoidal surgery, one may argue that the possibility of visualizing the tumor with imri may have weakened the effort to probe for occult tumor remnants. Furthermore, this was a retrospective analysis, which is inherently subject to bias. Finally, most NFAs recur within 5 years after surgery, 62 so our comparably short mean follow-up of 2.2 ± 2.1 years may have incompletely assessed tumor control. Conclusions The use of imri in transsphenoidal reoperations for NFA leads to significantly higher GTR rates, which prevents additional operations and reduces the number of unexpected remnants. The complication rates do not exceed those reported in the literature for primary transsphenoidal surgery. If complete tumor resection is not possible, imri guidance facilitates tumor volume reduction. Disclosure The authors report no conflict of interest concerning the materials or methods used in this study or the findings specified in this paper. Author contributions to the study and manuscript preparation include the following. Conception and design: Berkmann. Acquisition of data: all authors. Analysis and interpretation of data: Berkmann. Drafting the article: Berkmann. Critically revising the article: all authors. Reviewed submitted version of manuscript: all authors. Approved the final version of the manuscript on behalf of all authors: Berkmann. Statistical analysis: Berkmann. Study supervision: Buchfelder. References 1. Abosch A, Tyrrell JB, Lamborn KR, Hannegan LT, Applebury CB, Wilson CB: Transsphenoidal microsurgery for growth hormone-secreting pituitary adenomas: initial outcome and longterm results. J Clin Endocrinol Metab 83: , Ahn JY, Jung JY, Kim J, Lee KS, Kim SH: How to overcome the limitations to determine the resection margin of pituitary tumours with low-field intra-operative MRI during transsphenoidal surgery: usefulness of Gadolinium-soaked cotton pledgets. Acta Neurochir (Wien) 150: , Albayrak B, Samdani AF, Black PM: Intra-operative magnetic resonance imaging in neurosurgery. Acta Neurochir (Wien) 146: , Barker FG II, Klibanski A, Swearingen B: Transsphenoidal surgery for pituitary tumors in the United States, : mortality, morbidity, and the effects of hospital and surgeon volume. J Clin Endocrinol Metab 88: , Baumann F, Schmid C, Bernays RL: Intraoperative magnetic resonance imaging-guided transsphenoidal surgery for giant pituitary adenomas. Neurosurg Rev 33:83 90, Benveniste RJ, King WA, Walsh J, Lee JS, Delman BN, Post KD: Repeated transsphenoidal surgery to treat recurrent or residual pituitary adenoma. J Neurosurg 102: , Berg C, Meinel T, Lahner H, Mann K, Petersenn S: Recovery of pituitary function in the late-postoperative phase after pituitary surgery: results of dynamic testing in patients with 1173

9 S. Berkmann et al. pituitary disease by insulin tolerance test 3 and 12 months after surgery. Eur J Endocrinol 162: , Berkmann S, Fandino J, Müller B, Kothbauer KF, Henzen C, Landolt H: Pituitary surgery: experience from a large network in Central Switzerland. Swiss Med Wkly 142:w13680, Berkmann S, Fandino J, Müller B, Remonda L, Landolt H: Intraoperative MRI and endocrinological outcome of transsphenoidal surgery for non-functioning pituitary adenoma. Acta Neurochir (Wien) 154: , Berkmann S, Fandino J, Zosso S, Killer HE, Remonda L, Landolt H: Intraoperative magnetic resonance imaging and early prognosis for vision after transsphenoidal surgery for sellar lesions. Clinical article. J Neurosurg 115: , Berkmann S, Schlaffer S, Buchfelder M: Tumor shrinkage after transsphenoidal surgery for nonfunctioning pituitary adenoma. Clinical article. J Neurosurg 119: , Black PM, Zervas NT, Candia GL: Incidence and management of complications of transsphenoidal operation for pituitary adenomas. Neurosurgery 20: , Boelaert K, Gittoes NJ: Radiotherapy for non-functioning pituitary adenomas. Eur J Endocrinol 144: , Bohinski RJ, Warnick RE, Gaskill-Shipley MF, Zuccarello M, van Loveren HR, Kormos DW, et al: Intraoperative magnetic resonance imaging to determine the extent of resection of pituitary macroadenomas during transsphenoidal microsurgery. Neurosurgery 49: , Bradley KJ, Wass JA, Turner HE: Non-functioning pituitary adenomas with positive immunoreactivity for ACTH behave more aggressively than ACTH immunonegative tumours but do not recur more frequently. Clin Endocrinol (Oxf) 58:59 64, Brochier S, Galland F, Kujas M, Parker F, Gaillard S, Raftopoulos C, et al: Factors predicting relapse of nonfunctioning pituitary macroadenomas after neurosurgery: a study of 142 patients. Eur J Endocrinol 163: , Buchfelder M, Schlaffer SM: Intraoperative magnetic resonance imaging during surgery for pituitary adenomas: pros and cons. Endocrine 42: , Cappabianca P, Cavallo LM, Colao A, de Divitiis E: Surgical complications associated with the endoscopic endonasal transsphenoidal approach for pituitary adenomas. J Neurosurg 97: , Chen Y, Li ZF, Zhang FX, Li JX, Cai L, Zhuge QC, et al: Gamma knife surgery for patients with volumetric classification of nonfunctioning pituitary adenomas: a systematic review and meta-analysis. Eur J Endocrinol 169: , Cho HY, Cho SW, Kim SW, Shin CS, Park KS, Kim SY: Silent corticotroph adenomas have unique recurrence characteristics compared with other nonfunctioning pituitary adenomas. Clin Endocrinol (Oxf) 72: , Ciric I, Ragin A, Baumgartner C, Pierce D: Complications of transsphenoidal surgery: results of a national survey, review of the literature, and personal experience. Neurosurgery 40: , Coburger J, König R, Seitz K, Bäzner U, Wirtz CR, Hlavac M: Determining the utility of intraoperative magnetic resonance imaging for transsphenoidal surgery: a retrospective study. Clinical article. J Neurosurg 120: , Colao A, Cerbone G, Cappabianca P, Ferone D, Alfieri A, Di Salle F, et al: Effect of surgery and radiotherapy on visual and endocrine function in nonfunctioning pituitary adenomas. J Endocrinol Invest 21: , Comtois R, Beauregard H, Somma M, Serri O, Aris-Jilwan N, Hardy J: The clinical and endocrine outcome to transsphenoidal microsurgery of nonsecreting pituitary adenomas. Cancer 68: , Darakchiev BJ, Tew JM Jr, Bohinski RJ, Warnick RE: Adaptation of a standard low-field (0.3-T) system to the operating room: focus on pituitary adenomas. Neurosurg Clin N Am 16: , de Lara D, Ditzel Filho LF, Prevedello DM, Otto BA, Carrau RL: Application of image guidance in pituitary surgery. Surg Neurol Int 3 (Suppl 2):S73 S78, De Witte O, Makiese O, Wikler D, Levivier M, Vandensteene A, Pandin P, et al: [Transsphenoidal approach with low field MRI for pituitary adenoma.] Neurochirurgie 51: , 2005 (Fr) 28. Fahlbusch R, Ganslandt O, Buchfelder M, Schott W, Nimsky C: Intraoperative magnetic resonance imaging during transsphenoidal surgery. J Neurosurg 95: , Fahlbusch R, Keller B, Ganslandt O, Kreutzer J, Nimsky C: Transsphenoidal surgery in acromegaly investigated by intraoperative high-field magnetic resonance imaging. Eur J Endocrinol 153: , Fatemi N, Dusick JR, Mattozo C, McArthur DL, Cohan P, Boscardin J, et al: Pituitary hormonal loss and recovery after transsphenoidal adenoma removal. Neurosurgery 63: , Gerlach R, du Mesnil de Rochemont R, Gasser T, Marquardt G, Reusch J, Imoehl L, et al: Feasibility of Polestar N20, an ultra-low-field intraoperative magnetic resonance imaging system in resection control of pituitary macroadenomas: lessons learned from the first 40 cases. Neurosurgery 63: , Hardy J, Wigser SM: Trans-sphenoidal surgery of pituitary fossa tumors with televised radiofluoroscopic control. J Neurosurg 23: , Jane JA Jr, Thapar K, Alden TD, Laws ER Jr: Fluoroscopic frameless stereotaxy for transsphenoidal surgery. Neurosurgery 48: , Jones J, Ruge J: Intraoperative magnetic resonance imaging in pituitary macroadenoma surgery: an assessment of visual outcome. Neurosurg Focus 23(5):E12, Knosp E, Steiner E, Kitz K, Matula C: Pituitary adenomas with invasion of the cavernous sinus space: a magnetic resonance imaging classification compared with surgical findings. Neurosurgery 33: , Kopp C, Theodorou M, Poullos N, Jacob V, Astner ST, Molls M, et al: Tumor shrinkage assessed by volumetric MRI in long-term follow-up after fractionated stereotactic radiotherapy of nonfunctioning pituitary adenoma. Int J Radiat Oncol Biol Phys 82: , Landolt AM: Surgical management of recurrent pituitary tumors, in Schmidek HH (ed): Operative Neurosurgical Techniques, ed 4. Philadelphia: WB Saunders, 2000, pp Laws ER Jr, Fode NC, Redmond MJ: Transsphenoidal surgery following unsuccessful prior therapy. An assessment of benefits and risks in 158 patients. J Neurosurg 63: , Laws ER Jr, Trautmann JC, Hollenhorst RW Jr: Transsphenoidal decompression of the optic nerve and chiasm. Visual results in 62 patients. J Neurosurg 46: , Lewin JS, Nour SG, Meyers ML, Metzger AK, Maciunas RJ, Wendt M, et al: Intraoperative MRI with a rotating, tiltable surgical table: a time use study and clinical results in 122 patients. AJR Am J Roentgenol 189: , Littley MD, Shalet SM, Beardwell CG, Ahmed SR, Applegate G, Sutton ML: Hypopituitarism following external radiotherapy for pituitary tumours in adults. Q J Med 70: , Losa M, Franzin A, Mangili F, Terreni MR, Barzaghi R, Veglia F, et al: Proliferation index of nonfunctioning pituitary adenomas: correlations with clinical characteristics and longterm follow-up results. Neurosurgery 47: , Losa M, Valle M, Mortini P, Franzin A, da Passano CF, Cenzato M, et al: Gamma knife surgery for treatment of residual nonfunctioning pituitary adenomas after surgical debulking. J Neurosurg 100: , Lundin P, Pedersen F: Volume of pituitary macroadenomas: assessment by MRI. J Comput Assist Tomogr 16: , J Neurosurg / Volume 121 / November 2014