Original Article MRI Research of Diaphragma Sellae in Patients with Pituitary Adenoma

Int J Clin Exp Med 2015;8(8):12842-12849 www.ijcem.com /ISSN:1940-5901/IJCEM0009225

Original Article MRI research of diaphragma sellae in patients with pituitary adenoma

Liangfeng Wei*, Zhiyu Xi*, Shun’an Lin, Qingshuang Zhao, Junjie Jing, Shousen Wang

Department of Neurosurgery, Fuzhou General Hospital, Second Military Medical University, Fuzhou 350025, P. R. China. *Equal contributors. Received April 16, 2015; Accepted August 7, 2015; Epub August 15, 2015; Published August 30, 2015

Abstract: This study is to investigate the clinical significance of diaphragma sellae in patients with pituitary adenoma by MR images. A total of 47 cases of pituitary adenoma patients were enrolled in this study. Preoperative and postoperative MR scanning together with preoperative 3D-GE sequential scanning were performed. A series of parameters of diaphragma sellae were measured and compared. Tumor height was greater in patients with convex diaphragma sellae than that in patients with concave diaphragma sellae. The width and height of diaphragmal opening were positively related to tumor height. Diaphragmal opening width in the invasive group was greater than that of the non invasive group. Diaphragmal opening width in the non total resection group was significantly greater than that in the total resection group. Tumor resection rate was negatively correlated to diaphragmal opening width in the non total resection group. Lift angle of bilateral epidural around diaphragmal opening was positively related to tumor height. Enhanced 3D-GE images can perfectly display diaphragma sellae and parameters of tumor height and tumor invasiveness are related to diaphragmal opening diameter.

Keywords: Pituitary adenoma, diaphragma sellae, diaphragmal opening, MRI

Introduction with normal people. Farn et al. [11] found that while compared with the conventional spin Diaphragma sellae (DS) is a layer of epidural echo (SE) enhanced images, epidural struc- mater structure, covering the above of pituitary tures displaying in 3D gradient echo (GE) and forming the top of pituitary fossa. For sequential enhanced images was with a higher patients with pituitary adenoma (PA), DS has a percentage, longer fragments and better conti- special meaning in both tumor growth and nuity. Based on previous studies, in order to transsphenoidal approach operation [1, 2]. better display DS in patients with PA, 3D-GE Anatomy studies on DS morphology of normal sequential enhanced scanning was used and people are conducted [1, 3-6], however, no in- the scanning parameters had been adjusted in depth reports of DS are published for patients this study. The accurate parameters were mea- with PA. As early as 1986, Daniels et al. [7] sured to investigate the clinical application firstly observed DS in patients with PA in MR value of DS in patients with PA. images and they described it as a low shadow signal line above pituitary. Because contrast Materials and methods agent was not widely used in clinical at that time, nor did the existence of the sequence Patients’ data concept, the image displaying effect for DS was not ideal. With the development of MR technol- A total of 47 cases of patients admitted to our ogy, many scholars have focused on epiepidur- hospital and diagnosed as PA from March 2012 all structures such as DS using MR [8-10]. to April 2013 were enrolled. Among them, 22 However, they all paid attention to DS display- were male and 25 were female. They were aged ing effect but did not further analyze DS mor- from 23-70 years old, with an average age of phology not to mention compare PA patients 45.1 years old. Among them, 1 case was micro MRI research of diaphragma sellae

Figure 1. Measurement markers of all parameters used in 3D-GE sequential enhanced MRI coronal scanning. A. The right bordering of diaphragma sellae dura connecting to the lateral wall of cavernous sinus. B. Diaphragma sellae dura end on the right of diaphragmal opening. C. Diaphragma sellae dura end on the left of diaphragmal opening. D. The left starting point of diaphragma sellae dura. The distance between “b” and “c” was diaphragmal opening width. The sum of the distance of “a”-“b” and “c”-“d” was the width of diaphragma sellae dura on both sides of diaphragmal opening. The average value of the angle between “a”-“b”, “c”-“d” and the horizontal line was the lift angle of diaphragma sellae dura on both sides of diaphragmal opening. The arrows showed the pituitary stalk. adenoma (diameter < 10 mm), 41 cases were adrenocorticotropic hormone type, 7 cases of large adenoma (10 mm < diameter < 40 mm) gonadotropin type, 1 case of thyroid stimulat- and 5 cases were giant adenoma (diameter > ing hormone type and 5 cases of multiple hor- 40 mm). Tumor height was 8.90-60.11 mm, mone type. with an average height of 26.40 ± 11.59 mm. The main symptoms included headaches, The inclusion criteria were described as fol- blurred vision, amenorrhea, lactation, acro- lows. PA patients who had not taken surgery megaly, loss of libido, etc. The longest duration before admission. Patients who could cooper- was 10 years and the shortest was less than 1 ate with MR checking preoperative and postop- month. Postoperative immunohistochemistry erative. Patients those taken endonasal trans- confirmed that there were 18 cases of nonfunc- sphenoidal operation of tumor resection and tional adenoma, 7 cases of prolactin type, 7 the operators were the corresponding authors cases of growth hormone type, 2 cases of of this study. Patients with PA confirmed by sur-

12843 Int J Clin Exp Med 2015;8(8):12842-12849 MRI research of diaphragma sellae

Figure 2. Diaphragma sellae morphology in 3D-GE sequential enhanced MRI coronal images. A. Concave type. B. Horizontal type. The white arrow showed the pituitary stalk. C. Convex type. The sellar was pushed upward by the tumor and diaphragmal opening was expanded. The white arrow showed the pituitary stalk.

Table 1. Measuring data of pituitary adenoma and diaphragma MRI examination sellae of the 47 cases of patients All the patients received MRI Measurement parameters Measurement result examination before and after SE sequence surgery with Siemens trio Tumor height (26.40 ± 11.59) mm 3.0T MR (Global Siemens Tumor resection extent (91.10 ± 13.80)% Healthcare Headquarters, Er- 3D-GE sequence (refer to each marker of Figure 1) langen, Germany). Axial, coro- Diaphragmal opening width (7.67 ± 5.11) mm nal and sagittal plain and Lift angle of diaphragmal opening bilateral dura (46.53 ± 31.95)° enhanced scan images of SE Diaphragmal opening bilateral dura width (18.37 ± 6.60) mm sequential T1WI and T2W Diaphragmal opening bilateral dura thickness (0.94 ± 0.20) mm were collected. Coronal 3D- GE enhanced scan was taken preoperatively. The parameters employed were as fol- lows: TR/TE, 7/2 ms; Fov, 180 mm × 180 mm; matrix, 448 × 305; slice thickness, 0.8 mm. At 30 s after the injection of contrast agent, uninterrupted continuous scanning was carried out. The contrast agent was GD DTPA, with a dosage of 0.1 mmol/kg, i.v.

Image analysis

The collected MRI images were read by at least two Figure 3. Scatter diagram of the relationship between tumor height and the width of the diaphragm sellae opening. attending or rather seniority physicians and an attending or rather seniority neurosurgery. The exclusion criteria were PA patients geon. DS morphology was observed in 3D-GE taking operation because of recurrence. Prior coronal enhanced images. DS was the inward written and informed consent were obtained extended enhancing shadow line between bilat- from all patients and the study was approved eral cavernous segment of internal carotid by the ethics review board of the Fuzhou artery and supraclinoidal segment. In most General Hospital, Fujian Medical University. cases, DS extended inward to bilateral sides of

12844 Int J Clin Exp Med 2015;8(8):12842-12849 MRI research of diaphragma sellae

Table 2. Comparison of tumor invasive and diaphragmal probability was used for ranked data opening width comparison. Nonparametric test was Invasive Non invasive performed for data that did not meet Cases group group normal distribution. 21 26 Results Diaphragmal opening width (mm) 10.30 ± 5.79 5.55 ± 3.29 P value 0.002 Comparison of DS morphology

In order to understand the growth char- Table 3. Comparison of whether tumor was complete acteristics of PA, DS morphology of the resection or not with diaphragmal opening width patients were recorded and compared. Complete Incomplete DS were integrity in all the 47 cases and Cases resection resection fell into concave type, horizontal type 29 18 and convex type. The cases of the three Diaphragmal opening width (mm) 5.58 ± 4.14 11.04 ± 4.80 types were 3 (6.4%), 10 (21.3%) and 34 P value 0.000 (72.3%), respectively. Tumor height of the three types were (13.57 ± 4.42) mm, (22.03 ± 10.77) mm and (28.82 ± 11.29) hypophyseal stalk and combined with it. In mm, respectively. Tumor height of the convex some cases, there might be no hypophyseal type was significantly larger than that of con- stalk, leaving the free epidural border. The sur- cave type (P < 0.05). However, there were no rounding part of the ellipse edge, i.e, the significant difference in tumor height neither central part of diaphragma sellae dura that between the convex type and horizontal type or without dura covered, was identified as the dia- between horizontal type and concave type. In phragmal opening. The layer that maximally dis- summary, these results indicated that the played diaphragmal opening was selected. The tumor could lift up diaphragma sellae dura width, thickness, and lift angle of bilateral while it grew upward, however, there was no epiepidurall together with diaphragmal open- direct evidence showing that diaphragmal ing width were measured with PACS system opening could expand. (Figures 1 and 2). Tumor morphology was observed in SE enhanced images, meanwhile, Differences of diaphragmal openin width grading indexes like tumor height and SIPAP [12] were recorded (Table 1). In the enhanced In order to identify whether diaphragmal open- images before and after operation, tumor ing was related to tumor growth, diaphragmal boundary was artificially outlined and tumor openin widths in different groups were com- area of each layer was automatically measured pared. Diaphragmal opening width of the 47 with computer. Tumor volume was calculated cases of PA patients was 1.45-21.87 mm with by adding thickness of each layer preoperative an average of (7.67 ± 5.11) mm. Diaphragmal and postoperative, thereby, the degree of tumor opening width and tumor height were made into resection was calculated. The correlation am- scatter plot and analyzed with linear regres- ong DS, diaphragmal opening indexes and sion, and the result showed positive correlation tumor markers was compared. between the two parameters (r = 0.224, P = 0.003) (Figure 3). According to Hardy-Wilson Statistical analysis staging [13] and Knosp grading [14] classification, preoperative MRI of W-Hardy staging All data were analyzed by SPSS19.0 software above stage C and Knosp grading higher than (SPSS Inc, Chicago, IL, USA) and P < 0.05 was level 3 were considered as invasive PA group considered as statistically significant. All data and the conversely were taken as non invasive were analyzed with Kolmogorov-Smirnov test PA group. There were 21 cases (44.7%) in inva- and data met normal distribution were ex- sive group and 26 cases (55.3%) in non inva- _ pressed as x ± s. A linear regression was per- sion group. Diaphragmal opening width of the formed between the two variables. An indepen- invasive group was significantly higher than dent-sample t test was carried to compare the that of non invasive group, with statistically sig- difference between groups. χ2/Fisher exact nificant difference (Table 2). There were 29

12845 Int J Clin Exp Med 2015;8(8):12842-12849 MRI research of diaphragma sellae

neither between the invasive group and non invasive group nor between complete and incomplete groups. These results indicated that the tumor had upward pushing force all the time it grows even if it has passed the diaphragmal opening.

Comparison of bilateral epidural around diaphragmal opening width and thickness

To identify whether there was biological invasion or phago- Figure 4. Scatter diagram of the relationship between tumor height and lift cytosis exsisted, bilateral epi- angle of diaphragmal opening bilateral epidural. dural around diaphragmal opening width and thickness cases of tumor complete resection (61.7%) and were measured and compared. Average width 18 cases (38.3%) of incomplete resection. The of bilateral epidural around diaphragmal incomplete resection cases all were from the opening of the 47 cases was (18.37 ± 6.60) invasive group. Diaphragmal opening width of mm and there was no significant correlation incomplete resection cases was significantly (r = 0.277, P = 0.059) between epidural width larger than that of the complete resection and tumor height. Bilateral epidural around cases, with statistically significant differences diaphragmal opening width of invasive group (Table 3). In the 18 cases of non complete and non invasion group were (18.59 ± 6.33) resection patients, tumor resection rate and mm and (18.19 ± 6.93) mm, respectively, diaphragmal opening width were negatively with no significant difference between them. correlated (r = -0.576, P = 0.02). As a conclu- The width of bilateral epidurall around dia- sion, the above results argued that the larger phragmal opening in complete resection group the diaphragmal opening, the easier the tumor and incomplete resection group were (19.10 ± grows upward to diaphragmal opening and the 6.58) mm and (17.18 ± 6.63) mm, and there harder the tumor to be removed. was no significant difference. Average thickness of bilateral epidural around diaphragmal Comparison of the lift angle of bilateral epi- opening in the 47 cases was (0.94 ± 0.20) dural around diaphragmal opening mm, and there was no significant correlation (r = 0.050, P = 0.740) between epidural thick- To find out whether lift angle of diaphragmal ness and tumor height. There was no sig- opening caused by tumor was positively lin- nificant difference neither between invasive ked to tumor height, the lift angle of bilateral group and non invasive group nor between epiepidurall around diaphragmal opening complete resection group and incomplete was measured and compared. The average resection group. To sum up, these results sug- degree of the lift angle of bilateral epidural gested that no invasion or phagocytosis was around diaphragmal opening (i.e. bilateral dia- found no matter the PA was invasive or non phragma sellae around diaphragmal opening) invasive. of the 47 cases of patients was (46.53 ± 31.95). Lift angle and tumor height were made Discussion into scatter plot and analyzed with linear regression and the result indicated positive In this study, 3D-GE sequential MRI was adopt- correlation between the two parameters (r = ed to show DS and this method had received 0.586, P = 0.000) (Figure 4). There was no good results. Based on previous reports [7-11, significant difference in lift angle of bilateral 14, 15], a series of parameters used in this epidural around diaphragmal opening (P > 0.5) study had been modified. Short TR and TE were

12846 Int J Clin Exp Med 2015;8(8):12842-12849 MRI research of diaphragma sellae used to shorten the time of a single excitation It is reported that 35% PA are diagnosed as in order to adjust matrix and vision thus improv- invasive adenoma [16]. During the growth, inva- ing signal to noise ratio. A 0.8 mm ultrathin sive PA can destroy the medial wall of cavern- layer thickness and continuous scanning was ous sinus, sellar epidural, even the floor of sella adopted to increase DS appearing layer num- bone structure. As the pituitary fossa top, DS is bers as far as possible. Image collection was the only way by which PA growing to the saddle. started 30 s after injection of contrast agent so The middle of DS is diaphragmal opening, as to achieve rich intravenous development of which is a natural passage that leading to sad- DS epidural structures using the circulation dle. Just as the existence of diaphragmal open- period within contrast agents into the venous. ing, making the relationship of DS with tumor Besides, the image scanning was taken sing different from the medial wall of cavernous 3.0T MRI with high resolution, thus observing sinus and sellar epidural. To date, whether the satisfactory DS morphology. One minor short contact mean is infiltrating growth or just a sim- coming was that sagittal and axial images were ple expansion process, there is controversy. In not collected lacking of three-dimensional this study, histological technique had not been observation effect. The method of measuring used to test whether there was infiltration of and analyzing DS morphology in patients with tumor on DS, however, MR observation itself PA after DSon MRI has not been reported. also had some persuasion. One short coming was that as DS is relatively thin and currently Due to the limitation of specimens, previous MRI precision is limited, as a result, there are anatomical study on DS and PA was difficult to certain errors in actual measurement process. be linked together. As a result, there was no direct evidence for mutual influence between Yao Yiqun et al. [17] have taken biopsy of DS PA growth and DS, and the relationship was during surgery, and their results showed that mainly remained on theoretical speculation. there was infiltration of PA on part of DS. As DS None of the sporadical imaging researches on is an important structure for preventing the DS in patients with PA [7, 8, 15] measured the leakage of cerebrospinal fluid, direct biopsy is accurate parameters of DS. Daniels et al. [7] not widely recognized. Wang Jianxin et al. [18] has studied DS structure with MRI plain scan, proposed that invasive PA growth might not however, DS were observed only in some cases. play a destructive role by means of infiltration, In addition, the researchers focused mainly on but by expansion, which causing the thinning of DS displaying effect. In this study, enhanced surrounding structures. They also argued that MRI was used and DS displaying received ideal PA was monoclonal origin. Their conclusion result in each case. Cattin et al. [8] have com- could be applied to other parts including DS. pared SE sequential sagittal enhanced MRI of Defects of the medial wall of cavernous sinus, PA and normal people. They found that DS epi- sellar epidural and bone were seen and even dural thickness in some patients with PA was cavernous sinus top wall defect appeared in thickened. However, they did not compare the this study, however, all DS were complete. data systematically and they just observed There was no statistic difference between inva- from one point of view, which was not a com- sive group and non invasive group in DS thick- prehensive understanding. Nomura et al. [15] ness and bilateral epidural around diaphragmal used transsphenoidal approach parallel level opening comparison, which could not support scanning for observing sellar region. Pituitary the existence of DS infiltration. Lift angle of stalk and diaphragmal opening were discov- bilateral epidural around diaphragmal opening ered in MRI images in some pituitary micro was positively related to tumor height, suggest- adenoma patients and were measured. How- ing that tumor growth had the effect of upward ever, pituitary macro adenomas and giant ade- process. Diaphragmal opening width in inva- noma were not observed, thus, it was difficult sive group was significantly higher than that of to analyze the effect of tumor growth on DS. In non invasive group indicated that invasive PA this study, not only 1 case of micro adenoma could expand diaphragmal opening while pass- but more macro adenomas and giant adeno- ing from diaphragmal opening into saddle. mas were included, which, can comprehensive- Hence one can see that as the tumor grows ly reflect the interaction between tumor and upward, the main effect is squeezing DS upward DS. A short coming of this study was that rela- and the evidence of whether there is infiltration tively few cases were collected in this study. is not sufficient.

12847 Int J Clin Exp Med 2015;8(8):12842-12849 MRI research of diaphragma sellae

It was reported that postoperative residual rate [2] Guinto Balanzar G, Abdo M, Mercado M, of non functioning adenomas was 50% [19]. Nishimura E and Arechiga N. Diaphragma sel- Another researcher reported non functional lae: a surgical reference for transsphenoidal giant PA growing to saddle in 68 patients resection of pituitary macroadenomas. World receiving transsphenoidal operation. One stage Neurosurg 2011; 75: 286-93. complete resection accounted for only 20.6% [3] Kursat E, Yilmazlar S, Aker S, Aksoy K and Oygucu H. Comparsion of lateral and superior (20 cases), and the other 48 cases all had walls of the pituitary fossa with clinical empha- residual tumors [20]. Postoperative tumor sis on pituitary adenoma extension: cadaveric- residual of cavernous sinus was common in anatomic study. Neurosurg Rev 2008; 31: 91- clinical, and suprasellar part residual was often 9. seen. Besides tumor texture, DS is another rea- [4] Songtao Q, Yuntao L, Jun P, Chuanping H, son that causing tumor difficult to settle into Xiaofeng S. Membranous layers of the pituitary intrasellar. The larger the preoperative imaging gland: histological anatomic study and related of diaphragmal opening, the more difficulty the clinical issues. Neurosurgery 2009; 64: ons1- tumor to be totally extracted. 9. [5] Ju KS, Bae HG, Park HK, Chang JC, Choi SK In conclusion, 3D-GE enhanced imaging was and Sim KB. Morphometric study of the Korean used to display DS and ediaphragmal opening, adult pituitary glands and the diaphragma sel- and this method could be popularized in relat- lae. J Korean Neurosurg Soc 2010; 47: 42-9. ed research in the future. During the growth of [6] Perondi GE, Isolan GR, de Aguiar PH, Stefani PA, it could push DS upward. diaphragmal MA and Falaetta EF. Endoscopic anatomy of opening width and lift angle of diaphragmal sellar region. Pituitary 2013; 16: 251-9. opening bilateral epidural were positively relat- [7] Daniels DL, Pojunas KW, Kilgore DP, Pech P, ed to tumor height. The diaphragmal opening Meyer GA and Williams AL. MR of the diaphragm sellae. AJNR 1986; 7: 765-9. of invasive PA was larger than non invasive PA [8] Cattin F, Bonneville F, Andrea I, Barrali E and and the lager the tumor, the harder the tumor to Bonneville JF. Dural enhancement in pituitary be totally extracted. macroadenomas. Neuroradiology 2000; 42: 505-8. Acknowledgements [9] Dietemann JL, Correia Bernardo R, Bogorin A, Abu Eid M, Koob M, Nogueira Th, Vargas MI, This work was supported by the Twelfth Five Fakhoury W and Zollner G. Normal and abnor- Year Plan of Nanjing Military Medical Innovation mal meningeal enhancement: MRI features. J Key Project (2011Z034). We greatly appreciate Radiol 2005; 86: 1659-83. all the help from associate professor Qun [10] Smirniotopoulos JG, Murphy FM, Rushing EJ, Zhong and technician Ming Ma from Department Rees JH and Schroeder JW. Patterns of con- of Radiology of our hospital for technical assis- trast enhancement in the brain and meninges. tance in MR scanning and reading. Radiographics 2007; 27: 525-51. [11] Farn JW and Mirowitz SA. MR imaging of nor- Disclosure of conflict of interest mal meninges: comparison of contrast-enhancement pattern on 3D gradient-echo and None. spin-echo images. AJR 1994; 162: 131-5. [12] Meyer S, Valdemarsson S and Larsson EM. Address correspondence to: Shousen Wang, De- Classification of pituitary growth hormone pro- partment of Neurosurgery, Fuzhou General Hospital, ducing adenomas according to SIPAP: applica- Second Military Medical University, No. 156, tion in clinical practice. Acta Radiol 2011; 52: Xi’erhuanbei Road, Fuzhou 350025, P. R. China. 796-801. Tel: +86-591-24937080; Fax: +86-591-87640785; [13] Wilson CB. A decade of pituitary microsurgery. E-mail: [email protected] J Neurosurg 1984; 61: 814-33. [14] Knosp E, Steiner E, Kitz K and Matula C. PA References with invasion of the cavernous sinus space: a magnetic resonance imaging classification [1] Campero A, Matins C, Yasuda A and Rhoton AL. compared with surgical findings. Neurosurgery Microsurgical anatomy of the diaphragma sel- 1993; 33: 610-8. lae and its role in directing the pattern of [15] Nomura M, Tachibana O, Yamashima T, growth of pituitary adenomas. Neurosurgery Yamashita J and Suzuki M. MRI evaluation of 2008; 62: 717-23. the diaphragmal opening: using MRI parallel to

12848 Int J Clin Exp Med 2015;8(8):12842-12849 MRI research of diaphragma sellae

the transsphenoidal surgical approach. J Clin [19] Macro L, Alberto F, Francesca M, Maria R, Neurosci 2002; 9: 175-7. Raffaella B, Fabrizio V, Pietro M and Massimo [16] Hansen TM, Batra S, Lim M, Gallia GL, Burger G. Proliferation index of nonfunctioning PA: cor- PC, Salvatori R, Wand G, Quinones-Hinojosa A, relations with clinical characteristics and long- Kleinberg L and Redmond KJ. Invasive adeno- term follow-up results. Neurosurgery 2000; 47: ma and pituitary carcinoma: a SEER database 1313-8. analysis. Neurosurg Rev 2014; 37: 279-86. [20] Wang JX, Qi ST, Zeng H and Peng YP. Stagedly [17] Yao YQ, Li DH, Liu LP, Zeng EM and Li L. Transsphenoidal Operations on Giant Nonfun- Significance of diaphragma sellae dura biopsy ctioning Pituitary Adenoma with Suprasellar in diagnosis of pituitary adenoma invasive- Extension. Chinese General Practice 2011; 14: ness. Journal of Jiangxi Medical College 2009; 622-4. 49: 60-3. [18] Wang JX, Qi ST, Lu YT, Peng YP, Pan J, Fan J and Wang RZ. Pathomorphological study on effects of pituitary adenoma on sella floor’s dura. Chinese Journal of Neurosurgery 2012; 28: 688-92.

12849 Int J Clin Exp Med 2015;8(8):12842-12849