- Email: [email protected]
Magnetic resonance imaging of the pituitary gland
Chntcal Ra&ology (1986) 37, %14 © 1986RoyalCollegeof Radiologists
0009-9260/86/450009502 00
Magnetic Resonance Imaging of the Pituitary Gland BENJAMIN GLASER, MAXIM SHEINFELD, JOSEPH BENMAIR* and NOAM KAPLAN*?
Institute of Endocrinology, Kupat Holim, Carmel Hospital, Haifa, Israel; *Magnetic Resonance Imaging Center, Elscint, Herzlia, Israel; ?Rakah Institute of Physics, Hebrew University, Jerusalem, Israel
Magnetic resonance imaging (MRI) was performed on 22 patients with a wide spectrum of pituitary disease and on 17 control subjects, in order to determine whether MRI is useful in localising and determining the extent of involvement of large and small pituitary tumours. The results indicate that MRI, with an 0.5 T superconducting magnet (Elscint, Israel), can be used to visualise large pituitary tumours and to determine the extent of parasellar and suprasellar involvement. The 'empty sella' syndrome can also be easily and confidently diagnosed. Some microadenomas can be visualised despite the relatively thick slices (0.7 cm) used in this study. These findings suggest that MRI in its current state of development is very useful in evaluating large pituitary lesions. With further technical refinements, improving the spatial resolution and decreasing slice thickness, this technique may come to be of some importance in the evaluation of pituitary microadenomas.
Increased awareness of the symptomatology of pituitary disease, along with the ready availability of pituitary hormone assays, has resulted in earlier diagnosis of pituitary tumours. Frequently, these tumours are diagnosed biochemically before they are demonstrable radiographically, even using high-resolution computed tomography (CT). With modern treatment approaches, adequate and safe visualisation of the tumours is becoming increasingly important. Although transphenoidal microsurgery is becoming more and more widely available, long-term medical management is also becoming more acceptable, such as bromocriptine in acromegaly (Clayton et al., 1978) and hyperprolactinaemia (McGregor et al., 1979) and cyproheptidine, bromocriptine and valproic acid in adrenocorticotrophic hormone (ACTH) tumours (Krieger, 1983). It is clear that preoperative visualisation of a tumour will facilitate its surgical removal. Perhaps more importantly, visualisation of tumours prior to and during medical management may be of help in evaluating the efficacy of therapy, particularly if subtle changes in the tumour's 'consistency' can be detected. Magnetic resonance imaging (MRI) has several theoretical advantages over other imaging modalities when imaging the pituitary (Bydder et al., 1982). The lack of bone artefacts and the availability of multiplanar imaging may make it possible to discern the fine architecture of the gland. Furthermore, since the magnetic resonance image is the result of the interaction of at least three separate tissue characteristics (proton density; longitudinal relaxation time, T1; transverse relaxation time, T2), it is possible that, by choosing the proper Address correspondenceto: BenjaminGlaser, MD, Departmentof Endocrinology&Metabolism,HadassahUniversityHospital,PO Box 12000, 91120Jerusalem, Israel.
pulse sequences, one may be able to detect very subtle changes in tumour 'consistency'. This study was intended as a preliminary investigation, to determine the potential of MRI in the visualisation of large and small pituitary lesions and to lay the groundwork for further development of MRI in evaluating patients with pituitary disease. PATIENTS AND METHODS
Seventeen control subjects, 13 males and four females (aged 20-59 years), who were either healthy volunteers or patients with no known or suspected pituitary disease, were studied in order to determine the appearance of the normal pituitary gland by MRI. All were healthy and with no known or suspected endocrine disorder elicited by clinical history. Twenty-two patients were selected from the endocrine clinic of Carmel Hospital or from outside referrals. Patients were pre-selected to represent a wide range of pituitary disease. The clinical details are given in Table 1. Included are patients with large and small tumours producing prolactin, ACTH or growth hormone. Some were treated medically, while others were not. Four patients with 'empty sella' syndrome are included: three with normal pituitary function and one with hyperprolactinaemia. Ten patients were studied who had biochemical evidence for pituitary tumours, with negative X-ray CT scans (Exel 2002 CT scanner; Elscint, Israel). All studies were approved by the human use committee of Carmel Hospital. No study was performed without prior written, informed consent from the patient or volunteer. Patients and control groups were scanned at the MRI Center, Herzlia, using a 0.5 T superconductive magnet instrument (Gyrex 5000; Elscint, Israel). Image acquisition utilised a selective multislice technique with contiguous spacing. Slice thickness was 0.7 cm, matrix size 2562 and image reconstruction by two-dimensional Fourier transformation. One to seven contiguous slices were obtained simultaneously, depending on the repetition time used in each protocol. Acquisition time varied from 2 min to 12 min, depending on the specific protocol used and the number of data averages obtained. Two pulse sequence techniques were used. (1) Spinecho (SE), with a basic cycle consisting of a ~/2-T-scT-(lst echo)-T-sr-T-(2nd echo) sequence. The value of Twas 14 ms, yielding a first echo time (Te) of 28 ms and a second echo time of Te=56 ms. The basic cycle was repeated for phase encoding, with a repetition time Tr. The above sequence is described in the text by the notation SErr/r 0. (2) Inversion-recovery (IR) images were obtained by applying a basic cycle with an inverting (at) pulse followed after a time T~ by a magnetisation
10
CLINICAL RADIOLOGY
Table 1 - Clinical details
Case
Age (years)
Sex
Clinical dtagnosls*
Treatment
X-ray CT dmgnosls
MR1 diagnosis
1
69
M
No treatment
2
62
F
Panhypopituitarism Hyperprolactinaemia Acromegaly
3
76
M
No treatment
4
48
F
Bromocriptlne
5 6
55 30
M F
Bromocrlptine Bromocriptine
Large tumour Large tumour
7 8 9 10
53 43 50 64
F F M F
Panhypopituitarism Hyperprolactinaemia Panhypopituitarism Hyperprolactinaemia Hyperprolactinaemia Inappropriate TSH level Hyperprolactinaemia Normal pituitary function Normal pituitary function Normal pituitary function Hyperprolactinaemia
Large tumour, suprasellar extension Large tumour, parasellar extension Large tumour, parasellar extension Large tumour
No treatment No treatment No treatment Bromocriptine
'Empty 'Empty 'Empty 'Empty
11
34
F
Hyperprolactinaemia
Bromocnptine
Non-pituitary tumour
12
46
F
Hyperprolactinaemia
Bromocnptine
Normal
13 14 15
21 47 33
F M F
Hyperprolactinaemia Hyperprolactinaemia Hyperprolactinaemia
Bromocriptine Bromocriptine Bromocriptine
Normal Normal Normal
16
24
F
Hyperprolactmaemia
Bromocriptine
Normal
17
36
F
Hyperprolactinaemia
Bromocrlptme
Normal
18
40
F
Cushing's syndrome
Sodium valproate
Normal
19
36
F
Cushing's syndrome
No treatment
Normal
20 21
48 31
F F
Cushing's syndrome Cushing's syndrome
No treatment No treatment
Normal Normal
22
45
F
Acromegaly
Bromocriptine
Large tumour, parasellar extension
Large tumour, suprasellar extension Large tumour, parasellar extension Large tumour, parasellar extension Large tumour 'Empty sella' syndrome Large tumour Large tumour, suprasellar extension 'Empty sella' syndrome 'Empty sella' syndrome 'Empty sella' syndrome Large turnout 'Empty sella' syndrome Tumour invasion of sella Non-pituitary tumour Mlcroadenoma, definitive diagnosis Normal Normal Microadenoma, suspicious Microadenoma, suspicious Microadenoma, definitive diagnosis Microadenoma, definitive diagnosis Microadenoma, suspicious Normal Microadenoma, suspicious Large tumour, paraseUar extension
No treatment
sampling spin-echo sequence. The basic IR sequence was repeated with a repetition time Tr. These sequences are described in the text by the notation IRvr/v,. This is a preliminary study, so that a wide variety of pulse timings were used in an attempt to establish which will give the most reliable results. Because of limitations of scan time, any given patient was scanned with only two or three different sequences. The slice thickness used in this study (0.7 cm) was in the same range as the dimensions of the normal pituitary, making volume averaging a significant problem. In order to minimise this, every effort was made to obtain exact midline studies. Furthermore, lesions were considered significant only if they could be viewed clearly in two perpendicular planes. Lesions visualised in one plane only were reported as 'suspicious', but not diagnostic.
RESULTS Normal Subjects The normal pituitary gland could be easily visualised on midline sagittal section as an ellipsoid mass projecting into the sphenoid sinus. While the bone of the selta turcica was not visualised, frequently bone marrow could be seen in the dorsum sellae a n d / o r on the floor of the sella. The normal pituitary gland was homogeneous with a fiat or slightly concave superior surface. The
sella' sella' sella' sella'
syndrome syndrome syndrome syndrome
diaphragma sellae could occasionally be localised as a black line (total lack of signal) immediately above the pituitary gland. In one control patient, a 20-year-old female with surgically proven adrenal adenoma (Cushing's syndrome), a mild upward convexity of the anterior pituitary could be identified. Above the pituitary was a clear space filled with cerebrospinal fluid (CSF) and, above that, the optic chiasm and optic nerve. Suprasellar enlargement with and without optic nerve involvement could be easily determined, as could upward convexity of the gland in the sagittal plane. However, since the tips of the posterior and anterior clinoid processes could not be seen, precise measurement of the degree of suprasellar extension could be difficult, especially when large tumours disrupted the pituitary architecture. The relationship of the gland to the parasellar sinus was best evaluated on the coronal projection. The dimensions of normal pituitary glands are given in Table 2. In no case was a pituitary adenoma found in a control section. The height of the gland was measured on both the sagittal and the coronal projections, whereas the length was determined on the sagittal view and the width on the coronal section. Table 2 - N o r m a l pituitary d i m e n s i o n s (range) as measured by M R I
A B C D
Antermr-posterior Superior-inferior Right-left Volume (A×B xC×n:/6)
0.7-1 2 cm 0.6-0.9 cm 0.7-1.0 cm 0 28-0.41 cm 3
MRI OF TIlE PITUITARY GLAND
(a)
]l
(b)
Fig ~ - (a) M~dline sagittal and ( b ) c~r~na~ s e c t i ~ s ( S E ~ ' 2 ~ ) t~r~ug~ a large~ gr~wth-h~rm~ne-pr~ducing tum~ur m a 6 3 - y e a r ~ d fema~e ( Case 2) Note the enlarged gland with a convex superior border on the sagittal section. On the coronal section, note the downward slope of the pituitary floor on the right (arrow). Note also the markedly thickened calvanum on the sagittal section, demonstrated by a very wide black band (bone) between the soft tissue of the brain and the subcutaneous fat.
Patients Seven patients with previously k n o w n large pituitary tumours were studied. T h r e e patients (Cases 1-3) with untreated tumours (two prolactinomas, one acromegaly) d e m o n s t r a t e d large, n o n - h o m o g e n e o u s lesions, the extent of which correlated well with C T findings. O n e had suprasellar extension, while two (Fig. 1), had extension into the parasellar area. T w o patients with m a c r o p r o l a c t i n o m a s (Cases 4, 5; Fig. 2) were studied after p r o l o n g e d t r e a t m e n t with b r o m o c r i p t i n e with normalisation of serum prolactin concentrations. In both patients, the large t u m o u r mass was mostly replaced by fluid-filled cavities, indicating degeneration of the t u m o u r . A n o t h e r patient (Case 22), with acromegaly, was also studied after 1 year of high doses
Fig. 2- Midline sagittal section (8E520/28) in a 48-year-old female (Case 4) with a 20-year history of galactorrhoea and amenorrhoea. The scan was made after 22 months of bromocriptine therapy. Prior to initiation of therapy, a CT scan had demonstrated a large, solid tumour with suprasellar and parasellar extension. Note the enlarged pituitary fossa virtually entirely filled with CSF (black). A small remnant of the pituitary can be seen on the floor of the anterior portion of the fossa (arrow) The infundibulum is seen crossing through the CSF toward the residual gland.
(60 m g / d a y ) of bromocriptine. H e r e , too, fluid-like areas were seen within the otherwise solid t u m o u r mass. O n e patient (Case 6; Fig. 3) has b e e n r e p o r t e d previously (Spitz e t al., 1984) as having pituitary resistance to thryoid h o r m o n e s . She had m a r k e d l y elevated thyroid-stimulating h o r m o n e (TSH) and prolactin levels associated with elevated T 4 and T 3 concentrations and clinical thyrotoxicosis. T h e thyrotoxicosis had b e e n controlled 1 y e a r previously with 131I, but T S H concentrations r e m a i n e d elevated. C o m p u t e d t o m o g r a p h y p e r f o r m e d 2 years previously s h o w e d no evidence of turnout. M a g n e t i c resonance imaging revealed a large, h o m o g e n e o u s mass with suprasellar extension, which was c o n f i r m e d by a repeat C T scan. F o u r patients (Cases 7-10) with the ' e m p t y sella' synd r o m e were studied. While three had entirely n o r m a l pituitary functions, the fourth (Case 10) had inappropriately elevated serum prolactin ( > 1 0 0 n g / m l ) ,
Fig. 3 - A midline saglttal scan (SE550J28)in a 30-year-old female (Case 6) with a TSH-producing pituitary tumour. Note the large tumour, which has extended superiorly up to the optic chiasma. (arrow).
12
CLINICAL RADIOLOGY
(a) (b) Fig. 4 - (a) Midline saglttal and (b) coronal scans (SE1200/2s) in a 53-year-old female with normal pituitary function (Case 7). Routine skull radiography and computed tomography showed an enlarged balloon sella filled with CSF, suggestive of 'empty sella' syndrome. Note that, except for a small band of pituitary tissue in the posterior, inferior part of the fossa (arrow), the entire sella is filled with CSF. On the coronal view, the infundibulum (arrow) can be seen in the mldline coursing down through the CSF-filled fossa
suggestive of prolactinoma. In all four patients, MRI demonstrated an enlarged pituitary fossa, predominantly filled with CSF-like substance (Fig. 4). In the patient with hyperprolactinaemia there was evidence of some tumour within the pituitary fossa, suggesting secondary 'empty sella' syndrome following degeneration of a macroadenoma. The patient in Case 11 had previously been treated with radiation therapy for nasopharyngeal carcinoma,
after which menstrual irregularity and hyperprolactinaemia were noted. Computed tomography revealed a normal pituitary with an opacified sphenoid sinus; however, polytomography (Fig. 5a) revealed a defect in the sellar floor, suggestive of pituitary tumour. Magnetic resonance imaging (Fig. 5b) revealed a pituitary gland of normal appearance and a mass lesion in the sphenoid sinus, with extension through the floor of the sella turcica in the region of the bony defect.
(a)
(b)
Fig. 5 - A 36-year-old female (Case 11), known to have an inoperable nasopharyngeal tumour. Polytomography (a) showed a defect in the sellar floor (arrow) and CT (not illustrated) showed a normal pituitary with opacification of the sphenoid sinus. (b) A midline saginal MR scan (SEss0/28) shows the tumour in the sphenoid sinus and the extension of the tumour through the floor of the sella m to the pituitary fossa (arrow).
MRI OF THE PITUITARY GLAND
Ten patients, with a clinical diagnosis of pituitary adenoma but with normal pituitary CT and laminography images, were chosen for study. Of six with hyperprolactinaemia, two had normal M R I scans, two (Cases 12, 17) had lesions ranging from 2 mm to 4 mm in size, clearly visualised on at least two views. Two additional patients (Cases 15, 16) had abnormalities noted on one view only. Case 17 demonstrated the importance of pulse sequence choice in these studies. The first study (SE1500/28) revealed a normal, homogeneous gland. Repeated at SE550/28, MRI revealed a suggestion of an area of increased intensity anteriorly in the gland, which became even more evident in the last sequence (SEas0/2s), which is essentially a saturation-recovery study, very sensitive to differences in T 1. A lesion of similar appearance could be seen on coronal and sagittal views of a 46-year-old woman (Case 12) with hyperprolactinaemia. In one patient (Case 16) a relatively large lesion of increased intensity was seen in the pituitary gland on sagittal section. Coronal sections in this patient were of poor quality, so that an artefact related to the partial volume effect could not be excluded and the study was reported as 'suspicious' of a pituitary lesion. All lesions which were visualised in these four patients appeared as an increased magnetic resonance (MR) signal and at pulse sequences that suggested a short T 1, but a T2 which was similar to that of surrounding normal tissue. Four patients had Cushing's syndrome with normal CT appearances of the pituitary. The pituitary origin of excessive cortisol excretion was established by full suppression by 8 mg dexamethasone per day, but not by 2 mg per day. One patient (Case 18) who has been reported previously (Glaser et al., 1984), had a clear lesion (decreased MR intensity) in the posterior of the gland. Another (Case 19) had a lesion of similar appearance in the pituitary, but this was seen only on one view, so it was described as 'suspicious'. A third patient had a mildly enlarged, homogeneous gland (10 m m x l 0 m m x l 2 mm; volume, 0.63 cm 3) whose posterior aspect had a convex upward superior border. Coronal sections clearly demonstrated an asymmetry, with deviation of the stalk to the right, suggesting a space-occupying lesion on the left. A 1 cm tumour was found at that location during transphenoidal surgery. The fourth patient (Case 20) appeared normal on MRI. The lesions that were seen were best visualised as areas of decreased intensity on SE15~0/2s and 8E1500/56 scans, suggestive of a short T2 relative to surrounding tissue. DISCUSSION The results of this study suggest that magnetic resonance imaging is potentially useful in the study of the pituitary lesions. Although early MRI studies, using weaker magnets, implied that M R I was not useful in identifying small tumours (Buonanno et al., 1983), more recent studies (Kaufman et al., 1983; Kelly et al., 1983) using updated equipment have been more optimistic. In this study, normal pituitary architecture was readily visualised in the control group. The normal pituitary is seen as a homogeneous, ellipsoid mass with an image intensity similar to that of normal white matter. The height of the normal pituitary as reported here (6-9 mm) is somewhat greater than that originally
13
reported in CT scans (Syvertsen et al., 1979); however, more recent studies suggest that normal pituitary height can reach 8 mm (Hemminghytt et al., 1983) or even 9 mm in women of child-brearing age (Swartz et al., 1983). The ability to visualise the upper border of the gland in two planes (sagittal and coronal) may aid in distinguishing normal from enlarged glands, although the significance of a convex upward curvature in the coronal plane is controversial. Some CT (Taylor, 1982) and autopsy (Sage et al., 1982) studies suggest that the upper border of the normal gland is flat or convex downward. Others (Swartz et al., 1983), however, report an upward convexity of the gland in a large percentage of normal women. A difference in pituitary size between men and women has been reported from CT studies (Syvertsen et al., 1979), but it is likely that additional variability exists among women, dependent on age and parity. Macroadenomas were readily diagnosed in all six patients. Suprasellar and parasellar extension was also easily identified. Tumour regression after bromocriptine was seen as CSF-filled cavities in the area previously occupied by the tumour. The normal pituitary could still be seen in the inferior portion of the fossa, with the pituitary stalk coursing through the empty fossa. Primary 'empty sella' syndrome was easily demonstrated in three patients, whereas in a fourth, with elevated prolactin levels, 'empty sella' syndrome secondary to spontaneous tumour degeneration was suggested. In the patient with nasopharyngeal malignancy and abnormal sella tomography, extension of the tumour through the floor of the sella at the point of the bony defect was demonstrated. In the 10 patients with pituitary hyperfunction suggestive of neoplastic disease, microadenomas or asymmetry were visualised with confidence in four (two prolactinomas and two Cushing's syndrome) and suspected in another three patients (two prolactinomas and one Cushing's syndrome). In our clinic, the tendency is to treat these patients medically or with radiation, rather than by neurosurgery. As a result, surgical verification of our MRI findings will not be possible. This study was performed using contiguous, 0.7 cmthick slices. Since microadenomas are frequently smaller than this, and, indeed, the entire pituitary gland is in this size range, volume averaging is a significant problem. Averaging with the adjacent carotid artery may produce the appearance of a lesion of decreased intensity, whereas inclusion of bone marrow in the clinoid process in a slice will produce the appearance of a lesion of increased intensity. For this reason, we considered a study to be positive for a lesion only if the lesion could be visualised in both the sagittal and the coronal plane. A lesion seen in one plane only was considered 'suspicious', even if the perpendicular study was felt to be suboptimal (as was the case in three patients). Even with these criteria, significant errors can occur. For this reasion, our MRI system is currently being modified to allow for increased resolution in the plane of study, to the range of 0.5 mm, with a slice thickness of 2-3 mm. This should greatly improve the sensitivity of MR! in the visualisation of microadenomas of the pituitary. Although the number of patients was too small to draw definite conclusions, our data suggest that MR characteristics may differ with different turnout types.
14
CLINICAL RADIOLOGY
Previous studies have produced conflicting results regarding MR characteristics of pituitary tumours. T1 values similar to (Bydder et al., 1982), greater than (Araki et al., 1984) or less than (Buonanno et al., 1983; Mills et al., 1984) that of surrounding brain tissue have been reported. This apparent discrepancy may, in part, be caused by failure of some studies to specify what type of pituitary tumour was studied. In this study, prolactinomas appeared to have short T1 and similar T2 when compared to normal tissue, whereas ACTH-producing turnouts (Cushing's syndrome) have a short T2 relative to normal tissue. Therefore, the choice of pulse Sequences may depend on the specific tumour in question or, conversely, the MR characteristics of a tumour may be useful in defining its biological activity. The importance of MRI of pituitary lesions goes beyond its potential for increased sensitivity and specificity when compared with CT. These tumours are relatively benign and frequently occur in young people. Follow-up of these patients is usually over a long time and requires repeated examinations. Frequently, young women require a follow-up throughout pregnancy, during which time enlargement of the turnout may occur. The availability of a technique of imaging the turnouts without ionising radiation or other health hazard is, therefore, of importance. Furthermore, these tumours are increasingly being treated medically: bromocriptine for prolactinomas and acromegaly and cyproheptidine, bromocriptine and valproic acid for Cushing's disease. Determining changes not only in tumour size, but also in 'consistency' may be of great help in evaluating response to these forms of therapy, again without radiation risk to the patients. In conclusion, this study suggests that MRI is potentially of use in evaluating the pituitary gland. The technique, in its current state of development, can be of use in evaluating and following patients with large pituitary lesions, that is, those greater than 1 cm in diameter, and with the 'empty sella' syndrome. With improved spatial resolution and thinner slices, the potential for its use in localising microadenomas may be realised. Further studies, which subdivide pituitary microadenomas according to their physical and biochemical characteristics, will be needed to establish optimal pulse sequencies for visualisation of various types of tumour. Since the completion of this study, contiguous multi-slice operation with slice thickness of 1 mm and in-slice pixel size of 0 5 mm have been
introduced at the Herzlia MRI Center This development has motivated the planning of a new study concerned with microadenomas of the pituitary gland, in accord with the foregoing concluding remarks.
REFERENCES
Araki, T., Inouye, T., Suzuki, H., Machida, T. & Iio, M. (1984). Magnetic resonance imaging of brain tumors: measurement of 7"1. Radiology, 150, 95-98 Buonanno, F. S., Pykett, I. L., Brady, T. J. & Pohost, G. M. (1983). Clinical applications of nuclear magnetic resonance (NMR). Disease-a-Month, 29, 1-81. Bydder, G. M., Stelner, R. E., Young, I. R., Hall, A. S., Thomas, D. S., Marshall, J. et al. (1982). Chmcal, NMR imaging of the brain: 140 cases. American Journal of Roentgenology, 139, 215236. Also American Journal of Nuclear Radiology, 3, 459-480. Clayton, R. N., Vrionides, Y., Lynch, S. S., Butt, W. B. & London, D. R. (1978). Response of acromegaly to long term bromocrlptine therapy: a biochemical and clinical assessment. Acta Endocrmologica (Copenhagen), 89, 469-482. Glaser, B., Kahana, L., Elias, V. & Sheinfeld, M. (1984). Sodium valproate and metyrapone in the management of pituitary dependent Cushing's disease. Lancet, ii, 640. Hemminghytt, S., Kalkhoff, R. K., Daniels, D. L., Williams, A. L., Grogen, J. P. & Haughton, V. M. (1983). Computed tomographic study of hormone-secreting microadenomas. Radiology, 146, 65-69. Kaufman, B., Arafah, B. M., Brodkey, J. S , Pearson, O. H., Bonstelle, C. T., Han, J. S. et al. (1983). NMR in evaluating pituitary tumors (abstract). Radiology, 149, 1977. Kelly, W. M., Glover, (3. H., Norman, D., MacFall, J. R., Grlgsby, N., Wehrli, F. W. et al. (1983). High-resolution NMR imaging of the pituitary gland (abstract). Radiology, 149, 97. Krieger, D. T. (1983). Physiopathology of Cushing's disease. Endocrine Reviews, 4, 22-43. McGregor, A. M., Scanlon, M. F , Hall, K., Cook, D. B & Hall, R (1979). Reduction in size of a pituitary tumor by bromocriptine therapy. New England Journal of Medicine, 300, 291-293. Mills, C. M., Crooks, L E., Kaufman, L. & Brant-Zawadzki, M. (1984). Cerebral abnormalities: use of calculated T1 and T2 magnetic resonance images for diagnosis. Radiology, 150, 87-94. Sage, M. R., Blumbergs, P. C., Mulligan, B P. & Fowler, G. W. (1982). The diaphragma sellae: its relationship to the configuration of the pituitary gland. Radiology, 145, 703-708 Spitz, I. M., Sheinfeld, M., Glaser, B. & Hirsch, H. S. (1984). Inappropriate TSH secretion in association with hyperprolactinaemia. Postgraduate Medical Journal, 60, 328-335. Swartz, J. D., Russell, K. B., Basile, B. A., O'Donnell, P. C. & Porky, G. L. (1983). High resolution computed tomographic appearance of the intrasellar contents in women of childbearing age. Radiology, 147, 115-117. Syvertsen, A., Haughton, V. M., Williams, A. L. & Cusik, J. (1979). Computed tomography of the normal pituitary gland and microadenomas. Radiology, 133, 385-392. Taylor, S. (1982). High resolution computed tomography of the sella. Radiologic Chmcs of North America, 20, 207-236.
0009-9260/86/450009502 00
Magnetic Resonance Imaging of the Pituitary Gland BENJAMIN GLASER, MAXIM SHEINFELD, JOSEPH BENMAIR* and NOAM KAPLAN*?
Institute of Endocrinology, Kupat Holim, Carmel Hospital, Haifa, Israel; *Magnetic Resonance Imaging Center, Elscint, Herzlia, Israel; ?Rakah Institute of Physics, Hebrew University, Jerusalem, Israel
Magnetic resonance imaging (MRI) was performed on 22 patients with a wide spectrum of pituitary disease and on 17 control subjects, in order to determine whether MRI is useful in localising and determining the extent of involvement of large and small pituitary tumours. The results indicate that MRI, with an 0.5 T superconducting magnet (Elscint, Israel), can be used to visualise large pituitary tumours and to determine the extent of parasellar and suprasellar involvement. The 'empty sella' syndrome can also be easily and confidently diagnosed. Some microadenomas can be visualised despite the relatively thick slices (0.7 cm) used in this study. These findings suggest that MRI in its current state of development is very useful in evaluating large pituitary lesions. With further technical refinements, improving the spatial resolution and decreasing slice thickness, this technique may come to be of some importance in the evaluation of pituitary microadenomas.
Increased awareness of the symptomatology of pituitary disease, along with the ready availability of pituitary hormone assays, has resulted in earlier diagnosis of pituitary tumours. Frequently, these tumours are diagnosed biochemically before they are demonstrable radiographically, even using high-resolution computed tomography (CT). With modern treatment approaches, adequate and safe visualisation of the tumours is becoming increasingly important. Although transphenoidal microsurgery is becoming more and more widely available, long-term medical management is also becoming more acceptable, such as bromocriptine in acromegaly (Clayton et al., 1978) and hyperprolactinaemia (McGregor et al., 1979) and cyproheptidine, bromocriptine and valproic acid in adrenocorticotrophic hormone (ACTH) tumours (Krieger, 1983). It is clear that preoperative visualisation of a tumour will facilitate its surgical removal. Perhaps more importantly, visualisation of tumours prior to and during medical management may be of help in evaluating the efficacy of therapy, particularly if subtle changes in the tumour's 'consistency' can be detected. Magnetic resonance imaging (MRI) has several theoretical advantages over other imaging modalities when imaging the pituitary (Bydder et al., 1982). The lack of bone artefacts and the availability of multiplanar imaging may make it possible to discern the fine architecture of the gland. Furthermore, since the magnetic resonance image is the result of the interaction of at least three separate tissue characteristics (proton density; longitudinal relaxation time, T1; transverse relaxation time, T2), it is possible that, by choosing the proper Address correspondenceto: BenjaminGlaser, MD, Departmentof Endocrinology&Metabolism,HadassahUniversityHospital,PO Box 12000, 91120Jerusalem, Israel.
pulse sequences, one may be able to detect very subtle changes in tumour 'consistency'. This study was intended as a preliminary investigation, to determine the potential of MRI in the visualisation of large and small pituitary lesions and to lay the groundwork for further development of MRI in evaluating patients with pituitary disease. PATIENTS AND METHODS
Seventeen control subjects, 13 males and four females (aged 20-59 years), who were either healthy volunteers or patients with no known or suspected pituitary disease, were studied in order to determine the appearance of the normal pituitary gland by MRI. All were healthy and with no known or suspected endocrine disorder elicited by clinical history. Twenty-two patients were selected from the endocrine clinic of Carmel Hospital or from outside referrals. Patients were pre-selected to represent a wide range of pituitary disease. The clinical details are given in Table 1. Included are patients with large and small tumours producing prolactin, ACTH or growth hormone. Some were treated medically, while others were not. Four patients with 'empty sella' syndrome are included: three with normal pituitary function and one with hyperprolactinaemia. Ten patients were studied who had biochemical evidence for pituitary tumours, with negative X-ray CT scans (Exel 2002 CT scanner; Elscint, Israel). All studies were approved by the human use committee of Carmel Hospital. No study was performed without prior written, informed consent from the patient or volunteer. Patients and control groups were scanned at the MRI Center, Herzlia, using a 0.5 T superconductive magnet instrument (Gyrex 5000; Elscint, Israel). Image acquisition utilised a selective multislice technique with contiguous spacing. Slice thickness was 0.7 cm, matrix size 2562 and image reconstruction by two-dimensional Fourier transformation. One to seven contiguous slices were obtained simultaneously, depending on the repetition time used in each protocol. Acquisition time varied from 2 min to 12 min, depending on the specific protocol used and the number of data averages obtained. Two pulse sequence techniques were used. (1) Spinecho (SE), with a basic cycle consisting of a ~/2-T-scT-(lst echo)-T-sr-T-(2nd echo) sequence. The value of Twas 14 ms, yielding a first echo time (Te) of 28 ms and a second echo time of Te=56 ms. The basic cycle was repeated for phase encoding, with a repetition time Tr. The above sequence is described in the text by the notation SErr/r 0. (2) Inversion-recovery (IR) images were obtained by applying a basic cycle with an inverting (at) pulse followed after a time T~ by a magnetisation
10
CLINICAL RADIOLOGY
Table 1 - Clinical details
Case
Age (years)
Sex
Clinical dtagnosls*
Treatment
X-ray CT dmgnosls
MR1 diagnosis
1
69
M
No treatment
2
62
F
Panhypopituitarism Hyperprolactinaemia Acromegaly
3
76
M
No treatment
4
48
F
Bromocriptlne
5 6
55 30
M F
Bromocrlptine Bromocriptine
Large tumour Large tumour
7 8 9 10
53 43 50 64
F F M F
Panhypopituitarism Hyperprolactinaemia Panhypopituitarism Hyperprolactinaemia Hyperprolactinaemia Inappropriate TSH level Hyperprolactinaemia Normal pituitary function Normal pituitary function Normal pituitary function Hyperprolactinaemia
Large tumour, suprasellar extension Large tumour, parasellar extension Large tumour, parasellar extension Large tumour
No treatment No treatment No treatment Bromocriptine
'Empty 'Empty 'Empty 'Empty
11
34
F
Hyperprolactinaemia
Bromocnptine
Non-pituitary tumour
12
46
F
Hyperprolactinaemia
Bromocnptine
Normal
13 14 15
21 47 33
F M F
Hyperprolactinaemia Hyperprolactinaemia Hyperprolactinaemia
Bromocriptine Bromocriptine Bromocriptine
Normal Normal Normal
16
24
F
Hyperprolactmaemia
Bromocriptine
Normal
17
36
F
Hyperprolactinaemia
Bromocrlptme
Normal
18
40
F
Cushing's syndrome
Sodium valproate
Normal
19
36
F
Cushing's syndrome
No treatment
Normal
20 21
48 31
F F
Cushing's syndrome Cushing's syndrome
No treatment No treatment
Normal Normal
22
45
F
Acromegaly
Bromocriptine
Large tumour, parasellar extension
Large tumour, suprasellar extension Large tumour, parasellar extension Large tumour, parasellar extension Large tumour 'Empty sella' syndrome Large tumour Large tumour, suprasellar extension 'Empty sella' syndrome 'Empty sella' syndrome 'Empty sella' syndrome Large turnout 'Empty sella' syndrome Tumour invasion of sella Non-pituitary tumour Mlcroadenoma, definitive diagnosis Normal Normal Microadenoma, suspicious Microadenoma, suspicious Microadenoma, definitive diagnosis Microadenoma, definitive diagnosis Microadenoma, suspicious Normal Microadenoma, suspicious Large tumour, paraseUar extension
No treatment
sampling spin-echo sequence. The basic IR sequence was repeated with a repetition time Tr. These sequences are described in the text by the notation IRvr/v,. This is a preliminary study, so that a wide variety of pulse timings were used in an attempt to establish which will give the most reliable results. Because of limitations of scan time, any given patient was scanned with only two or three different sequences. The slice thickness used in this study (0.7 cm) was in the same range as the dimensions of the normal pituitary, making volume averaging a significant problem. In order to minimise this, every effort was made to obtain exact midline studies. Furthermore, lesions were considered significant only if they could be viewed clearly in two perpendicular planes. Lesions visualised in one plane only were reported as 'suspicious', but not diagnostic.
RESULTS Normal Subjects The normal pituitary gland could be easily visualised on midline sagittal section as an ellipsoid mass projecting into the sphenoid sinus. While the bone of the selta turcica was not visualised, frequently bone marrow could be seen in the dorsum sellae a n d / o r on the floor of the sella. The normal pituitary gland was homogeneous with a fiat or slightly concave superior surface. The
sella' sella' sella' sella'
syndrome syndrome syndrome syndrome
diaphragma sellae could occasionally be localised as a black line (total lack of signal) immediately above the pituitary gland. In one control patient, a 20-year-old female with surgically proven adrenal adenoma (Cushing's syndrome), a mild upward convexity of the anterior pituitary could be identified. Above the pituitary was a clear space filled with cerebrospinal fluid (CSF) and, above that, the optic chiasm and optic nerve. Suprasellar enlargement with and without optic nerve involvement could be easily determined, as could upward convexity of the gland in the sagittal plane. However, since the tips of the posterior and anterior clinoid processes could not be seen, precise measurement of the degree of suprasellar extension could be difficult, especially when large tumours disrupted the pituitary architecture. The relationship of the gland to the parasellar sinus was best evaluated on the coronal projection. The dimensions of normal pituitary glands are given in Table 2. In no case was a pituitary adenoma found in a control section. The height of the gland was measured on both the sagittal and the coronal projections, whereas the length was determined on the sagittal view and the width on the coronal section. Table 2 - N o r m a l pituitary d i m e n s i o n s (range) as measured by M R I
A B C D
Antermr-posterior Superior-inferior Right-left Volume (A×B xC×n:/6)
0.7-1 2 cm 0.6-0.9 cm 0.7-1.0 cm 0 28-0.41 cm 3
MRI OF TIlE PITUITARY GLAND
(a)
]l
(b)
Fig ~ - (a) M~dline sagittal and ( b ) c~r~na~ s e c t i ~ s ( S E ~ ' 2 ~ ) t~r~ug~ a large~ gr~wth-h~rm~ne-pr~ducing tum~ur m a 6 3 - y e a r ~ d fema~e ( Case 2) Note the enlarged gland with a convex superior border on the sagittal section. On the coronal section, note the downward slope of the pituitary floor on the right (arrow). Note also the markedly thickened calvanum on the sagittal section, demonstrated by a very wide black band (bone) between the soft tissue of the brain and the subcutaneous fat.
Patients Seven patients with previously k n o w n large pituitary tumours were studied. T h r e e patients (Cases 1-3) with untreated tumours (two prolactinomas, one acromegaly) d e m o n s t r a t e d large, n o n - h o m o g e n e o u s lesions, the extent of which correlated well with C T findings. O n e had suprasellar extension, while two (Fig. 1), had extension into the parasellar area. T w o patients with m a c r o p r o l a c t i n o m a s (Cases 4, 5; Fig. 2) were studied after p r o l o n g e d t r e a t m e n t with b r o m o c r i p t i n e with normalisation of serum prolactin concentrations. In both patients, the large t u m o u r mass was mostly replaced by fluid-filled cavities, indicating degeneration of the t u m o u r . A n o t h e r patient (Case 22), with acromegaly, was also studied after 1 year of high doses
Fig. 2- Midline sagittal section (8E520/28) in a 48-year-old female (Case 4) with a 20-year history of galactorrhoea and amenorrhoea. The scan was made after 22 months of bromocriptine therapy. Prior to initiation of therapy, a CT scan had demonstrated a large, solid tumour with suprasellar and parasellar extension. Note the enlarged pituitary fossa virtually entirely filled with CSF (black). A small remnant of the pituitary can be seen on the floor of the anterior portion of the fossa (arrow) The infundibulum is seen crossing through the CSF toward the residual gland.
(60 m g / d a y ) of bromocriptine. H e r e , too, fluid-like areas were seen within the otherwise solid t u m o u r mass. O n e patient (Case 6; Fig. 3) has b e e n r e p o r t e d previously (Spitz e t al., 1984) as having pituitary resistance to thryoid h o r m o n e s . She had m a r k e d l y elevated thyroid-stimulating h o r m o n e (TSH) and prolactin levels associated with elevated T 4 and T 3 concentrations and clinical thyrotoxicosis. T h e thyrotoxicosis had b e e n controlled 1 y e a r previously with 131I, but T S H concentrations r e m a i n e d elevated. C o m p u t e d t o m o g r a p h y p e r f o r m e d 2 years previously s h o w e d no evidence of turnout. M a g n e t i c resonance imaging revealed a large, h o m o g e n e o u s mass with suprasellar extension, which was c o n f i r m e d by a repeat C T scan. F o u r patients (Cases 7-10) with the ' e m p t y sella' synd r o m e were studied. While three had entirely n o r m a l pituitary functions, the fourth (Case 10) had inappropriately elevated serum prolactin ( > 1 0 0 n g / m l ) ,
Fig. 3 - A midline saglttal scan (SE550J28)in a 30-year-old female (Case 6) with a TSH-producing pituitary tumour. Note the large tumour, which has extended superiorly up to the optic chiasma. (arrow).
12
CLINICAL RADIOLOGY
(a) (b) Fig. 4 - (a) Midline saglttal and (b) coronal scans (SE1200/2s) in a 53-year-old female with normal pituitary function (Case 7). Routine skull radiography and computed tomography showed an enlarged balloon sella filled with CSF, suggestive of 'empty sella' syndrome. Note that, except for a small band of pituitary tissue in the posterior, inferior part of the fossa (arrow), the entire sella is filled with CSF. On the coronal view, the infundibulum (arrow) can be seen in the mldline coursing down through the CSF-filled fossa
suggestive of prolactinoma. In all four patients, MRI demonstrated an enlarged pituitary fossa, predominantly filled with CSF-like substance (Fig. 4). In the patient with hyperprolactinaemia there was evidence of some tumour within the pituitary fossa, suggesting secondary 'empty sella' syndrome following degeneration of a macroadenoma. The patient in Case 11 had previously been treated with radiation therapy for nasopharyngeal carcinoma,
after which menstrual irregularity and hyperprolactinaemia were noted. Computed tomography revealed a normal pituitary with an opacified sphenoid sinus; however, polytomography (Fig. 5a) revealed a defect in the sellar floor, suggestive of pituitary tumour. Magnetic resonance imaging (Fig. 5b) revealed a pituitary gland of normal appearance and a mass lesion in the sphenoid sinus, with extension through the floor of the sella turcica in the region of the bony defect.
(a)
(b)
Fig. 5 - A 36-year-old female (Case 11), known to have an inoperable nasopharyngeal tumour. Polytomography (a) showed a defect in the sellar floor (arrow) and CT (not illustrated) showed a normal pituitary with opacification of the sphenoid sinus. (b) A midline saginal MR scan (SEss0/28) shows the tumour in the sphenoid sinus and the extension of the tumour through the floor of the sella m to the pituitary fossa (arrow).
MRI OF THE PITUITARY GLAND
Ten patients, with a clinical diagnosis of pituitary adenoma but with normal pituitary CT and laminography images, were chosen for study. Of six with hyperprolactinaemia, two had normal M R I scans, two (Cases 12, 17) had lesions ranging from 2 mm to 4 mm in size, clearly visualised on at least two views. Two additional patients (Cases 15, 16) had abnormalities noted on one view only. Case 17 demonstrated the importance of pulse sequence choice in these studies. The first study (SE1500/28) revealed a normal, homogeneous gland. Repeated at SE550/28, MRI revealed a suggestion of an area of increased intensity anteriorly in the gland, which became even more evident in the last sequence (SEas0/2s), which is essentially a saturation-recovery study, very sensitive to differences in T 1. A lesion of similar appearance could be seen on coronal and sagittal views of a 46-year-old woman (Case 12) with hyperprolactinaemia. In one patient (Case 16) a relatively large lesion of increased intensity was seen in the pituitary gland on sagittal section. Coronal sections in this patient were of poor quality, so that an artefact related to the partial volume effect could not be excluded and the study was reported as 'suspicious' of a pituitary lesion. All lesions which were visualised in these four patients appeared as an increased magnetic resonance (MR) signal and at pulse sequences that suggested a short T 1, but a T2 which was similar to that of surrounding normal tissue. Four patients had Cushing's syndrome with normal CT appearances of the pituitary. The pituitary origin of excessive cortisol excretion was established by full suppression by 8 mg dexamethasone per day, but not by 2 mg per day. One patient (Case 18) who has been reported previously (Glaser et al., 1984), had a clear lesion (decreased MR intensity) in the posterior of the gland. Another (Case 19) had a lesion of similar appearance in the pituitary, but this was seen only on one view, so it was described as 'suspicious'. A third patient had a mildly enlarged, homogeneous gland (10 m m x l 0 m m x l 2 mm; volume, 0.63 cm 3) whose posterior aspect had a convex upward superior border. Coronal sections clearly demonstrated an asymmetry, with deviation of the stalk to the right, suggesting a space-occupying lesion on the left. A 1 cm tumour was found at that location during transphenoidal surgery. The fourth patient (Case 20) appeared normal on MRI. The lesions that were seen were best visualised as areas of decreased intensity on SE15~0/2s and 8E1500/56 scans, suggestive of a short T2 relative to surrounding tissue. DISCUSSION The results of this study suggest that magnetic resonance imaging is potentially useful in the study of the pituitary lesions. Although early MRI studies, using weaker magnets, implied that M R I was not useful in identifying small tumours (Buonanno et al., 1983), more recent studies (Kaufman et al., 1983; Kelly et al., 1983) using updated equipment have been more optimistic. In this study, normal pituitary architecture was readily visualised in the control group. The normal pituitary is seen as a homogeneous, ellipsoid mass with an image intensity similar to that of normal white matter. The height of the normal pituitary as reported here (6-9 mm) is somewhat greater than that originally
13
reported in CT scans (Syvertsen et al., 1979); however, more recent studies suggest that normal pituitary height can reach 8 mm (Hemminghytt et al., 1983) or even 9 mm in women of child-brearing age (Swartz et al., 1983). The ability to visualise the upper border of the gland in two planes (sagittal and coronal) may aid in distinguishing normal from enlarged glands, although the significance of a convex upward curvature in the coronal plane is controversial. Some CT (Taylor, 1982) and autopsy (Sage et al., 1982) studies suggest that the upper border of the normal gland is flat or convex downward. Others (Swartz et al., 1983), however, report an upward convexity of the gland in a large percentage of normal women. A difference in pituitary size between men and women has been reported from CT studies (Syvertsen et al., 1979), but it is likely that additional variability exists among women, dependent on age and parity. Macroadenomas were readily diagnosed in all six patients. Suprasellar and parasellar extension was also easily identified. Tumour regression after bromocriptine was seen as CSF-filled cavities in the area previously occupied by the tumour. The normal pituitary could still be seen in the inferior portion of the fossa, with the pituitary stalk coursing through the empty fossa. Primary 'empty sella' syndrome was easily demonstrated in three patients, whereas in a fourth, with elevated prolactin levels, 'empty sella' syndrome secondary to spontaneous tumour degeneration was suggested. In the patient with nasopharyngeal malignancy and abnormal sella tomography, extension of the tumour through the floor of the sella at the point of the bony defect was demonstrated. In the 10 patients with pituitary hyperfunction suggestive of neoplastic disease, microadenomas or asymmetry were visualised with confidence in four (two prolactinomas and two Cushing's syndrome) and suspected in another three patients (two prolactinomas and one Cushing's syndrome). In our clinic, the tendency is to treat these patients medically or with radiation, rather than by neurosurgery. As a result, surgical verification of our MRI findings will not be possible. This study was performed using contiguous, 0.7 cmthick slices. Since microadenomas are frequently smaller than this, and, indeed, the entire pituitary gland is in this size range, volume averaging is a significant problem. Averaging with the adjacent carotid artery may produce the appearance of a lesion of decreased intensity, whereas inclusion of bone marrow in the clinoid process in a slice will produce the appearance of a lesion of increased intensity. For this reason, we considered a study to be positive for a lesion only if the lesion could be visualised in both the sagittal and the coronal plane. A lesion seen in one plane only was considered 'suspicious', even if the perpendicular study was felt to be suboptimal (as was the case in three patients). Even with these criteria, significant errors can occur. For this reasion, our MRI system is currently being modified to allow for increased resolution in the plane of study, to the range of 0.5 mm, with a slice thickness of 2-3 mm. This should greatly improve the sensitivity of MR! in the visualisation of microadenomas of the pituitary. Although the number of patients was too small to draw definite conclusions, our data suggest that MR characteristics may differ with different turnout types.
14
CLINICAL RADIOLOGY
Previous studies have produced conflicting results regarding MR characteristics of pituitary tumours. T1 values similar to (Bydder et al., 1982), greater than (Araki et al., 1984) or less than (Buonanno et al., 1983; Mills et al., 1984) that of surrounding brain tissue have been reported. This apparent discrepancy may, in part, be caused by failure of some studies to specify what type of pituitary tumour was studied. In this study, prolactinomas appeared to have short T1 and similar T2 when compared to normal tissue, whereas ACTH-producing turnouts (Cushing's syndrome) have a short T2 relative to normal tissue. Therefore, the choice of pulse Sequences may depend on the specific tumour in question or, conversely, the MR characteristics of a tumour may be useful in defining its biological activity. The importance of MRI of pituitary lesions goes beyond its potential for increased sensitivity and specificity when compared with CT. These tumours are relatively benign and frequently occur in young people. Follow-up of these patients is usually over a long time and requires repeated examinations. Frequently, young women require a follow-up throughout pregnancy, during which time enlargement of the turnout may occur. The availability of a technique of imaging the turnouts without ionising radiation or other health hazard is, therefore, of importance. Furthermore, these tumours are increasingly being treated medically: bromocriptine for prolactinomas and acromegaly and cyproheptidine, bromocriptine and valproic acid for Cushing's disease. Determining changes not only in tumour size, but also in 'consistency' may be of great help in evaluating response to these forms of therapy, again without radiation risk to the patients. In conclusion, this study suggests that MRI is potentially of use in evaluating the pituitary gland. The technique, in its current state of development, can be of use in evaluating and following patients with large pituitary lesions, that is, those greater than 1 cm in diameter, and with the 'empty sella' syndrome. With improved spatial resolution and thinner slices, the potential for its use in localising microadenomas may be realised. Further studies, which subdivide pituitary microadenomas according to their physical and biochemical characteristics, will be needed to establish optimal pulse sequencies for visualisation of various types of tumour. Since the completion of this study, contiguous multi-slice operation with slice thickness of 1 mm and in-slice pixel size of 0 5 mm have been
introduced at the Herzlia MRI Center This development has motivated the planning of a new study concerned with microadenomas of the pituitary gland, in accord with the foregoing concluding remarks.
REFERENCES
Araki, T., Inouye, T., Suzuki, H., Machida, T. & Iio, M. (1984). Magnetic resonance imaging of brain tumors: measurement of 7"1. Radiology, 150, 95-98 Buonanno, F. S., Pykett, I. L., Brady, T. J. & Pohost, G. M. (1983). Clinical applications of nuclear magnetic resonance (NMR). Disease-a-Month, 29, 1-81. Bydder, G. M., Stelner, R. E., Young, I. R., Hall, A. S., Thomas, D. S., Marshall, J. et al. (1982). Chmcal, NMR imaging of the brain: 140 cases. American Journal of Roentgenology, 139, 215236. Also American Journal of Nuclear Radiology, 3, 459-480. Clayton, R. N., Vrionides, Y., Lynch, S. S., Butt, W. B. & London, D. R. (1978). Response of acromegaly to long term bromocrlptine therapy: a biochemical and clinical assessment. Acta Endocrmologica (Copenhagen), 89, 469-482. Glaser, B., Kahana, L., Elias, V. & Sheinfeld, M. (1984). Sodium valproate and metyrapone in the management of pituitary dependent Cushing's disease. Lancet, ii, 640. Hemminghytt, S., Kalkhoff, R. K., Daniels, D. L., Williams, A. L., Grogen, J. P. & Haughton, V. M. (1983). Computed tomographic study of hormone-secreting microadenomas. Radiology, 146, 65-69. Kaufman, B., Arafah, B. M., Brodkey, J. S , Pearson, O. H., Bonstelle, C. T., Han, J. S. et al. (1983). NMR in evaluating pituitary tumors (abstract). Radiology, 149, 1977. Kelly, W. M., Glover, (3. H., Norman, D., MacFall, J. R., Grlgsby, N., Wehrli, F. W. et al. (1983). High-resolution NMR imaging of the pituitary gland (abstract). Radiology, 149, 97. Krieger, D. T. (1983). Physiopathology of Cushing's disease. Endocrine Reviews, 4, 22-43. McGregor, A. M., Scanlon, M. F , Hall, K., Cook, D. B & Hall, R (1979). Reduction in size of a pituitary tumor by bromocriptine therapy. New England Journal of Medicine, 300, 291-293. Mills, C. M., Crooks, L E., Kaufman, L. & Brant-Zawadzki, M. (1984). Cerebral abnormalities: use of calculated T1 and T2 magnetic resonance images for diagnosis. Radiology, 150, 87-94. Sage, M. R., Blumbergs, P. C., Mulligan, B P. & Fowler, G. W. (1982). The diaphragma sellae: its relationship to the configuration of the pituitary gland. Radiology, 145, 703-708 Spitz, I. M., Sheinfeld, M., Glaser, B. & Hirsch, H. S. (1984). Inappropriate TSH secretion in association with hyperprolactinaemia. Postgraduate Medical Journal, 60, 328-335. Swartz, J. D., Russell, K. B., Basile, B. A., O'Donnell, P. C. & Porky, G. L. (1983). High resolution computed tomographic appearance of the intrasellar contents in women of childbearing age. Radiology, 147, 115-117. Syvertsen, A., Haughton, V. M., Williams, A. L. & Cusik, J. (1979). Computed tomography of the normal pituitary gland and microadenomas. Radiology, 133, 385-392. Taylor, S. (1982). High resolution computed tomography of the sella. Radiologic Chmcs of North America, 20, 207-236.