Education Resource from the Society for Endocrinology
J Byrne
Department of Neuroradiology, The Radcliffe Infirmary, Woodstock Road, Oxford OX2 6HE
Summer School 13-16 July 2004
St Anne's College, Oxford University,
UK
Magnetic resonance imaging (MRI) is the optimum imaging method with high quality computer tomography (CT) an acceptable alternative.
The advantages of MRI are:-
The axial plane is a relatively
poor technique for demonstrating the vertical relationships of structures
lying between the floor of the IIIrd ventricle
and sella turcica, therefore assessment of the pituitary and hypothalamus
is easiest in the sagittal and coronal planes. MRI in these planes
provides the
optimum demonstration of the pituitary stalk and the relationships
of intrasellar contents with the cavernous and sphenoid sinuses as well as
the contents
of the chiasmatic cistern.
Various technical refinements to pituitary MRI have been advocated
but given the enormous number of potentially useful MRI sequence protocols,
basic scanning
methods are remarkably similar between centres. It is agreed that the
structures of the sella region are best imaged using T1-weighted (T1W)
sequences,
since cerebrospinal fluid provides inherent tissue contrast with the
surrounding structures.
In practice T1W spin echo sequences are performed with repetition times (TR) of 500-600msec, echo times (TE) of 15msec and 2 or more excitations. Scanning is performed in coronal and sagittal planes using matrix size 256 x 256, to give 3mm thick contiguous slices. Typically scanning takes 5-8 minutes at 1.5T for each sequence. An alternative approach is to use a T1W gradient echo technique with a three-dimensional (3D) Fourier synthesis so that subsequent computer manipulation allows the imaged sample to be viewed in any plane. Some centres perform both sequence types because image clarity is superior in the former but the 3D sequence allows post-processing to review and clarify any suspicious areas.
The disadvantage of MRI (apart from logistical difficulties of its availability and cost) is its relative insensitivity to pathological calcification and lack of signal from corticated bone.
The need for routine intra-venous administration of paramagnetic agents, such as gadolinium, is controversial. The pituitary gland and stalk enhance whilst the hypothalamus and optic chiasm don’t if the blood-brain barrier is intact. Surrounding blood vessels, meninges and the mucosa of the paranasal sinuses all enhance. On T1W unenhanced MRI it is normally easy to identify these structures. The same is not true of CT and intravenous injection of iodinated contrast media has been advocated to improve tissue contrast. CT, unlike MRI, can’t distinguish blood flow without contrast agents, nor intracranial arteries and veins. After contrast enhancement, intra-dural arteries can be differentiated on CT but similar enhancement levels of arterial and venous blood within the cavernous sinus prevents demonstration of the intracavernous carotid arteries as separate structures.
Fast MRI sequences have been used to study the timing of intravenously administered gadolinium uptake by the hypophysis. So called dynamic scans are obtained at 20-30 second intervals and have been used to demonstrate that the stalk and posterior lobe enhance 20 seconds after gadolinium injection and that enhancement extends into the anterior portion of the gland within 80 seconds. Dynamic scanning has been developed as a means of increasing detection rates of microadenomas.
CT may be required to demonstrate or exclude pathological calcification and is more sensitive than plain film radiography. CT scanning remains the primary imaging modality for a small proportion of patients who are unable to undergo MRI because of extreme claustrophobia, cardiac pacemakers or other corporal metal implants such as intracranial aneurysms clips and traumatic metallic fragments. CT using multislice scanner can now be performed so that images can be reformatted into the coronal or sagittal planes by computer post-processing.
The disadvantages of CT over MRI should not be exaggerated. Radiation exposure is a definite concern when an individual undergoes multiple follow-up scans but the accuracy of CT is similar to MRI for diagnosis of macroadenoma. It will provide virtually all the pre-operative data for transphenoidal hypophysectomy, including the bony structure of the ethmoid and sphenoid sinuses, though not an accurate assessment of the position of the carotid arteries. Pre-operative angiography may then be necessary and angiography should always be considered when CT raises the possibility of a thrombosed aneurysm. Intra-arterial angiography is now virtually obsolete since MR and CT angiography (MRA, CTA) are capable of identifying the positions of the intra-cavernous and supra-clinoid carotid arteries and differentiate pituitary mass lesions from aneurysms. Very rarely the diagnoses of a substantially thrombosed aneurysm requires intra-arterial digital subtraction angiography (IA-DSA). Angiography (i.e. venography)continues to have a role during catheter navigation for venous sampling in patients being investigated for causes of Cushing’s syndrome.
On CT adenomas are isodense or hypodense relative to brain tissue and show variable patterns of enhancement after radiographic contrast media administration.
On MRI signal return is typically similar to that of brain on both T1W and T2W sequences. Cysts or areas of necrosis cause foci of moderate hypointensity on T1W and hyperintensity on T2W sequences and heterogeneous enhancement with gadolinium. Signal due to haemorrhage is more specific but temporally variable because the magnetic effects of iron within red blood cells change as they break up and haemoglobin degrades. In general these are best appreciated on T1W MRI since within days of haemorrhage concentrations of methaemoglobin increase in areas of haemorrhage causing T1 recovery time shortening and bright signal on this sequence. In the acute period after haemorrhage (less than 3 days) MRI changes are non-specific but acute haemorrhage is hyperdense on CT. As haematomas liquidify in the next 2-3 weeks, their density reduces and they become hypodense on CT. Methaemoglobin and haemosiderin can persist for months in the brain so that chronic haemorrhage is easily identified but the temporal evolution of haemorhage within pituitary adenomas is unknown because bleeding may be asymptomatic. Though the scan appearances of tumours in patients presenting acutely with pituitary apoplexy will usually reflect the extent of haemorrhage and/or necrosis and gadolinium enhancement will occur at the margins of necrotic areas.
Scanning should help the surgeon by differentiating tumour from normal gland and by identifying likely areas of local invasion. Tumour invasion of the cavernous sinus, sphenoid bone and extension into the chiasmatic cistern are evident on MRI. Such behaviour has been identified surgically and histologically in all tumour types.
Asymmetry of the cavernous sinuses, displacement of the medial walls and of the carotid arteries are inconsistent features of invasion. Displacement of the lateral wall is a more reliable sign. Administration of gadolinium is helpful in identifying normal pituitary gland and distinguishing it from tumour.
The timing of post-operative scanning is important since early post-surgical changes due to local swelling (in the first 1-2 weeks) and surgical packing materials, used in the transphenoidal exposure, may show little reduction in the mass of a pituitary tumour. Biological packing material returns mixed signal with fat being hyperintense and muscle isointense on T1W MRI. Re-expansion of the normal pituitary gland and reabsorption of packing material is usually evident on follow-up scans at 3 months. Early scanning is therefore only useful to investigate possible surgical complications and scanning to identify residual tumour is best delayed for at least 3 months. The signal returned by residual tumour is usually the same as that of tumour at presentation but gadolinium enhancement is less helpful at distinguishing it from normal gland after operation.
Imaging has a continuing role in follow-up of treated patients. Demonstration of tumour regression or recurrence relies on comparisons between follow-up scans.
The protocol for post-operative follow-up imaging in Oxford is for a baseline study to be obtained 4 months after hypophysectomy followed by interval MRI, 12 months and 5 years later. More frequent scans are obtained if the patient’s visual fields change or histological examination of the resected tumour suggests local invasiveness. Patients treated medically for functional macroadenomas are also monitored by serial imaging, in combination with biochemical and clinical follow up assessments.
In most patients the presence of a microadenoma is assumed from biochemical testing and imaging is undertaken to confirm an intrasella source and to guide its transphenoidal excision. To identify adenomas less than 10mm in size demands the highest standards of imaging technique and interpretation. Microadenomas show little inherent contrast to normal pituitary tissue on CT and scanning requires intravenous radiographic contrast agents to demonstrate non-enhancement of the microadenoma against a background of normal gland enhancement. On MRI, microadenomas are typically hypointense relative to normal gland on T1W sequences and this inherent contrast may or may not be amplified by scanning after enhancement with intravenous gadolinium. It is generally accepted that MRI is more accurate.
An alternative approach to increase the detection rate for microadenoma is dynamic MRI. This technique employs rapid sequential imaging to show temporal differences in gadolinium uptake between adenoma and normal gland. In this way microadenomas which enhance shortly after normal gland enhancement can be identified. Dynamic scanning with various fast imaging techniques have been used to identify microadenomas not detected on conventional MRI. However the improved sensitivity (67% compared with 52%) enjoyed by dynamic MRI is offset by reduced specificity (80% versus 100%) compared to conventional MRI. The sensitivity of unenhanced high resolution MRI for pituitary microadenoma is in the order of 60-80%. Conventional scanning with contrast-enhancement detects 5-10% more lesions and dynamic scanning a further 5-10% of lesions. Detection rate can thus be improved but at the expense of a higher false positive rate.
The pituitary fossa and suprasella regions are sites of a gamut of non-pituitary tumours that may cause endocrine disturbances by virtue of being located within or close to the hypothalamic-pituitary axis.
The imaging differential is best considered by study of the effect of a mass lesion on the optic chiasm.
Lesions that arise above the chiasm:-
Craniopharyngioma, haemangioblastoma,
glioma (usually astrocytoma), hamartoma of the hypothalamus, lipoma and ependymoma
within the anterior part of the
IIIrd venticle.
Lesions that arise below the chiasm:-
Meningioma, aneurysm, schwannoma (particularly of the trigeminal nerve), chordoma,
chondrosarcoma, lymphoma and metastases. These lesions should be considered
in the differential diagnosis of pituitary macroadenomas as well as the rarer
tumours of the neurohypophysis: pilocystic astrocytoma and granular cell tumour
(also known as choristoma).
Lesions that arise in the chiasmatic cistern:-
Optic nerve glioma, meningioma,
craniopharyngioma, aneurysm and metastasis.
Several inflammatory conditions may affect the sellar and parasellar tissues causing symptoms and signs similar to pituitary neoplasms.
These include:-
Pituitary Abscess
Lymphocytic hypophysitis
Wegener’s granulomatosis
Langerhans cell histiocytosis
Tuberculous meningitis
Granulomatous leptomeningitis (sarcoid granulomas)
Nuclear medicine techniques such as positron emission tomography (PET) or single-photon emission tomography (SPECT) have been used to obtain in-vivo characterisation of pituitary tissue and to differentiate it from other neoplasms. The presence of octreotide-binding somatostatin receptors in non-functioning adenomas cause them to take up 111In-DTPA-octreotide but meniogiomas may also express somatostatin receptors and take up somatostatin receptor specific isotopes. However tracers that bind to the enzyme monoamino oxidase b have been used to differentiate meningioma from pituitary adenoma using PET and more recently a D2 dopamine receptor specific isotope, [18F] fluoro-ethyl-spiperone to differentiate non-functioning adenomas from craniopharyngioma and meningioma. PET using tracers such as [18F] fluorodeoxyglucose and [11C] methionine can be used to study rates of glucose metabolism and protein synthesis. However the current availability of scanners limits the use of PET to selected patients and research.
The opinions expressed in this paper are those of the speaker and do not necessarily reflect the views of the Society
Revised:
05-Nov-2004