Although the amenorrhea-galactorrhea syndrome and other clinical manifestations of pituitary disorders have been known from antiquity, true advances in the understanding of their physiology, pathology, and clinical correlations have occurred only since the beginning of this century. In the past three decades in particular, we have witnessed explosive advances in immunohistopathologic diagnosis, imaging techniques, hormonal assays, receptor mediated pharmacologic therapy, trans-sphenoidal microsurgical techniques, and stereotactic simultaneous multi port focused beam radiation therapy.

Sporadic cases of pituitary carcinoma have been reported since the turn of this century, but the authenticity of some of the early reported cases is in doubt. Carcinomas of the sphenoid and other paranasal sinuses invading the cavernous sinus, sellar floor, and pituitary gland have been mistaken for primary pituitary carcinoma. Metastases to the pituitary from remote primary sites such as the lung and kidneys have led to mistaken diagnoses as well. Such errors occurred in part because of sole reliance on certain morphologic features and tinctorial properties with a few elementary stains with light microscopy, without having the benefit of specific immunohistochemical stains and the electron microscope, the blessings of contemporary endocrine pathology.

Definition

The operational definition of pituitary carcinoma has been the subject of debate. Zulch questioned whether such an entity exists. Jefferson made a distinction between "malignant pituitary adenomas," which he believed to be those tumours that were locally invasive and had cytologic features of malignancy, and true "pituitary carcinomas," which showed unequivocal evidence of distant metastases, either in the cerebrospinal axis or elsewhere in the body. Subsequent experience has shown that the cytologic features of malignancy and the tendency to metastasize are two attributes of malignant neoplasms, which often occur together but not always. Many pituitary neoplasms with disseminated metastases have retained absolutely benign histologic characteristics; others with evidence of anaplasia, pleomorphism, and increased mitotic activity, the cytologic hallmarks of malignancy, have failed to metastasize during the patient's lifetime. Such a paradox is of course not unique with pituitary neoplasms; it has been observed with other endocrine and nonendocrine neoplasms, notably meningiomas. Until we know the molecular basis of the two often associated phenomena dedifferentiation and the tendency to disseminate, we may not understand the basis behind this paradoxical dichotomy. Recent characterization of growth factors secreted by certain human pituitary tumours may shed some light on this topic.

The emerging consensus with regard to classifying pituitary adenomas is as follows: (I) benign-neoplasms that typically are confined to the sella turcica and that, when they grow, do so by compression rather than by infiltration, along the path of least resistance; the general direction of growth is upward into the suprasellar cistern; (2) locally invasive-neoplasms that show restricted local invasion, into the cavernous sinus or into the sphenoid sinus through the floor of the sella; generally these tumours retain the biological characteristics of benign neoplasms; (3) malignant (pituitary carcinomas)-tumours that show evidence of distant metastasis, either within the cerebrospinal axis or elsewhere. Thus, this operational classification rests solely on the biological behaviour of the tumour rather than its histologic characteristics. The mere size of the tumour is, in itself, not a criterion; pituitary tumours can assume enormous proportions yet remain biologically benign. Even so, it is of interest that all pituitary carcinomas reported to date have been macro adenomas with aggressive locally invasive properties. To the best of our knowledge, metastases have not been observed to originate from a microadenoma confined to the pituitary parenchyma, regardless of its functional type.

A qualifying remark has to be made with regard to the tendency for local invasion and the biological behaviour of a tumour. It is well known that local invasiveness alone is not a criterion for malignancy. In fact, Selman and associates found that 85 percent of patients with benign pituitary neoplasms in whom the dura mater was routinely examined showed evidence of microscopic dural invasion. Yet, locally infiltrative tumours that exhibit flagrant violation of anatomic barriers turn out to be malignant. Many examples may be cited, such as tumours that grow downward beyond the sphenoid sinus into the nasopharynx, oral cavity, or nose; laterally well beyond the cavernous sinus along the sphenoid ridge or along the floor of the middle cranial fossa and then through the foramen ovale into the infratemporal fossa; or posterolaterally along the trigeminal ganglion into the cerebellopontine angle. Similarly, although compression of the frontal lobes superiorly and the temporal lobes laterally may be observed with large benign macroadenomas, actual infiltration of the medial temporal lobe, causing partial complex seizures, or infiltration of the hypothalamus, causing diabetes insipidus, proves to be a sign of malignancy.

Incidence

Pituitary carcinomas are relatively rare. It is said that much less than 1 percent of all pituitary neoplasms turn out to be malignant. Thus, many neurosurgeons may not come across a single case during their lifetime. Malignant tumours can be either functional or nonfunctional. Within the functional category, adrenocorticotropic hormone (ACTH)-producing tumours most commonly metastasize; next in frequency are the prolactin-secreting neoplasms; growth-hormone-secreting tumours rarely produce metastases. However, some of the tumours reported as nonfunctional in the earlier literature before the days of prolactin assays might indeed have been prolactinomas; thus, the actual incidence of malignant prolactinomas may be higher than has been reported. It is of interest that ACTH-secreting pituitary carcinoma is reported to occur in the dog.

Routes of Spread

Malignant pituitary neoplasms spread by means of local invasion and through the venous system, the cerebrospinal fluid (CSF) pathways, and the lymphatics. The first three routes are the most common ones. Spread by local invasion has already been alluded to.

Spread through the venous system occurs initially through the cavernous sinus; posterior spread can then occur through the petrosal veins all the way to the jugular vein. Retrograde spread through the cortical draining veins may affect the superior sagittal sinus.

Spread by CSF pathways may involve the supratentorial, infratentorial, or spinal compartments. In the supratentorial space, the tumour may spread to the interpeduncular fossa. Spread along the cortical hemispheric surface may lead to metastases in the frontal, temporal, parietal or occipital lobes. Spread to these. lobes seems to occur more often through CSF pathways than through the bloodstream. Two observed phenomena confirm this belief. First, the lesions tend to be subpial or close to the ventricular surface rather than deep in the white matter; second, patients with intracranial metastases often do not have evidence of systemic spread; if the dissemination were through the bloodstream, one should see metastases elsewhere. Subarachnoid spread in the infratentorial compartment leads to metastases in the cerebellopontine angle, in the ependymal lining of the fourth ventricle, and in the cerebellum.

Spread along the spinal subarachnoid space may lead to metastases in the spinal cord or cauda equina.

The normal pituitary gland does not have any lymphatics. However, lymphatic spread of pituitary tumours can occur after the tumour has invaded the base of the skull.

Systemic spread through the bloodstream has been reported to occur in the lungs. liver, kidneys. heart, bones of the spinal column, scalp, and skull. ACTH-secreting pituitary carcinomas more frequently result in systemic spread, particularly to the liver, compared to other functional carcinomas. The hepatic metastases tend to be small and multiple. The metastatic tumours invariably resemble the primary pituitary tumour in histology, including immunohistochemical characteristics. Secretory granules may be apparent on electron microscopy. The chromogranin A stain is positive even in nonfunctional tumours, and one study 'suggests that serum chromogranin A levels may be elevated in patients harbouring nonfunctioning pituitary tumours.

Clinical Features and Therapy

The evolution of symptoms and their rate of progression depend on the biological behaviour of the tumour. In some cases, the tumour presumably starts as a benign lesion with a long clinical course, but with multiple recurrences and eventually disseminated metastases. In such patients, the initial clinical course is indistinguishable from that of a benign pituitary neoplasm. In other cases, the tumour is probably a carcinoma de novo. In such instances, the course is fulminant, with evidence of rapid local invasion with visual failure, multiple cranial nerve palsies involving the nerves related to the cavernous sinus, appearance of the tumour in the oronasal cavities, and evidence of disseminated metastases in the cerebrospinal axis and elsewhere.

In patients with functional tumours, symptoms and signs of endocrine hyperactivity are superimposed on those related to local invasion and disseminated metastases. Patients with ACTH-secreting pituitary carcinoma tend to do particularly poorly because they suffer gravely not only from the various deleterious manifestations of hypercortisolism but also from local tumour invasion and disseminated metastases. In a similar fashion, patients with growth hormone excess suffer from that condition, as well as from the effects of tumour invasion and/or distant metastases. However, patients with nonsecreting or prolactin-secreting tumours suffer more from tumour invasion and metastases than from hormonal effects.

Facial pain is thought to be a particularly ominous sign in patients with a known pituitary tumour.  However, it does not automatically imply malignancy. Tic-like pain is thought to occur only when the trigeminal ganglion rather than the peripheral divisions of the nerve are involved. Involvement of the ganglion presupposes destruction of the dorsum sellae and invasion through the dura into the ganglion, an event more likely to occur with aggressive tumours. In one instance, the facial pain was shown to be associated with bromocriptine intake.

The onset of diabetes insipidus early in the course of the disease is again an unfavourable sign; it may imply that the hypothalamus is infiltrated rather than just compressed. Patients with benign pituitary neoplasms with suprasellar extension seldom manifest diabetes insipidus preoperatively.

The age of onset of symptoms is no different from that of benign pituitary adenomas. The youngest patient with pituitary carcinoma reported to date was 7.5 years old when the initial symptoms appeared.

The symptoms and signs of central nervous system metastases, of course, depend on the location of the metastases. Those in the cerebral hemispheres produce symptoms pertaining to the appropriate lobe of the brain; cerebellar signs may predominate if the metastasis is to the posterior fossa; spinal metastases may produce paraplegia if they involve the cord or polyradiculopathy if they involve the cauda equina.

There is no evidence to validate the theory that radiation therapy may convert a benign adenoma to a carcinoma. Nor is there evidence to support the notion that surgical intervention promotes the dissemination of the tumour. The constitutive nature of the tumour cell determines how it is to behave.

Patients with Nelson's syndrome tend to have aggressive invasive neoplasms that go on to develop into frank carcinomas. This does not imply that prior adrenalectomy is a prerequisite for the development of an ACTH-producing pituitary carcinoma; several cases of carcinoma of the pituitary have been reported in patients with Cushing's disease from a pituitary tumour who had not undergone adrenalectomy.

In prolactin-secreting carcinomas, the response of the metastases to bromocriptine varies. In some cases, shrinkage of the metastases with normalization of the prolactin level has occurred; in others, metastases have occurred while the patient was on bromocriptine therapy. Even those who respond to bromocriptine initially eventually become resistant to the treatment. After bromocriptine therapy has failed, the use of other dopamine agonists has not conferred any additional benefit. Some reports suggest that positron emission tomography scanning to determine D2 dopamine receptor density and metabolic activity of the tumour using 11C methionine uptake may help to predict and assess the effectiveness of dopamine agonist therapy.

5-Fluorouracil and methotrexate have been tried on occasion to treat disseminated metastases.  However, limited experience precludes any generalization with regard to their efficacy.