Invasion, proliferation, and metastatic ability are three ways in which tumors take root, grow, and advance. Although proliferative tumors can be invasive, and invasion is a prerequisite to metastasis, none of these qualities automatically links to the others in a given tumor.Thus, each should be considered as sufficient, but not necessary, for a tumor to qualify as aggressive.
Invasion
Invasion, proliferation, and metastatic ability are three ways in which tumors take root, grow, and advance. Although proliferative tumors can be invasive, and invasion is a prerequisite to metastasis, none of these qualities automatically links to the others in a given tumor.Thus, each should be considered as sufficient, but not necessary, for a tumor to qualify as aggressive.
Invasion
Invasion means, first and foremost, dural invasion. Because it lines the sides and floor of the sella, the dura sits adjacent to the pituitary gland. Thus, it stands between the pituitary tumor and several of the nearby anatomic compartments into which the tumor may extend, i.e. the sphenoid sinus and cavernous sinuses. The true incidence of dural invasion has been reported as 40% from the Mayo Clinic by gross description, but this figure climbs to 85% with microscopic sampling of dura harvested at surgery. Pituitary tumors, like meningiomas, can invade dura and adjacent bone but even aggressive varieties do not usually invade the brain. ‘Benign’ adenomas with an indolent clinical course can and often do invade, thus invasion alone is not a perfect indicator for malignancy in pituitary tumors.
In addition, microinvasion of normal gland by tumor occurs frequently along the pseudocapsule that most pituitary tumors form around their periphery. Even in benign adenomas, failure to recognize (and resect) such zones of invasion at surgery is one of the leading causes of tumor recurrence over time.Tumors with a greater-than-average chance of invasive behavior or recurrence include those that secrete thyrotropin, large prolactinomas (especially those in males) associated with very high serum prolactin levels, nonfunctional tumors ultrastructurally defined as silent adenoma subtype 3, and silent corticotroph adenomas (which areimmunopositive for adrenocorticotropin, yet do not produce hypercortisolism). Thinned by the intrasellar pressure imposed by a growing tumor, the floor of the sella can ultimately be penetrated, with tumor extension into the sphenoid sinus or less commonly, the clivus. Tumor more often extends into the suprasellar cistern, not by invading but simply by stretching and fenestrating the diaphragma sellae, an arachnoid layer that separates the CSF-containing subarachnoid space above from the intrasellar space below. In some patients a dumbbell-like extension of the pituitary tumor reaches through the small opening in the diaphragma traversed by the pituitary stalk. A truly large pituitary tumor pushing up into third ventricle may ultimately cause hydrocephalus if it reaches the foramina of Munro. The most common result of suprasellar extension is compression of the optic chiasm and/or nerves leading to a loss of the temporal portions of the visual field. This is usually an indenting phenomenon, and tumor pushing into the third ventricle has usually not invaded the floor of that space, but rather has stretched the floor to its breaking point. Occasional patients will ultimately develop invasion of the brain or optic chiasm, but this phenomenon is uncommon and must raise the possibility of true malignancy; and the possibility that the tumor is not a pituitary adenoma at all, but instead a sarcoma or metastasis from an extrapituitary carcinoma arising in breast or lung. Pituitary sarcoma most frequently arises in patients with acromegaly, in whom chronic stimulation by insulinlike growth factor-I is a likely culprit encouraging transformation of an adenoma, or in those who have had sellar irradiation.
Lateral extension into one or both cavernous sinuses may also occur without predilection for any particular pituitary tumor subtype. Although surgeons have long believed that the dura is physically breached by the tumor, recent evidence gathered during endoscopic procedures and cadaveric dissections shows that gaps in the medial wall of the sinus (i.e. in the lateral wall of the sella) can allow an adenoma to enter the sinus without any impediment. As such holes are variably present (in 10% of cadavers examined), it is possible that some tumors showing cavernous sinus entry are indeed invasive, but many are not. It is therefore imprudent to use such entry (which in any case is imprecisely shown by magnetic resonance imaging (MRI) unless it is profound) as a criterion predictive of anaplastic behavior.
Once across the dural wall, the adenoma fills the sinusoidal spaces within the sinus and ultimately can encase the carotid artery and cranial nerves. The cavernous sinus typically expands as its dural boundaries are stretched by tumor. Cranial nerve dysfunction occurs only in the late stages of such invasion. It is, however, possible for cranial nerve malfunction (typically an isolated oculomotor paresis) to occur due to compression of the cavernous sinus rather than to frank invasion. Such compression can be relieved by surgery and allows the preoperative cranial neuropathies to improve over 3–6 months. Complete return to normal extra-ocular motility is not, however, guaranteed. Patients who undergo decompressive transsphenoidal surgery, but have encasement of the cranial nerves by tumor within the cavernous sinus, have much less chance for neurologic improvement. The internal carotid artery usually resists narrowing from tumor compression quite well, but arterial narrowing can occur and may be devastating to patients with poor collateral flow through an anatomically incomplete circle of Willis. Proliferation
The histological hallmarks of malignancy are often absent in patients with invasive pituitary adenomas. Refinements in pathological analysis looking at such factors as TP53 status, apoptosis, epidermal growth factor receptor expression, and in particular proliferation indices all correlate well with the degree of invasiveness and are most pronounced in the frankly metastatic pituitary tumors. Unfortunately, they are no more than loose predictors of the likelihood of recurrence after surgical excision. Much controversy swirls in the pathology literature about the clinical utility of various measures, direct and indirect, of the overall proliferative capacity of a tumor. As oncogene amplification (H-ras, c-myc) and mutation of tumor suppressor genes (TP53, Rb) has been demonstrated much more frequently in pituitary carcinoma than in more benign tumors, such genetic alterations have been proposed as the mechanism underlying tumor progression along the spectrum of malignancy. However, the presence of such alterations does not correlate with recurrence risk in more benign grades.
Methods of assessing proliferation include immunohistochemical demonstration of the per cent of cells expressing Ki-67 antigen (e.g. MIB-1 labeling index) or the proliferating cell nuclear antigen, both expressed in cells that have entered the cell cycle.These measures are imperfect in that such indices show sufficient variability to give some indolent, non-invasive adenomas an index higher than the low index seen in some invasive, anaplastic adenomas; similar overlap occurs between the index range for anaplastic pituitary tumors and pituitary carcinomas.
Several surrogate markers of aggressive behavior have been proposed from a molecular genetic perspective. These include TP53 status, in which mutated gene product shows positive on immunostains. In a study by Thapar et al., positive staining occurred in no benign tumors, 15% of invasive adenomas, and in all carcinomas tested. Activity of telomerase, a reverse transcriptase that extends telomere length and cellular lifespan, correlates with MIB-1 labeling, as does expression of pituitary tumor transforming gene, but it too shows much variability.
The drawback to all these approaches to the problem of predicting future proliferation (and recurrence) is the exclusive attention paid to proliferation while ignoring on-going cell death (i.e. apoptosis and autophagy). Both coexist, and the ability of any tumor to grow depends on the balance struck between the two opposing processes. Although indices for proliferation are well established, those measuring cell death are not. One study compared proliferative and apoptotic indices in tumors with higher growth rates by MRI vs those with lower growth rates. Inability to detect a difference between these groups may well reflect insufficient measuring tools, rather than a failure of the concept. Metastasis
Metastases from pituitary tumors are the hallmark of the true pituitary carcinoma.They are quite rare, with about 130 cases reported, but the incidence is likely higher than this limited number would imply, since pituitary tumor patients do not routinely undergo a staging work-up or radionuclide imaging that could detect systemic metastasis, nor is every identified case reported. Although the majority of such tumors originate from clinically nonfunctional adenomas, all functional adenoma subtypes have been reported to produce pituitary carcinomas. In half of such patients, the tumor disseminates within the neuraxis and most of these (although not all) show cytological features of malignancy, including pleiomorphism, nuclear atypia, and mitotic figures. Occasional patients will show leptomeningeal dissemination with positive cytology for malignant cells within spinal fluid. Half of patients with metastases to extraneural sites have an ACTH-secreting tumor. Half of pituitary carcinomas metastasize to the liver with lesser proportions metastasizing to bone, lung, and/or lymph nodes. In the overall statistics, pituitary carcinomas are nonfunctional half the time and secrete ACTH in 22%, growth hormone in 13%, and prolactin in 11%. Although nonfunctional adenomas are thus somewhat over-represented, given that immunohistochemical staining was not performed on all reported cases it is possible that some are actually gonadotropinsecreting tumors.
Pituitary carcinoma can show either an indolent or an aggressive course, and survival ranges from one month to ten years or more. Most start as histologically benign adenomas that initially grow slowly, but progress over years to the more invasive variety. Because by definition, metastasis requires invasion of blood vessels by tumor, non-invasive adenomas cannot metastasize. As mentioned above, radiotherapy does cause sarcomatous change in a few patients with benign adenomas (many of whom have growth hormone-secreting tumors, which seem to be more prone to such transformation), but more than half of patients with pituitary carcinomas have had no previous sellar exposure to irradiation, so its role is a transforming agent remains unclear.
Treatment protocols are difficult to devise in any consistent way for these rare neoplasms. Even aggressive treatment confers relatively short survival. A classic oncologic approach can rarely be applied to pituitary tumors due to anatomical constraints. Such an approach would by definition include radical resection of the primary site and of surgically accessible metastatic sites, followed by radiotherapy and then cytotoxic chemotherapy. No chemotherapy protocols have been devised, and treatment has typically been conducted on an ad hoc basis with occasional patients showing response to chemotherapy which is usually temporary and which does not persist after treatment stops. In one series lomustine and 5-fluorouracil were given to seven patients with aggressive tumors, four of whom had frank carcinoma. Temporary shrinkage of tumor occurred, but all died of their disease from 3–65 months after starting treatment. Our practice is to use sarcoma regimens (often including isofosfamide) for such lesions, but we too have had limited success in controlling their growth. Radiotherapy is also very commonly used as an adjunct to incomplete surgical excision, but even stereotactic radiosurgery with its high dose per fraction shows poor success in controlling these tenacious tumors. Although the development of skull base approaches over the past 20 years had made surgery feasible within the cavernous sinus, it is not free of risk nor is it typically curative unless the morbid and difficult procedure of cavernous sinus exenteration is undertaken.We have attempted to use such approaches in select patients with aggressive intracavernous disease that is expanding actively and threatening the integrity of the optic apparatus.
Both radical cavernous sinus exeneration and partial cleanout of the cavernous sinus with preservation of neurovascular structures have been tried. In frankly malignant tumors, neither approach has ultimately succeeded although some palliation has been obtained thereby. Invasive but not markedly proliferative tumors can usually be debulked with a reasonable expectation for continued indolence by the tumor, and thus an extended life expectancy of reasonable quality. In general, given the dismal natural history of this pituitary carcinoma, patients with this most malignant variety of pituitary tumor should undergo as aggressive a resection as anatomical and technical factors permit. Metastatic sites inside or outside the neuraxis are usually treated with local radiotherapy.
