Prolactinomas may be defined as pituitary tumors that autonomously secrete prolactin, a polypeptide hormone.  An outline of the diagnostic evaluation is presented, and the current status of medical, surgical, and radiation treatment is evaluated. The management of prolactinomas is emphasized, as it has undergone significant revisions over the course of the last decade. These changes represent the clinical application of an enhanced understanding of the pathophysiology of prolactin secretion and its systemic effects, advances made in neuroendocrinology, and a heightened recognition of the pathologic changes associated with hyperprolactinemia. In addition, the introduction of sophisticated neuroradiologic diagnostic techniques and a variety of new therapeutic options has had a profound impact on how prolactinomas are diagnosed and treated. Given the relatively short period over which these data have been introduced into the literature, it is not surprising that the management of prolactinomas is controversial.

Physiology

Prolactin is produced and secreted by an erythrosinophilic subtype of acidophilic cells in the adenohypophysis, under the dual control of hypothalamic neurohormones that act as releasing and inhibiting factors. The hypothalamic modulation of prolactin secretion, unlike that of other anterior pituitary hormones, is exerted mainly through inhibition. Prolactin is tonically released from the pituitary gland. A prolactin inhibitory factor (PIF) is produced in the hypothalamus and released into the portal venous system. There is strong evidence to suggest that dopamine is the naturally occurring PIF that acts at the D2 receptors in the pituitary gland. PIF prevents unrestrained release of prolactin by the adenohypophysis. Sectioning of the pituitary stalk and the presence of certain hypothalamic lesions that interfere with the normal delivery of PIF to the anterior lobe increase serum prolactin levels. The prolactin inhibitory mechanism is mediated by the hypothalamic catecholamines dopamine and norepinephrine, and it is blocked by dopaminergic blocking agents such as phenothiazines, tricyclic antide­pressants, methyldopa, and reserpine. A prolactin-releasing factor (PRF), possibly regulated by serotonin, has been identified in hypothalamic extracts.

The physiologic effects of prolactin are not completely understood, but it is believed to act in concert with several other hormones, including estrogen, progestins, oxytocin, corticosteroids, growth hormone, and insulin.

Normal physiologic stimuli that result in prolactin release include exercise, stress, sleep, pregnancy, and nipple stimulation. In humans, prolactin is essential for stimulation of breast tissue growth and for initiation and maintenance of lactation, provided that glandular breast tissue has been appropriately primed by the actions of several other hormones, including estrogen, progestins, corticosteroids, growth hormone, and insulin. Small amounts of prolactin are also necessary for progesterone production by granulosa cells; hyperprolactinemia, however, inhibits progesterone production. Prolactin secretion increases steadily during pregnancy and peaks at parturition. During the postpartum period, plasma prolactin rapidly declines to normal levels. Suckling produces a brisk rise in prolactin secretion and may be involved in the "milk-ejection reflex." This rise in prolactin level is blunted after three months of nursing. Although the function of prolactin in males is unclear, the hormone seems to be necessary for normal sperm production. Hyperprolactinemia has been shown to inhibit 5-α-reductase, which converts inactive testosterone to the biologically active dihydrotestosterone, a hormone that must be present in high concentrations in the testicular tubules for spermatogenesis to occur.

A variety of hormones and endocrine disorders affect serum prolactin levels. Prolactin secretion is increased by the administration of estrogens. Patients with primary hypothyroidism may have hyperprolactinemia stemming from an increased responsiveness to the prolactin-releasing activity of thyrotropin-releasing hormone (TRH).

Hyperprolactinemia has an adverse effect on the skeleton. In 1980, Klibanski and colleagues drew attention to the possibility that hyperprolactinemia may contribute to osteoporosis. They described decreased bone density in 14 hyperprolactinemic women, which correlated with the relative or absolute estrogen deficiency that may accompany hyperprolactinemia. These findings are supported by the research of Cann and colleagues, who found decreased spinal mineralization in amenorrheic women.  Greenspan and colleagues found an analogous relationship between osteoporosis and hyperprolactinemic hypogonadism in males. The report of an osteoporotic fracture in a middle-aged man with a prolactinoma lends support to this association. Sartorio and colleagues found that the concentration of osteocalcin, a specific marker of bone formation. was significantly lower in 29 patients with microprolactinomas than in the control population. These authors also noted an inverse relationship between osteocalcin and prolactin levels once therapy was instituted. The effects of hyperprolactinemia are reversible once therapy directed at normalizing the serum prolactin level is instituted.

Pathology

As with other pituitary tumors, prolactinomas may be classified according to gross size into microadenomas (<10 mm in greatest diameter) and macroadenomas (> 10 mm in greatest diameter). Tumors that invade the cavernous sinus, and thus are not readily amenable to total surgical excision, are referred to as invasive tumors.

The traditional microscopic classification of pituitary tumors on the basis of their staining characteristics with hematoxylin and eosin as chromophobic, acidophilic, basophilic, and mixed adenomas has little functional significance. Over the past several years, the introduction of radioimmunoassay techniques, electron microscopy, and immunocytochemistry has permitted closer study of the pathology of pituitary adenomas.

Immunohistochemical classification has revealed that there are two types of eosinophilic adenohypophyseal cells: lactotrophs and somatotrophs. The former normally secrete prolactin, and the latter, growth hormone. Eosinophilic adenoma cells may secrete either or both, or may be functionally inactive. Most prolactinomas are composed of lactotrophs. Depending on the density of the secretory granules, prolactinomas may be chromophobic (sparsely granulated) or eosinophilic (densely granulated).

Electron microscopic studies reveal characteristic ultrastructural features in prolactinoma cells, including nebenkerns, which are bodies composed of whorls of rough endoplasmic reticulum, and misplaced exocytosis (granule secretion into the intercellular space). Similar changes are also seen in late pregnancy and with estrogen therapy.

Hyperprolactinemia may occur in association with a variety of pituitary adenomas. Kovacs and Horvath found that mixed growth hormone/prolactin adenomas-tumors composed of both neoplastic somatotroph and lactotroph cell lines-occurred in 7 percent of cases. Approximately 40 percent of patients with acromegaly have an elevated prolactin level, which in most cases is due to the stalk effect. Hyperprolactinemia has also been attributed to hyperplasia of the prolactin-producing cells of the anterior pituitary, in the absence of an actual adenoma.

Clinical Presentation

Prolactinomas are the most common functional pituitary tumour encountered in humans, accounting for approximately 25 percent of all pituitary tumors and 3 percent of all intracranial tumors. The incidence of prolactinomas in the general population is unknown. In a study undertaken to determine the incidence of hyperprolactinemia and prolactinomas, Miyake and colleagues screened 4803 men and estimated the incidence of prolactinomas in males to be approximately 1: 1600.

The aetiology of prolactinomas remains unknown. It is of interest that there has been a definite increase in the number of pituitary adenomas in women of childbearing age, while the incidence in men and older women seems to have remained stable. Whether this increase is due to heightened recognition of the tumour or to exposure to as yet unidentified etiologic factors is not known. Although estrogen administration has been associated with an increased incidence of pituitary adenomas in laboratory animals, the role of oral contraceptives in the pathogenesis of prolactinomas in humans is unknown. It seems unlikely that the low dosage of estrogen contained in birth control pills would stimulate the development of a pituitary adenoma. Certainly, no clear relationship has yet been identified between the intake of oral contraceptives and the occurrence of prolactinomas, despite widespread use of these drugs. However, a case report of the development of an invasive macroprolactinoma after prolonged estrogen replacement therapy, although anecdotal, would hint that a relationship may exist.

No long-term studies have defined the natural history of prolactinomas. However, a few small studies with limited follow-up suggest that the progression of untreated prolactinomas is both slow and unpredictable. In a series of 38 patients with microprolactinomas followed with serial CT studies for an average of 31.7 months, Sisam and colleagues did not demonstrate significant tumour growth in any patient. However, Weiss and colleagues, in a series of 27 similar patients followed for 6 years, noted evidence of tumour progression in 3 patients (10 percent). These findings are consistent with a recent prospective report in which 30 untreated women with hyperprolactinemia were monitored for 3 to 7 years. In this study, serum prolactin levels increased in 6 of the women, decreased in 10, and did not change in the other 14.

Empirical clinical experience has indicated that untreated macroprolactinomas continue to enlarge and cause compressive as well as endocrinologic symptoms. The symptoms related to microprolactinomas are restricted to the effects of hyperprolactinemia. However, small, incidental pituitary adenomas are found frequently during postmortem examination in patients who had no recognizable endocrinopathy during life.

Prolactin-secreting tumors may present with a mass effect and/ or a hyperprolactinemic endocrinopathy. The mass effect may cause compression of the adjacent hypothalamus or normal pituitary tissue, resulting in hypothalamic dysfunction or hypopituitarism, respectively. Other symptoms resulting from the mass effect include impairment of visual acuity and field defects (typically bitemporal hemianopia), headaches, and, more rarely, third, fourth, and sixth cranial nerve palsies and hydrocephalus.

Hyperprolactinemia suppresses the hypothalamic-pituitary­gonadal axis, which may result in gonadal dysfunction in both sexes. Prolactinomas are a well-recognized albeit infrequent cause of arrested puberty. The typical clinical presentation in females is the amenorrhea-galactorrhea (AG) syndrome (Forbes-Albright syndrome). Spontaneous galactorrhea occurs in about 30 percent of women and less frequently in men with prolactinomas. It is not known why only some patients develop galactorrhea, but its occurrence does not seem to be related to serum prolactin levels. Furthermore, hyperprolactinemia in females may suppress and/or interfere with the menstrual cycle and result in primary or secondary amenorrhea and infertility.

Although it is now well recognized that hyperprolactinemia­induced hypogonadism has a deleterious effect on the skeleton in both sexes, the epidemiologic and overall clinical significance of this effect is not known. Osteoporosis and its sequelae are rare presenting clinical features in patients with prolactinomas. However, with a heightened awareness of these changes, further studies may indicate that the true prevalence of osteoporosis is higher than is currently thought. It is noteworthy that this effect can be reversed once therapy directed at normalizing the serum prolactin level is instituted.

Men with prolactinomas may present with decreased libido, impotence, and oligospermia. Prolactinomas tend to be larger in men than in women at the time of clinical presentation. This may be related to the fact that hyperprolactinemia in females is more clinically overt than in males. The clinical picture in males is typically dominated by symptoms of a para- or suprasellar mass effect rather than by hyperprolactinemia.

The endocrinopathy associated with prolactinomas is due to the hyperprolactinemia and is symptomatically identical to that resulting from any other cause of elevated serum prolactin levels. Although several different conditions can cause hyperprolactinemia, the three most commonly identified are ingestion of certain drugs, particularly phenothiazines; primary hypothyroidism; and a pituitary tumour. Thus, in the diagnostic evaluation of patients with hyperprolactinemia due to a suspected pituitary tumour, it is essential to obtain a detailed history and to perform the appropriate tests [e.g., magnetic resonance imaging (MRI) and evaluation of thyroid function] to identify the cause of the elevated serum prolactin value.

Diagnostic Evaluation

Diagnostic evaluation for a suspected prolactinoma is divided into endocrine testing and neuroimaging studies.

  Endocrine Testing

Endocrine studies include baseline pituitary target organ tests and at least one determination (preferably two) of the fasting serum prolactin level. The pituitary target organ tests fulfil two purposes. First, they determine the presence and extent of pretreatment pituitary endocrine dysfunction, and second, they provide a yardstick by which treatment efficacy and endocrinologic complications can be measured.

It is frequently difficult to identify the cause of hyperprolactin­emia in patients with pituitary tumors who have a moderately ele­vated serum prolactin value (i.e., of 60 to 120 ng/ml). Such modest elevations in serum prolactin may result either from compression of the pituitary stalk and/or hypothalamus or from autonomous secretion by the tumour. Furthermore, lesions such as an intrasellar craniopharyngioma, an aneurysm, or a nonfunctional pituitary adenoma can cause a modest hyperprolactinemia. The mechanism by which these sellar and parasellar lesions cause modest elevations of serum prolactin is due to stalk compression which, in turn, interferes with the normal delivery of PIF to the anterior pituitary lobe; this is referred to as the stalk effect.

A fasting level of serum prolactin over 150 ng/ml usually indicates that the cause of the hyperprolactinemia is a pituitary adenoma. Very high serum prolactin levels (> 1000 ng/ml) signify invasiveness, which usually means that the tumour has extended into the cavernous sinus.

Provocative tests have been used to determine prolactin reserve both in prolactin deficiency states and in cases of autonomous prolactin release. Stimulation with TRH is the most efficient test for determining prolactin reserve. Baseline fasting thyroid-stimulating hormone (TSH) and prolactin levels are determined, and 500 ng of TRH is given intravenously over 30 s. The serum prolactin and TSH levels are determined at 15,30,45, and 60 min, with the patient in the supine position to prevent orthostatic hypotension. A less than twofold increase in the level of prolactin or TSH is indicative of loss of pituitary reserve.

Chlorpromazine (CPZ), a dopamine antagonist that inhibits PIF, can also be used to assess prolactin reserve. The CPZ test is performed by determining fasting serum prolactin levels before and 60, 120, and 180 min after an intramuscular injection of 50 mg CPZ. A less than twofold increase of prolactin levels from the baseline after the injection of CPZ is considered to show loss of reserve if the baseline prolactin value is low, and loss of hypothalamic control if the baseline level is high.

Although the prolactin response to provocative tests such as the TRH and CPZ tests has been useful in identifying structural lesions in the vicinity of the pituitary stalk and gland, these tests have not served to differentiate between the mechanisms that can be responsible for elevated prolactin levels in patients (i.e., pituitary tumour versus other causes of hyperprolactinemia). The response (in the serum prolactin level) to provocative tests is blunted or absent in all such situations. Failure of prolactin to rise to two to three times the baseline value is characteristic of virtually all pituitary tumors or other space-occupying lesions of the sella, whether their mechanism of action is related to autonomous secretion or to disruption of the PIF pathways.

These provocative tests are unnecessary when prolactin levels are> 150 ng/ml or when a tumour is identified on either computed tomography (CT) or MRI, as the diagnosis is already confirmed and the tests would add little to the investigation. When the prolactin levels are mildly elevated and the neuroimaging studies are equivocal or nondiagnostic, the provocative tests may be useful. In these instances, the tests differentiate between elevated prolactin levels due to abnormal prolactin secretion and normal variations.

  Neuroimaging Studies

Neuroimaging studies have evolved significantly in recent years. The imaging studies that have had the greatest impact in this area are high-resolution CT scanning and MRI. These tests not only facilitate the diagnosis, but have proved useful in planning and executing surgery and in detecting tumour recurrence postoperatively.

CT examinations of the sella are performed with the neck hyperextended to achieve an imaging plane that is perpendicular to the sella floor in a direct coronal plane. This minimizes dental artefacts. Thin collimation (1.5 to 2.0 mm) is required. A contrast agent is administered routinely and is excluded only in cases with suspected acute haemorrhage. Prolactin-secreting microadenomas are identified as discrete, focal, hypodense lesions; however, high­resolution CT studies may lack the sensitivity and specificity to confirm or exclude a prolactin-secreting microadenoma. In a retrospective study of 51 patients with prolactin-secreting microadenomas, Davis and colleagues found that 6 patients had a normal CT study. Typically, macroadenomas are isodense to the normal gland or inhomogeneous with mixed iso- or hypodense areas. There are no CT radiographic features that distinguish prolactinomas from other pituitary tumors.

MRI studies of the sellar and parasellar region require thin slices «2.5 mm) and a high field strength scanner (at least 0.35 T), and are routinely performed both without and with gadolinium enhancement. T1-weighted images performed with a short TE, short TR spin-echo sequence provide excellent anatomic detail of the sella, optic chiasm, and cavernous sinus. Most prolactinomas are isointense to hypointense to the normal gland and cerebral cortex on the T1-weighted sequence and are variable in intensity with T2 weighting. In an MRI study of 115 patients with pituitary macroadenomas, Lundin and colleagues found that diffuse invasion of the base of the skull was most common in prolactinomas and was associated with a lower frequency of suprasellar extension.

Cerebral angiography is rarely indicated in the neurodiagnostic evaluation of prolactinomas. However, when an intracranial aneurysm with intrasellar extension cannot be ruled out by either CT or MRI, angiography may be considered.

  Management

There is some controversy over the optimal management of prolactinomas. For instance, questions remain regarding both the extent and the form of the most appropriate diagnostic endocrinologic and radiologic evaluations, the role of irradiation, and the management of residual and recurrent prolactinomas. This uncertainty focuses on whether medical or surgical therapy (i.e., bromocriptine administration or transsphenoidal microsurgery) is the treatment of choice for prolactinomas. Although both have been shown to provide efficacious control or cure, the precise indications, contraindications, and potential benefits of these treatment options await clearer definition.

Prolactinomas can be managed conservatively, medically, surgically, or by irradiation. The most appropriate mode of therapy depends on a number of factors, including the size of the tumour; the level of the serum prolactin; the patient's age, overall health, and associated surgical risk factors; the patient's tolerance of or compliance with medical therapy; and whether the patient desires fertility.

  Conservative Therapy

There are exceptional situations in which it may be appropriate not to institute any definitive therapy for a prolactinoma, for instance in a young woman with a microprolactinoma and modest hyperprolactinemia (e.g., 60 to 90 ng/ml) who does not desire pregnancy. This management plan is only acceptable if osteoporosis, with the attendant increased risk of fracture, is not present. These patients should be followed with periodic clinical, MRI, and prolactin studies as well as bone densitometry. In all other cases, we believe that a more aggressive management plan is warranted. Hyperprolactinemic hypogonadism per se is an indication for therapy, even when the risk of a mass effect from a microadenoma is minimal and the patient does not request fertility or improved sexual function.

  Medical Therapy

The introduction of effective medical treatments has offered practical alternatives to surgical treatment. The ergot derivative bromocriptine is the current standard pharmacotherapy for prolactinomas.

Bromocriptine, a dopamine agonist, is a potent inhibitor of the synthesis of prolactin messenger RNA as well as the release of prolactin by the pituitary gland. It has been shown that bromocriptine causes significant cellular shrinkage in prolactinomas, owing to a decrease in protein synthetic activity and to a reduction in the size of related intracellular organelles. This reduction in size is reversible; the tumors will return to their pretreatment size within days of bromocriptine withdrawal. This observation implies that the reduction in bulk in prolactinomas treated with bromocriptine is mainly due to a reduction in cell size. However, Mori et al. have shown that bromocriptine also has a definite although quantitatively less significant cytocidal effect on human prolactinoma cells.

Most patients tolerate bromocriptine well. The most common side effects are nausea, nasal congestion, dizziness, hypotension, syncope, and sedation. Other complications include psychosis, headaches, leukopenia, thrombocytopenia, vasoconstriction, abdominal cramps, urinary incontinence, and cerebrospinal fluid rhinorrhea.

Not all prolactinomas are responsive to bromocriptine. Using cultures of prolactinoma cells, Pellegrini and colleagues noted differences in the density of dopaminergic binding sites and inhibition of adenylate cyclase activity between cells derived from bromocriptine-responsive and bromocriptine-resistant tumors. Further investigation by the same group found differences in response to bromocriptine that may be explained by the existence of either receptor or postreceptor defects in prolactinomas. This suggestion finds support in the pathologic study by Mori and colleagues, who noted two distinct histologic appearances in human prolactin-secreting adenomas treated with bromocriptine. As yet, there is no reliable method to identify those patients who will respond to bromocriptine.

Several new dopamine agonists have been introduced, all of which have chemistry and side effects similar to those of bromocriptine. Pergolide and cabergoline have the advantage of an extended plasma half-life, which allows once daily and once weekly therapy, respectively. A new long-acting nonergot dopamine agonist, CV 205-502, has been shown in preliminary studies to be as effective as bromocriptine and to have fewer side effects. In addition, CV 205-502 may be effective in cases of prolactinoma that are unresponsive to bromocriptine treatment. However, long­term assessment of this drug is required before its role in the management of prolactinomas is established.

Enthusiasm over the initial results with bromocriptine therapy of prolactinomas led some clinicians to recommend that the drug be used as primary treatment of these neoplasms, reserving surgery only for therapeutic failures. We believe this approach to be unsound, for the following reasons: (1) indications for surgical treatment as first-line therapy exist; (2) approximately 10 percent of prolactinomas fail to respond to bromocriptine; (3) there is a subset of patients who cannot tolerate bromocriptine; (4) bromocriptine therapy must usually be life­long, and patient compliance may vacillate; and (5) the results of surgery in appropriately selected patients are comparable to those of pharmacotherapy.

Despite these reservations, we do recommend bromocriptine for many patients with prolactinomas. Patients who have a serum prolactin level between 150 and 500 ng/ml and whose tumors are noninvasive are candidates for bromocriptine. This group of patients are also excellent candidates for transsphenoidal surgery. Women with microprolactinomas who have a modestly elevated serum prolactin level < 150 ng/ml and who desire pregnancy are considered candidates for bromocriptine therapy. In these cases, the primary goal of therapy is to reduce the level of serum prolactin so that a normal menstrual/ovulatory cycle is achieved, which will usually allow conception to occur. The drug is discontinued once the patient becomes pregnant. The potential teratogenic effects of bromocriptine outweigh the small risk of a mass effect due to enlargement of the gland or microtumor during pregnancy. It is for this reason that bromocriptine as a sole therapy is contraindicated in patients with macroprolactinomas who desire pregnancy. We also advocate the use of bromocriptine in patients with a prolactinoma who have very high levels of serum prolactin (> 1000 ng/ml). The surgical cure rate with these locally invasive tumors is poor. In cases of persistent or recurrent hyperprolactinemia, bromocriptine is usually given following surgery.

Although bromocriptine is not tumoricidal, its ability to dramatically reduce the size of a prolactinoma makes it a potentially useful preoperative adjunct. However, Landolt and colleagues found that chronic preoperative treatment with bromocriptine adversely affected the outcome of transsphenoidal surgery. They speculated that drug-induced fibrosis impeded a complete surgical removal of the tumour. However, in a series of 40 patients with prolactinomas, 20 of whom were treated with bromocriptine, Perrin and colleagues found that preoperative bromocriptine treatment neither increased fibrosis of the gland nor adversely affected surgical outcome.

Also to be considered in macroprolactinomas is the possibility that preoperative treatment with bromocriptine will cause enough reduction in tumour size that subsequent management by surgery will provide a better chance of achieving a cure. Perrin and colleagues reviewed the records of 40 patients with prolactinomas, 20 treated preoperatively with bromocriptine and 20 not so treated. They found that the surgical cure rate of the bromocriptine-treated group was higher than that of the control group, both for microprolactinomas (87.5 percent versus 50 percent) and for macroprolactinomas (33 percent versus 17 percent). Their study was retrospective and involved a relatively small group of patients. The experience of Weiss and colleagues with 19 patients suggests that preoperative tumour shrinkage is therapeutically beneficial. They observed that patients whose tumors responded favourably to preoperative pharmacologic manipulation had a better surgical cure rate.

One uncommon clinical situation that may present management problems and deserves special mention is a prolactinoma that presents in a pregnant patient. We recommend that medical or surgical treatment be withheld until the end of the pregnancy if the tumour is a microadenoma without any evidence of mass effect. These patients should be followed closely to check for evidence of significant tumour enlargement during the pregnancy, which may occur as a result of physiologic pituitary enlargement during pregnancy, tumour expansion, or both. However, if the pregnant patient presents with a macroadenoma, or if there are mass-effect symptoms such as visual loss, transsphenoidal microsurgical removal of the tumour is recommended.

Surgical Treatment

Surgery is an effective treatment for prolactinomas. Transsphenoidal microsurgery, the current method of choice, is reliable, safe, and effective. A number of investigators have reported cure rates of prolactinomas following transsphenoidal surgery on the order of 70 percent.

  Criteria for Cure

Patients who remain asymptomatic, have a normal serum prolactin level, and, more important, have negative MRI studies for a period of at least 5 years, may be considered to have been cured by the surgical procedure. Favourable indicators of a cure include the following:

1. Normalization of serum prolactin values to <25 ng/ml. This finding indicates that all hypersecreting tumour tissue was removed by the surgery. If the value remains normal for more than 5 years, one can assume that a cure was achieved, with little chance of tumour recurrence. Persistent hyperprolactinemia following transsphenoidal resection of a prolactinoma may be due to persistent tumour or to damage to the pituitary stalk by the tumour or the surgery, or both. When the postoperative value remains < 100 ng/ml and shows no tendency to increase, the persistent hyperprolactinemia usually reflects stalk damage.

2. Cessation of galactorrhea and resumption of normal menstrual periods. These clinical indicators of cure nearly always coincide with normalization of serum prolactin values. In exceptional cases, menstrual periods return and galactorrhea ceases even with levels of prolactin slightly above normal. In other cases, these symptoms may continue despite normalization of serum prolactin levels.

3. Observations made by the surgeon. In the case of a moderate sized or small tumour confined to the sella, an experienced neurosurgeon can usually determine whether gross total removal of tumour has been accomplished.

  Indications for Surgery

The indications for surgery include (1) a patient with a noninvasive prolactinoma (by MRI) and a serum prolactin level between 150 and 500 ng/ml, who does not wish to take long-term medication; (2) a woman with a noninvasive tumour (by MRI) and a prolactin level <500 ng/ml, who desired pregnancy; and (3) a woman with a macroprolactinoma who desires pregnancy. The surgical goal in this case is to eliminate the risk of tumour enlargement and mass effect during pregnancy. Bromocriptine may be used until conception occurs to normalize the serum prolactin level in those cases in which this goal was not achieved, and can then be discontinued during pregnancy with a reduced risk of mass effect from tumour enlargement. Other indications for surgery are as follows. Transsphenoidal surgery may be used as a preliminary debulking operation to enhance the effect of bromocriptine in cases of macroprolactinomas with serum prolactin levels >500 ng/ml, with or without evidence of invasiveness (the smaller lesion should allow an easier control of residual tumour with bromocriptine). The rare prolactinoma that causes pituitary apoplexy should be regarded as a surgical emergency. Surgery should be used for a prolactinoma in a pregnant patient presenting with a mass effect, especially one that is causing progressive visual loss. Transsphenoidal pituitary adenomectomy is also indicated in patients with prolactinomas that do not respond to primary bromocriptine treatment. Recurrent prolactinomas can usually be treated by surgery followed by medical and/or radiation therapy.

  Factors Influencing Surgical Results

Tumour size, preoperative level of serum prolactin, and presence of extrasellar tumour extension have been identified as prognostic factors that influence the outcome of transsphenoidal removal of prolactinomas. A number of studies have shown that the best results are obtained in patients with microadenomas. In the series of Hardy and colleagues, serum prolactin levels were normal following transsphenoidal surgery in 90 percent of patients with localized microadenomas, in 53 percent of those with enclosed adenomas, and in 43 percent of those with invasive adenomas. Similar results were obtained by Chang and colleagues, who reported return of menses in 16 of 17 women with prolactin-secreting microadenomas and in 2 of 7 with macroadenomas.

The degree of hyperprolactinemia is of more prognostic significance than tumour size in predicting the outcome of surgery. The influence of the serum prolactin level on cure rates has been investigated in a number of studies. Hardy and colleagues showed that transsphenoidal surgery for prolactinomas was followed by normalization of prolactin levels in 59 of 80 cases (74 percent). In this series, menses returned in 63 percent (50 cases), and a normal pregnancy ensued in 36 percent (29 cases). Faria and Tindal reported normalization of prolactin levels following transsphenoidal surgery in 55 of 72 (76 percent) patients whose preoperative prolactin level was <200 ng/ml and in only 13 of 28 (46 percent) patients whose preoperative prolactin level was >200 ng/ml. Domingue and colleagues defined therapeutic failure as unsuccessful resolution of amenorrhea within a follow-up period of at least 18 months. Therapeutic failure was encountered in 32 percent (29 of 91) of their patients with prolactinomas. They found that therapeutic failure occurred in patients who had higher preoperative serum prolactin levels (>200 ng/ml) and in those in whom total tumour removal was not achieved.

These and other studies have demonstrated an inverse relationship between cure rate and the preoperative level of serum prolactin. On average, the cure rate with serum prolactin levels between 200 and 500 ng/ml ranges between 48 and 68 percent. The cure rate drops to between 2 and 36 percent with prolactin levels between 500 and 1000 ng/ml, and to only 6 to 22 percent with preoperative prolactin levels> 1000 ng/ml.

In summary, patients with microadenomas are more likely to have preoperative prolactin levels <200 ng/ml, and it is in this group that transsphenoidal surgery achieves the best results, with a cure rate on the order of 75 percent. Patients with macroadenomas usually have preoperative prolactin levels >200 ng/ml, and the cure rate in this group falls progressively as the levels of prolactin increase.

  Type of Surgical Approach

The low operative morbidity and mortality rates and the ability to remove the tumour and spare the normal pituitary gland in most patients are the primary advantages of modern transsphenoidal microsurgery. It is recommended in virtually all patients with a prolactinoma in whom surgery is indicated. The exceptions to this surgical approach are: (1) significant extrasellar extension of the tumour into the anterior and/or middle cranial fossa, and (2) a suprasellar tumour with an unusual dumbbell configuration, in which the tumour extends through a relatively small opening in the diaphragma sellae and then expands into a larger suprasellar component. In these situations, the lesion is best exposed and managed by craniotomy or by a combined transsphenoidal-craniotomy approach.

We do not believe that a conchal type of sphenoid sinus-that is, a sphenoid sinus with little or no pneumatization -is a contraindication to the transsphenoidal approach. The bond can be removed safely with the use of an air-driven, high-speed, angled drill. In these situations, intraoperative skull films or televised fluoroscopy is used for localization and for monitoring the progress of the drilling.

  Surgical Technique

The details of transsphenoidal microsurgery, including the exposure along the nasal septum, entry into the sphenoid sinus and floor of the sella turcica, and method of tumour removal, are variable. However, a few technical points warrant specific attention. Regardless of whether a sublabial or endonasal incision is used, it is important that the surgeon develop a plane along one side of the nasal septum, allowing this structure to be spared. This minimizes the risk of creating a post­operative nasal deformity and, more important, reduces the technical difficulties of a future transsphenoidal operation should one be necessary. It is equally important not to damage the pituitary gland during opening of the dura. The dura mater can be opened safely by making a slow, deliberate incision with a # 11 blade on a bayonet knife holder under relatively high magnification. Once an adequate opening is made, it can be enlarged appropriately with fine angled microscissors. Because many tumors associated with abnormalities in prolactin secretion are < 10 mm in diameter, lacerations in the gland with attendant subcapsular bleeding make it difficult if not impossible to detect the subtle differences between the normal gland and a small tumour. This differentiation is a prerequisite to a successful removal of the tumour and sparing of the gland.

Prolactin microadenomas are usually situated laterally in the gland; larger pituitary tumors usually erode the floor of the sella turcica, and in these cases the tumour commonly extrudes into the operative wound upon removal of the floor of the sella and opening of the dura mater. Frequently, the pituitary gland is compressed and flattened against the dorsum sellae or diaphragma sellae by the tumour.

  Surgical Morbidity and Mortality

The relatively low incidence of complications following transsphenoidal surgery is one of the appealing features that has popularized this procedure. The complication rate from large series of transsphenoidal procedures is approximately 4 percent. The most common complications are cerebrospinal fluid leakage, hypopituitarism, and diabetes insipidus.

Recent series report an incidence of postoperative cerebrospinal fluid rhinorrhea of approximately 3 percent. It occurs more often if there is a congenitally large diaphragmatic aperture, a large tumour volume, or prior surgery or irradiation, or if the diaphragma sellae is disrupted either iatrogenically or by tumour invasion. A disruption of the diaphragma may not always be apparent intraoperatively, or even in the immediate postoperative period. Postoperative downward herniation of the arachnoid into the evacuated sella, with associated incompetence of the diaphragma, may occur years later. Initially, the treatment of postoperative cerebrospinal fluid rhinorrhea consists of serial lumbar punctures to lower the cerebrospinal fluid pressure to <5 cmH2O. If the cerebrospinal fluid leak persists, an indwelling spinal subarachnoid catheter is inserted and left in situ for 3 days. The attached sterile reservoir is placed at the height of the lumbar puncture site. Should this fail to halt the cerebrospinal fluid leak, the transsphenoidal wound is reopened, and the sella and sphenoid sinus repacked with adipose tissue and fibrin glue. Possible sequelae of cerebrospinal fluid rhinorrhea include meningitis and tension pneumocephalus.

The potential effect of transsphenoidal surgery on pituitary endocrine function is an important consideration, especially if surgery is aimed at restoring fertility. McLanahan and colleagues performed complete endocrine tests before and 10 days after transsphenoidal surgery in 97 women with prolactinomas, (72 percent) of whom harboured microadenomas. Of 65 patients who had normal pituitary endocrine function preoperatively, 50 had normal function postoperatively, 10 had a temporary impairment in one or more axes which fully recovered, and 5 had permanent damage to one or more axes. Of 32 patients who had impairment of at least one axis preoperatively, 11 showed documented improvement in endocrine function, 19 showed no change, and 2 became worse. It can be concluded from these data that approximately 7 percent of patients (7 of 97) will experience damage to one or more pituitary target organ axes as a sequel to transsphenoidal surgery performed by a surgeon experienced in transsphenoidal techniques.

Diabetes insipidus is a well-recognized complication of transsphenoidal surgery. The resultant fluid and electrolyte imbalance is usually mild and temporary. However, excessive manipulation of and damage to the posterior pituitary lobe or the proximal portion of the pituitary stalk may cause permanent injury.

Rarely, direct manipulation of the parasellar and intracranial structures may cause complications. These include haemorrhage from the cavernous sinus, cranial neuropathies (usually involving cranial nerves IV and III), carotid artery injury, intraventricular and subarachnoid haemorrhage, hypothalamic damage, and new postoperative visual deficits.

  Radiation Treatment

Conventional external radiation and stereotactic irradiation have been used to treat prolactinomas. In a retrospective study of 29 patients with nonfunctional or prolactin-secreting pituitary tumors who received radiotherapy alone, Rush and Newall reported normalization of serum prolactin levels in 7 of 10 patients (70 percent) with hyperprolactinemia. Sheline and colleagues found that radiotherapy alone decreased the average prolactin level in 75 to 90 percent, and normalized it in 30 percent of patients with prolactinomas. Tsagarakis and colleagues followed the effect of external beam megavoltage radiotherapy (4500 cGy at total daily doses of 180 cGy or less) in 36 patients with prolactinomas for 3 to 11 years (mean of 8 years). The serum prolactin fell into the normal range in 18 (50 percent) of the 36 patients, Two patients had a level higher than that at presentation, with one demonstrating tumour recurrence on CT scan. Of note is the high level of pituitary dysfunction. Eight patients (23 percent) had gonadal deficiency, and 34 patients (94 percent) had growth hormone deficiency, while disturbances of the pituitary-thyroid and pituitary-adrenal axis were evident in 5 patients (14 percent).

The adverse effects of brain irradiation are lasting and irreversible. These include hypopituitarism, optic nerve or chiasmal injury, brain radionecrosis, and carcinogenesis. Given the low risk/ benefit ratio of medical and surgical therapies for prolactinomas, and the known complications associated with sellar irradiation, the authors believe that the place for radiation treatment in the management of prolactinomas is limited. Radiation therapy should only be considered in patients who have a macroprolactinoma or invasive tumour in which a primary gross total resection could not be achieved, or for recurrent tumors associated with mass effect, if medical or surgical treatment is contraindicated or deemed ineffectua1. Routine irradiation after transsphenoidal microsurgical gross total removal of a prolactinoma is not an appropriate management strategy.

  Residual and Recurrent Tumour

Although the recurrence rate of prolactinomas following transsphenoidal surgery has not been studied adequately, a number of large series suggest that overall this rate is approximately 5 to 10 percent. The incidence of tumour persistence or recurrence is increased in patients treated with a frontal craniotomy, in those who receive irradiation as the sole treatment, and in those with macroadenomas or invasive prolactinomas. No pathologic markers have been identified that can predict which prolactinomas are apt to recur.

Medical, surgical, and radiation treatments have been used to treat recurrent prolactinomas. The criteria for choosing the most appropriate option differ from those for primary treatment. Although the factors that influenced the choice of primary treatment still remain valid, unresponsiveness to a particular therapy, surgical complications associated with reoperation, side effects from prolonged drug therapy, and prior cranial irradiation must also be considered.

The treatment of recurrent prolactinomas must be individualized. In our experience, the vast majority of recurrent prolactinomas can be managed medically with bromocriptine. Surgery via the transsphenoidal approach may occasionally be appropriate, especially in a patient with serum prolactin values <500 ng/ml and an MRI that does not show gross cavernous sinus invasion by the tumour. If the tumour exhibits extrasellar extension, a transcranial intradural approach is used. On occasion, the recurrent tumour may be debulked using both approaches, either simultaneously or consecutively. In these situations, bromocriptine is a useful preoperative adjunct. The administration of postoperative medical or radiation therapy is recommended, Clinical judgment regarding the individual case dictates the form and duration of these treatments.