Conventionally fractionated radiotherapy has a long established place in the therapy of pituitary adenoma.

In 1909 Beclere published his famous case report entitled: "the radiotherapeutic treatment of tumours of the hypophysis. gigantism and acromegaly". In this paper he described the first case:A young girl of sixteen suffering from hypophysomegaly. The xrays showed a notable enlargement of the sella turcica. She suffers from violent attacks of cephalalgia, severe visual troubles, gigantism, genital infantilism and excess adiposity …"

Concerning radiotherapy, he wrote:"The seances have now been carried out once a week for ten weeks. At each seance, the hypophysis was treated by cross fire through four or five different areas on the fronto-temporal region, the skin dose on each being 3H. The attacks of cephalalgia have completely disappeared. Still important is the improvement in the visual troubles."

Some fifty years ago, in an era before modern MR scanning, Sheline described recurrence free intervals at 5 and 10 years of only 25% and 9% following transcranial surgery for adenomas of the pituitary. However, with the routine delivery of post operative radiotherapy, the recurrence free statistic became 79% disease free at 10 years, and subsequent data confirmed this protection from relapse. Brada et al (2) found an actuarial progression-free survival rate of 94% at five and 88% at ten years (slightly worse control rates for secretory than non-functioning adenomas) with the routine use of post-operative radiotherapy. Other data have unequivocally demonstrated the slow but progressive decline in the circulating hormone product of secreting tumours over many years, following conventionally fractionated external beam radiotherapy for acromegaly (3), Cushings Disease, Nelson's syndrome(4,.5) and prolactinoma (6) - reviewed below. By observance of careful modern planning and dose prescription delivery, the risks to the normal surrounding nervous system should be extremely low.

The term radiotherapy refers to the delivery of ionising radiation onto tumours to arrest their growth (by DNA damage and preventing mitosis), whilst not destroying the structure or function of the surrounding normal tissues. As cellular damage mediated by ionising radiation is not tumour selective, radiotherapy must exploit other differences to enhance the therapeutic ratio and clinical effectiveness.

The first and most obvious objective is to deposit more radiation on the tumour than on the surrounding normal tissues - this is achieved by careful planning techniques. In conventional radiotherapy, it is usual practice to make an individually constructed mask to immobilise the patient's head and then to bring in at least three beams from different directions to 'cross-fire' the pituitary/tumour and thus concentrate the dose on the target, whilst sparing the normal surrounding structures that do not lie in this tight cross-fire zone.

Nowadays, linear accelerators generating megavoltage photon beams are employed. These are deeply penetrating beams with higher doses at depth, per unit dose deposited on the surface/skin than the old-fashioned deep X-ray therapy machines (DXT orthovoltage). The megavoltage, linear accelerator generated beams are also much better collimated than DXT beams (i.e.less side scatter to adjacent organs). The three usual beam directions are from an antero-superior approach and two lateral approaches, but in some of the newer stereotactically mapped or IMRT techniques there are up to five or six beams approaching from more directions, and each utilising microcollimation to shape or conform the beam to the patient's tumour/target volume. In Tomotherapy the beam is a continuous spiral and the thin ‘fan’ beam is modified in its dimensions continuously throughout therapy to more sophisticatedly conform the high dose of treatment to the individual tumour’s dimensions – this is a form of Intensity Modulated Radiation Therapy (IMRT).

The second feature to improve the clinical efficacy of radiotherapy is the delivery of the total radiation dose in a small number of 'fractions'. It has been known for some ninety years that fractionation (the splitting up of a large total dose into small daily doses - fractions) allows more sparing to the normal tissues than to tumours. Normal tissues are under homeostatic influences to sense and repair cellular damage/depletion whereas tumour cells have lost this sense – Thus, fractionation allows a differentially damaging effect to be unleashed on the tumour versus that on normal tissues (with repair capacity).

For all the above reasons, patient are currently treated, immobilised in an individually constructed plastic mask. A dose prescription of 45 Gy is delivered in 25 fractions of 1.8 Gy each on week-days over five weeks.

Note that all this :- fractionation and multiple cross-firing portals is, in principle, what Beclere was practising 100 years ago, albeit with the old-fashioned orthovoltage x-ray therapy.

"Stereotactic radiosurgery" is a more recently introduced radiotherapeutic treatment method for pituitary adenoma and is a generally less understood mode of radiation therapy outside interested radiotherapy or neurosurgical circles. Its potential has only been realised in the last few years, since the advent of high Tesla MRI facilities and the increased sophistication of the planning and treatment hardware. Stereotactic technology allows precise target definition, and thereby allows concentration of the treatment beam of radiation more specifically on the mapped target (e.g. a discreet target within the pituitary fossa). There are several methods of delivering this therapy. The first is a charged particle beam (protons) that allows the precise deposition of dose due to manipulation of the Bragg peak; as this method is not generally available, it will not be further discussed here.

The other methods employ megavoltage photons and are rather more generally available. This is the exact same modality of ionising radiation therapy that conventional linear accelerators generate - the point being that there is no difference in the radiobiological interaction/relative biological efficacy of the beams employed in radiosurgical methods vis a vis conventional radiation therapy. The difference lies in the exactitude of the targeting (no margin of "safety" taken around the lesion) and the sharp "fall-off' of dose at the edge of the radiosurgical treatment volume - i.e. better sparing of adjacent normal structures. Consequently, a single large dose of this photon radiation can safely be deposited because surrounding normal tissues are not subjected to such a large percentage of the prescsribed dose. Such a large single dsoe is more cytocidal than a similar total dose of fractionated radiation.

The first method is the Swedish gamma unit (Gamma Knife, Electa instruments, AB Linkoping, Sweden), a second method is the linear accelerator method (the planning software acquiring the name: "x-knife" Radionics, Mass, USA) and latterly the Cyberknife (Accuray, USA) has entered the field.

In the current gamma unit (Gamma Knife), there are 201 fixed cobalt sources, each as a thin rod of l mm diameter, the long axis of which is oriented along a radius of a hemisphere (the helmet into which the patients head fits for therapy). The centre point (or isocentre) of this hemisphere is the point at which the stereotactic coordinates of the mapped intracranial target are positioned, and the 201 emissions are thereby simultaneously concentrated on this target. The patient’s head is rigidly confined within four bars of a pinned stereotactic frame – pinnend to the head under local anaesthetic. By using a combination of different ‘shots’ a highly conformal dose plan is achieved, bespoke for even the most irregular shaped target volume.

Many centres (particularly neurosurgical units possessing Gamma Knife technology) are moving from conventional, fractionated radiotherapy to stereotactic radiation/radiosurgery by single dose Gamma Knife methodology for the post operative therapy of pituitary adenoma. For example, the MR delineated residuum that lies in the cavernous sinus – a difficult place for the surgeon to clear in trans-sphenoidal surgery – represents an excellent example of the ‘good target’ that Gamma Knife can ‘focus on’ and upon which it can deliver an obliterative dose of radiation safely. The method seems less attractive when the whole pituitary fossa and adjacent tissues are at risk of tumour regrgowth and here conventional radiotherapy still scores better – Neurosurgeons working with Gamma Knife, in isolation from good radiotherapy facilities, may be reticent to concede this point.

The $65 million question is: which method is best? Early Swedish Gamma Knife data on the outcome of treating pituitary adenoma were not persuasive for the use of Gamma Knife technology (reviewed in a separate section later in this manuscript), but latterly the coupling of modern MR imaging into the planning process and the improved technology of the equipment has allowed a much more sophisticated process to more accurately deliver high focused radiation onto more certainly imaged targets; we therefore may be able to discount the relatively poor early results. Although the Gamma Knife data are not as mature as the conventionally fractionated radiotherapy results, nevertheless, it has become a popular and effective alternative . The Cyberknife is probably a rival to the Gamma Knife but the data on therapy for pituitary adenoma is still to be forthcoming.

The l linear accelerator method again involves the stereotactic mapping. Verhey et al (1998) noted that the conformity index (the "moulding" of the high dose isodoses to the target volume) was considerably better on the gamma unit than the x-unit, but at the cost of a higher internal dose gradient. Plowman and Doughty (1999) took these observations further with specific reference to pituitary adenomas. The usual clinical situation in pituitary adenoma radiotherapy is for the target volume to abut the hypothalamus and optic chiasm. These workers compared the optimal dose plans for both technologies. They found that the gamma unit gave the faster fall-off on the rostral border (fig 1). This is important; Plowman and Doughty found that, for example, whereas the optic chiasm might receive 20% of the marginal adenoma radiation dose on the linac radiosurgery plan, it only received 10% on the gamma unit (Gamma Knife) plan. Taking this further, to illustrate the importance of this difference, one can imagine a typical patient referred to a radiosurgical facility with a recurrence of adenoma following previous surgery and radiotherapy (during which the optic chiasm has received full dose). The recurrent tumor comes to within 5mm of the chiasm and the treating clinician is faced with delivering a therapeutic radiosurgical dose to the recurrent tumour without overdosing the previously irradiated and highly radiosensitive chiasm. We normally try to keep the dose to the chiasm down to 250 cGy or less in this circumstance. That the gamma unit allows us to deliver twice the dose to the tumour margin for the same dose to the chiasm is obviously advantageous in this circumstance; the fact that there is a higher internal dose gradient to the centre of the tumour is also advantageous and so, on both counts, the gamma unit is preferred (fig 2).

A study of the usefulness of conventionally fractionated radiotherapy for pituitary adenoma associated with acromegaly was reported by Cicarelli et al (1998). These workers followed a cohort of patients, treated by the modern three field, linear accelerator method, using a prescription of 45Gy in 25 fractions over 35 days (exactly as we now still prescribe). Seventy three patients were followed long term, 61 unoperated and 12 post-operative patients. All data points of growth hormone measurement were performed off medical therapy. The importance of this single institution series is that it included some relatively advanced cases that would not nowadays be diagnosed so late in the course of their disease - due to advances in imaging and management. Thus the mean pre-radiotherapy serum GH level was 103 +/- 14 mU/l, (51.5 +/- mU/l) - *to convert mU/l to g/l divide by a factor of three. Following radiotherapy, a slow but progressive decline in the GH level occurred over the next decade such that the GH level at the ten year follow up point was 12.2 +/- 2.4 mU/l (6.1 +/- 1.2 mU/l). Furthermore, in 31 patients who had been followed for more than 10 years there was a significant further fall in the next five year period. These data are important and are complemented by other data, all of which confirms the progressive fall in GH; the level of GH is approximately 50% of the baseline figure by two years post-radiotherapy and 25% by five years.

More recently, Jenkins et al (1999) pooled data on 372 conventionally irradiated acromegalic patients to present a modern series of results. The mean baseline GH level was 24.2 mU/l which fell to 12.5 mU/l by two years and 7.8 mU/l by five years.

A level of serum GH of less than 5 mU/l (the target 'safe' level, associated with a normal age-related serum IGF-1) was achieved in 44% by five years and 59% by ten years and 100% by twenty years. Given that over 50% of the fall in the GH occurs in the first two years, it is obvious that the patients with the lowest initial GH levels will achieve this clinically useful definition of 'cure'. In the series reported by Jenkins et al (1999), 70% of those who had an initial, pre-radiotherapy GH level of less than 10 mU/l had achieved a GH of less five by 1.7 years, whereas 45% of those whose initial GH level was between 10 and 30 achieved this but after 4.7 years. In those patients who presented with a GH of more than 60 mU/l, only 31% had achieved a GH level of under 5 mU/l by 7.6 years. Larger tumours tend to be correlated with higher GH levels in the serum and are anyway less easy to permanently control by radiotherapy, thus arguing for primary surgery for larger tumours and higher serum GH levels.

Thus only a small, intrasellar, growth hormone secreting adenoma, where the GH is less than 20 would one routinely consider primary radiotherapy. In unfit patients, the indications for primary radiotherapy obviously broaden, but interim medical therapy with somatostatin analogues will be required. Radiotherapy is used in the post-operative situation where there is macroscopic residuum, a very large tumour at the outset and where the post-operative GH level remains raised.

Following the introduction of the IGF- 1 assay, some authorities advocated the measurement of this rather than GH in the response assessment. Powell et al (2000) reported that more than 60% of patients achieved a normal IGF-1 level at ten years after radiotherapy which increased to 80% on longer follow-up. Biermasz et al (2000) reported that 70% of patients at ten years and 84% at fifteen years had achieved normal IGF-I levels. Jenkins et al (1999) reported normalisation of IGF-I levels in 27% of patients by 3 years, 53% by 7 years and 56% by 10 years after radiotherapy. However, IGF- 1 for GH is not a simple substitution. Whilst there is a statistical correlation between a single IGF-l level and a mean GH day curve value, it must be remembered that other factors may affect the IGF- 1 result, for example hepatic function and diet and also that some recent data suggest that it is local paracrine/autocrine IGF-1 production/concentration rather than serum concentrations that mediate the growth promoting effects of GH. Thus, GH assessment of the response of acromegaly to therapy will continue to be employed, as well as serum IGF-1 .

There appear to be no differences in the response of tumours that co-secrete prolactin as well as GH as compared to the pure GH secreting tumours. However, there is another phenomenon going on in the post-irradiation period which influences the results of serial monitoring of prolactin levels. Whilst tumour derived prolactin levels fall progressively, post-radiation effects on the hypothalamus lead to damage to the dopaminergic, inhibitory control over the pituitary lactotroph cells with a consequent tendency for the prolactin levels to rise (Cicarelli et al l 989), although this usually gradually then returns to normal over time.

Most authorities would now agree that dopamine agonist medical therapy is first line therapy for patients with prolactinomas. However, whether the long term use of such medical therapy is optimal management is more questionable, given the risks of tumour rebound on discontinuation of therapy, and, for women contemplating pregnancy. the risk of tumour swelling during pregnancy.

Surgery is excellent therapy for pro lactinomas, used after medically induced shrinkage of the tumour back into the fossa but the long term cure rates of surgery, particularly for macroadenomas has been suboptimal and recurrence is common. There are now good data supporting the use of radiotherapy in the achievement of long term control/cure.

A study from St Bartholomew's Hospital , reported by Tsagarakis et al (1991), described the long term outcomes of 36 women who had been irradiated for prolactinoma and followed. Twelve patients had macroprolactinomas and 24 had microprolactinomas; all prolactin data were assessed off dopamine agonist medical therapy. Before radiotherapy, serum prolactin levels ranged from 1150 to 34,000 mU/I (57 to 1700ng/ml). After radiotherapy, with a mean follow-up of 8.5 years (range 3-14 years) the serum prolactin fell to normal (Less than 360 mU/l) in 18 of 36 patients (50%) and to just above normal (in the 378-780 mU/I range) in a further 10 patients (28%). In two patients the prolactin levels rose and in one there was evidence of tumour progression on scan. Twenty three per cent of patients developed post radiotherapy gonadal deficiency, 94% GH deficiency and 14% TSH and ACTH deficiencies. The authors concluded that there was a progressive fall over time in the serum prolactin levels, and no growth on scans, in the vast majority of patients demonstrating the efficacy of this form of therapy in prolactinoma patients.

Although many clinicians have, in the past, treated microprolactinoma patients with radiotherapy, to restore menses and allow conception off dopamine agonist therapy and free from the risks of pregnancy induced tumour growth, this treatment method is now usually reserved for macroprolactinomas.

In a minority of patients receiving conventional radiotherapy for pituitary adenoma, the serum prolactin may rise in the late follow-up period due to radiation effects on the hypothalamo-pituitary inhibitory pathway on the pituitary lactotroph, as mentioned above. Such rises are relatively small but can serve to confuse the interpretation of hormonal data in the follow-up period of all pituitary adenoma patients, including prolactinoma patients.

It has been established for some thirty years that radiotherapy is useful in the management of pituitary dependent Cushing's disease. Orth and Liddle (1971) reported a relatively low remission rate but Ross et al (1979) and Lamberts et al (1980) reported that 50% and 52% of patients followed for a mean of 4 and 6.5 years respectively achieved a remission after radiotherapy, albeit in combination with medical therapy; the use of medical therapy and the use of adrenalectomy being potential 'obscurers' of long term efficacy of radiotherapy. A study of the usefulness of conventionally fractionated radiotherapy in Cushing's disease and Nelson's syndrome was reported from St. Bartholomew's Hospital (Howlett et al 1989). These workers studied 52 patients with Cushing's disease or Nelson's syndrome followed long term after radiotherapy. All patients had been treated by modern linear accelerator technology and to 45Gy in 25 fractions over 35 days. Twenty one patients had received radiotherapy as primary therapy for their Cushing's disease, all having achieved initial normalisation of circulating cortisol levels and complete medical remission on metyrapone medical therapy, and were under follow up with a mean interval of 9.5 years after radiotherapy.

Twelve of the twenty one patients (57%) were off all medical therapy and in clinical remission with a normal mean cortisol levels throughout the day. However, only two demonstrated completely normal plasma cortisol responses to dynamic testing. Medical therapy requirements progressively fell and only four patients were still on medical therapy, although five patients had undergone some alternative therapy adrenalectomy or transsphenoidal hypophysectomy because of inadequate control.

Fifteen patients received radiotherapy for Nelson's syndrome developing after bilateral adrenalectomy and had been followed for a mean of 9.6 years. Fourteen of these fifteen patients had responded with progressive depigmentation, shrinkage of the adenoma on scanning and a median ACTH level reduced to 16% of the original pre-radiotherapy value. Of nine patients who received radiotherapy after unsuccessful transphenoidal surgery, five were of all therapy at an early mean follow up point of three years, and of four who received prophylactic radiotherapy after transsphenoidal surgery only one had subsequently developed Nelson's syndrome.

This publication by Howlett et al was important in that it was a relatively large series of patients followed long term after modern radiotherapeutic therapy. It established that radiotherapy has a definite role as primary or second-line therapy for Cushing's disease and successfully achieves long-term control of the hyperpigmentation and tumour enlargement in Nelson's syndrome.

In the last decade, several surgical series have demonstrated that, with earlier diagnosis of pituitary tumours in the modern scanning era, and following modern transsphenoidal surgery, there are some patients who have a low risk of later relapse and do not require post-operative radiotherapy. In a predominantly surgery only series, Laws and Thapar (32) found ten year recurrences of 8% for acromegaly, 12% for Cushing's disease, 17% for functionless pituitary adenoma and 24% for prolactinoma. Comtois et al (33) reviewed a large series of functionless tumour patients treated by operation alone and found a 21% recurrence rate at 6 years. In a more recent series Lillehei et (34) looked at 32 functionless patients who had radical resections and no post-operative tumour identifiable on post-operative scanning (MR). There were only 6% recurrences at 5.5 years - both relapsing patients being salvaged by radiotherapy at the time of relapse. However, in 1999, a counterweight opinion came from Oxford (Turner et al 1999) - a centre with acknowledged expertise in pituitary adenoma management. A cohort of 65 patients were reported with functionless pituitary adenomas who had undergone radical surgery, without post-operative radiotherapy and by one specialist surgeon. The lifetable analysis of the group demonstrated an 18% recurrence rate at 5 years but 44% at ten years. Thus, from the modern literature, one can assess that following surgery alone for pituitary adenoma there is an up to 20% risk of recurrence at five years and up to 44% risk by ten years

Clearly, the surgeon's report and post-operative MRI scans are of importance in selecting patients for a careful 'watch" policy but we know from secretory tumour hormone data that the surgeon's assessment is imperfect. Serial MRI scanning in follow-up is mandatory in such patients. For patients with secretory adenomas, there are hormone markers and the definitions of endocrine cure are largely agreed. These have been the post-operative parameters by which we have assessed the need for post-operative radiotherapy until recently.

Nowadays. the concept of prospectively defining "aggressive " adenomas is a potentially important one. The first criterion is that of invasiveness, several studies demonstrating that this correlates with higher relapse rates (35-37). MRI/CT scans help to define invasiveness: cavernous sinus invasion can be concluded with certainty if the tumour encircles the intracavernous carotid artery and complete erosion of the dorsum sellae/clivus indicates invasiveness (over a pressure only effect) and presence of tumour within the sphenoid sinus indicates more than a pressure effect on the fossa floor. Adenomas infiltrating parasellar tissues including dura, bone, cavernous sinuses, paranasal sinuses, subarachoid space and leptomeninges are obviously invasive/aggressive. However, in one study by Selman et al (38), it was found that there was microscopic evidence of dural invasion in 88% of intrasellar macroadenomas and 94% ofthose with suprasellar extension: this suggests that dependence upon this criterion alone would overestimate the need for radiotherapy - and ann also that the "cottage-loaf" effect through the diaphragma sellae is usually associated with microscopic infiltration of this structure. Otherwise, routine histology (including the morphology of constituent cells) is not of great help in the definition of aggressive adenomas (39).

A number of immunohistochemical staining techniques have been studied in this regard. Specifically, determination of proliferation rates of pituitary adenomas by immunohistochemical detection of PCNA, Ki-67 and MIB-l have been reported to be positive in those tumors that will prove to be "aggressive" (40-45). For example Thapar et al (44) demonstrated that there were higher mean proliferative indices in invasive adenoma ( and pituitary carcinoma), when they employed MIB -l immunohistochemistry to study the proliferation rates among 70 patients. The implication of their results, as would be as anticipated, is that the growth fraction of pituitary adenomas predicts clinical behaviour.

Other work suggested that decreased immunohistochemical staining for nm-23 (a purine binding product of a metastasising suppressor gene) may be another predictor of aggressive behaviour (46). p-53 expression, overexpression of epidermal growth factor and increased protein kinase activity may all also correlate (47-49).

However, the absence of immunoreactivity for the tumour suppressor gene product, p-27, whilst correlating with tumour malignancy, did not prove useful in defining aggressive or recurrent adenoma in one recent study (50).

The next logical step is to select the immunohistochemical predictors of aggressive behaviour that most closely correlate with recurrence rates and integrate them with the previously mentioned criteria for selecting the need of post-operative radiotherapy.

When post-operative radiotherapy is recommended, and in our current practice this includes all patients presenting with large tumours or features of invasiveness, as outlined above, it is our current policy to normally select conventionally fractionated radiotherapy. Such radiotherapy concentrates the beam on the target volume (the whole fossa and any tumour extension beyond) by a "cross-fire" technique of several portals, each pointing at the target and with the patient immobilised in a tight fitting mask. These portals are conformed to irregular shaped volumes. Stereotactic radiotherapy techniques such as the Gamma Knife are used for specific well delineated targets (areas of known residual disease that are well demarcated on MR).

In summary, radiotherapy is less de rigeur in the post operative setting than ten to twenty years ago and the widespread availability of surveillance MR imaging has rendered this a safer policy – as relapses can be picked up earlier. Nevertheless, radiotherapy is an effective means of reducing the recurrence rates (and controller of uncured endocrine syndromes) in pituitary disease and should be a well studied and practised subject.

There are several potential complications to pituitary radiation, due to the nature and radiosensitivity of the pituitary itself and to that ofthe adjacent delicate nervous system: optic pathway and hypothalamus in particular.

The reason that radiotherapists now confidently assert that the risks to the optic chiasm are low is that the pathological basis for radiation morbidity to the chiasm via its vasa nervorum is better understood. Not surprisingly, the major factors in the radiation prescription that contribute to the morbidity risks are the total dose and the dose per fraction. Whilst employing radiotherapy for pituitary adenoma in a dose range of 4200-5900 cGy in Boston in 1963-1973, Harris and Levine (7) found radiation induced optic neuropathy in 4 out of 55 patients, but in none who received a daily dose per fraction of less than 250 cGy. Aristazabal et al (8) reviewed 122 patients treated for pituitary adenoma with radiotherapy and found optic neuropathy in 4 cases - all four patients occuring in a subgroup of 26 patients who had received more than 4600 cGy total dose and all four patients had received a daily dose of more than 200 cGy per fraction. Several other publications support the conclusion that large daily dose fractions put the optic chiasm at greater risk (9-12). A long-term, tumour compressed chiasm with optic atrophy on fundoscopy prior to radiation therapy must be a more vulnerable one. In the modern era these late visual sequelae should be of historic interest only, but they nevertheless serve to warn us of the dangers of badly conducted radiotherapeutic practice

In the last decade there are few recorded instances of such damage in the conventional radiotherapeutic literature and this testifies to the now general acceptance of these dose limitations for safety. In single arm series, using conventionally fractionated (200 cGy/ day or less) radiotherapy, Brada et al (2) recorded two cases of late and otherwise unexplained visual deterioration in a series of 411 patients treated with doses up to 5000 cGy (neither led to blindness). Jones (13) found no cases of late optic neuropathy in 332 patients treated to 4500 cGy in 180 cGy daily fractions. The perception of the risk of current standard conventionally fractionated dose prescription radiotherapy causing optic pathway damage should now be revised.

The total dose prescribed and probably the dose per daily treatment fraction are also the major factors determining the risk and speed of onset of any radiation induced hypopituitarism (15,16). This risk increases with time after therapy. The GH axis is the most sensitive to the late effects of radiotherapy (17), the radiation induced defect occurring largely at hypothalamic level (1 8). Rises in prolactin post-radiotherapy also have their origins at hypothalamic level (19) as may the observation of occasional early or accelerated puberty post-pituitary radiation (20,21). The gonadotrophin and corticotrophin axes are the next most sensitive to radiation damage, the TSH axis is least sensitive, and diabetes insipidus very rarely occurs. The reasons for this rank order of sensitivity to radiation is not known. As O'Halleran and Shalet (14) point out, the observation that radiation induced changes occur at hypothalamic level is relevant to therapy: it may be physiologically advantageous to treat gonadotrophin deficiency by pulsatile gonadotrophin secreting hormone and GH deficiency by GHRH analogues.

The overall incidence of hypopituitarism is greater the more disturbed the pituitary function is prior to radiotherapy (i.e. tumour or surgery related). These factors temper the acceptance of exact 'percentage" statistics on the incidence of post-radiotherapy hypopituiarism. Nevertheless, in our patients with acromegaly , 25% required new endocrine replacement therapy by five years after radiotherapy and the need continued to rise in the next five years (22). Feek et al (23) reported that by 10 years post radiotherapy, 47% of patients were hypogonadal, 30% were hypoadrenal, and 16% were hypothyroid. These incidences were 70%, 54% and 38% respectively when surgery preceded radiotherapy.

Whether the late risk of carcinogenesis is as high as 1-2% remains contentious (13, 24-26). In a long-term follow-up study of 334 patients treated for pituitary adenoma by radiotherapy, Brada et al (24) described the development of two gliomas, two meningiomas and one meningeal sarcoma, concluding that there is an increased risk of developing second tumours following this procedure of 1.9% at 20 years. Tsang et al (25) observed the occurrence of four gliomas (two in the brainstem) in the long-term follow-up of 367 irradiated pituitary adenoma patients. Jones (18) reported the occurence of one glioma, one petrosal spine tumour and two cases of myelogenous leukaemia ( one acute and one chronic) in the late follow-up of 332 irradiated pituitary adenoma patients. However he observed the occurrence of one glioma in an equal group of unirradiated pituitary adenoma patients and warned of use of background population statistic comparisons, rather than disease specific groups, advice which has been reinterated (27) and which may seem more pertinent in this age of GH tumour-growth-factor research. Bliss et a] (28) observed one malignant brain lymphoma in a patient who had also previously undergone chemotherapy for Hodgkin's disease, and one meningioma in the long-term follow-up of 193 irradiated pituitary adenoma patients. If we total these four series, we find that, in the late follow-up of 1226 irradiated pituitary adenoma patients, there have been seven gliomas, two benign meningiomas (although the Tsang et al reference might not have recorded benign tumours) and one malignant meningeal sarcoma/parasellar fibrosarcoma. When compared to background population incidences ( e.g three per thousand for glioma), albeit with whatever shortcomings these may have due to non -disease specificity, it is clear that any excess risks are small.

There was a concern, first voiced in the pediatric brain treatment literature, that localised radiotherapy to this area of brain led to more neuropsychological late changes than partial brain radiotherapy received by other brain sites or to the whole brain (28). However, although some workers have explored the possibility of neurocognitive late morbidity following radiotherapy for pituitary adenoma, any changes were found to be unrelated to radiotherapy (29-31). Grattan-Smith et al (29) from Australia found more neuropsychological disturbances than expected in pituitary adenoma patients, but this was unrelated to whether the patients had received radiotherapy. An English study (30) found poor social adjustment, including mood disorders (31) related to therapy ( but as part of a multifactorial problem, to which surgery and hormonal imbalance also contributed). Further research is required before and after therapy in hypothalamo-pituitary tumour patients.

Figure 1. Top panel: Coronal MRI of pituitary adenoma that is extensive across the whole fossa and lies only just below the chiasm. The case for radiosurgery here is weak as conventional radiotherapy will more easily encompass the whole extent of disease and carry less risk (dose) to the chiasm which is so close to the target volume. Bottom panel: Coronal MRJ scan of small discrete pituitary adenoma. If surgery is not curative, the optimal radiation therapy method for such a tumour is controversial. Whilst stereotactic radiosurgery at first seems ideal for such a well demarcated and small lesion, well away from the chiasm, is it too focal for a case that surgery has failed to cure and would it jeopardise subsequent conventional radiotherapy, for which the 'pedigree' is better?

Figure 2. Sagittal MR scans of pituitary adenoma with superimposed Gamma Knife isodosimetry. Left panel demonstrates standard isodosimetry with the fifty per cent isodose well-encompassing the target/tumour; the optic chiasm is outlined in purple and the twenty per cent isodose is seen to just 'touch' this structure. On the right, the added facility of blocking portals facing the chiasm has been employed (blocking pattern shown in bottom panel) with the added advantage (particularly in a retreatment situation - where the patient has already received radiotherapy) viz. the reduction in dose to the chiasm to the ten per cent isodose is seen - see isodoses in the top right photo.

Craniopharyngioma constitutes 8% of all childhood tumours and 2% of adult tumours of the central nervous system (CNS) - with a median age of presentation of8 years and two thirds of patients presenting before the age of 20 years. Grossly, craniopharyngiomas are cystic, solid or partly both, not obviously encapsulated and not separated by leptomeminges from the brain parenchyma. Indeed, spread in the brain tissues , particularly tuber cinereum and hypothalamus, make radical resection potentially hazardous as it is difficult to define surgical planes between tumour and critical central nervous system.Thus, while all authorities agree that surgical excision is optimal primary initial management, there is persisting controversy as to whether radical resection should be attempted due to its significant risk of morbidity. However, without radical resection, the regrowth rate of subtotally resected tumours is high. In a pooled series of 111 subtotally resected craniopharyngioma patients, Amacher found that 75% regrew, requiring further treatment. In a San Francisco series, the surgeons considered that they had achieved complete excision in only 10% of 74 consecutive craniopharyngioma patients. Thus, it is clear that safe complete excision is both difficult to achieve and difficult to forecast even at the end of the operation. In patients with incomplete tumour excision, fewer than half survive for ten years and 50-75% have recurrent disease within 2-5 years, in the absence of radiotherapy.

Accumulating data heavily support the practice of post-operative radiotherapy. Heideman and his colleagues pooled the actuarial 5 and 10 year survival rates from various series to obtain the following results: 1) "total resection" : 58-100% and 24-100%, 2) subtotal excision :37-71% and 31-52%, and 3) subtotal excision and post-operative radiotherapy : 69-95% and 62-84% respectively. Additionally, neuropsychological function was better preserved in the combined modality group. The case for post-operative radiotherapy therefore appears to be overwhelming. As for pituitary adenoma, there has been interest in the use of stereotactic radiosurgery for this disease which is clearly a well demarcated volume of benign tumour in the brain of usually young patients. The use of such focal radiation technology at first seems compelling (Fig 3). However. the optic chiasm usually lies within the treatment volume and is likely to receive the full dose of this single high dose technique of radiation therapy and complications are likely. The problem of recurrent cystic craniopharyngioma is an occasional but very difficult one for clinicians interested in this disease. Beta ion ising radiation has been shown in vivo and in tissue culture in vitro to destroy epithelial lining of craniopharyngioma cysts.The early Scandinavian work established that instillation of a beta emitting isotope into a craniophayngioma cyst resulted in the therapeutically useful reduction in cyst wall secretion. We recently collated the published experience since that time. From a total of 149 cysts treated, 121 either reduced in size (and more importantly stopped re-filling) or were obliterated. Our own experience to date is with nine cases and we have used 90-yttrium colloidal solution to effect the dose to the secreting epithelium of the cyst of 200 Gy (Fig 4). The future will undoubtedly hold more for stereotactic delivered radiation therapy for solid craniopharyngiomas components and intracystic instillation of isotopes for cystic disease.

Figure 3. Axial CT scans of recurrent craniopharyngioma before (top panels) and after (bottom panels) sterotactic radiosurgery demonstrating the complete response of this solid recurrence.

Figure 4. Recurent cystic craniopharyngioma before (left panels) and after (right panels) intracystic instillation of 90-yttrium colloidal solution, demonstrating a good response on plain lateral skull films with intracystic instillation of contrast (lesion arrowed), and CT (left panels) and MR (right panels) scans.

Formerly called Histiocytosis X, LCH is a disease characterised by the aggressive infiltration of tissues by Langerhans cells (a unique subset of macrophages) and in up to 15% of childhood cases (the commonest age group) diabetes insipidus occurs due to LCH affecting the posterior pituitary and hypothalamus. Medical therapy (steroids and chemotherapy with agents such as vinca alkaloids or etoposide or immunotherapy with cyclosporin) is used as therapy particularly where there is multi-system disease. Low dose radiotherapy (1200-1500 cGy in 8-10 fractions) to this brain region may be useful in limiting the diabetes insipidus where this defect is the dominant problem. Whenever cytotoxic chemotherapy or radiotherapy is to be used for 'benign' disease, the pretreatment counselling has to be thorough.

When sarcoidosis affects the nervous system, the sites of predisposition for lesions are the hypothalamus, pituitary and the third ventricle, as well as cranial nerves (most frequently the facial and optic nerves) as part of a basal meningitis. Diabetes insipidus occurs in 33% of patients with neurosarcoidosis and other features of hypothalamo-pituitary dysfunction frequently accompany this, e.g. somnolence/sleep disorder, weight gain, impotence/hypogonadism, amenorrhoea and panhypopitutiarism. The hypothalamus is the usual site of first involvement in this region but the anterior pituitary itself may be affected, although its involvement by itself(without diabetes insipidus) is rare. Imaging may demonstrate 'space occupying' granulomata or thickened basal meninges and the CSF may show a raised lymphocyte count and protein as well as ACE levels, but importantly may also be normal.

Corticosteroid treatment in high dosage (e.g. prednisolone at 80 mg per day) may be useful therapy in the acute labile phase but the disease is not particularly steroid sensitive and those patients demonstrating a partial response may need longer term maintenance. Isolated reports of responses to low dose equivalent radiotherapy justify the inclusion of the problem here although the current author has yet to see a convincing response having irradiated two or three patients in a fifteen year period.

Primary intracranial germ cell cell tumours (IGCT) are rare and most commonly ocur in the first two decades of life. The incidence varies geographically; thus, although they account for 4-10% of all childhood brain tumours in Japan, the comparable figure in the West is 0.2-2%. The annual incidence in the USA is 40 per year.IGCT are histologically a heterogenous group of tumours and their clinical presentation also differs according to their site of origin. Germinomatous germ cell tumours or germinomas (GGCT) - the homologue of seminoma - are the most exquisitely chemo and radio-sensitive. Non-germinomatous germ cell tumours (NGGCT), comprising choriocarcinoma, teratoma and embryonal sinus (yolk sac) tumur and embryonal carcinoma have a poorer overall response to therapy and prognosis. Human chorionic gonadotropin-beta (hCG) may be produced by germinomas in small amounts but is secreted by choriocarcinoma elements of NGGCT in large amounts. We described that when paired measurements are taken in secreting IGCT, the CSF:blood ratio of hCG shows a 10:1 ratio. Alpha-foetoprotein is secreted by yolk sac elements of NGGCT. Detection of these markers in the serum or CSF is diagnostically and prognostically useful. IGCT most commonly occur in the pineal and then the suprasellar region at the hypothalamus; less commonly they occur in other midline brain sites. Synchronous IGCT occuring at pineal and suprasellar locations is a well recognised phenomenon (Fig 5). Whereas there is a definite male preponderance in the incidence of pineal IGCT this sex inequality is not found in suprasellar IGCT where there is either an equal sex incidence or even a slight female preponderance. In our small experience, dual site primaries did show a male preponderance.

Figure 5. Sagittal MR scans of a dual site intracranial germ cell tumour before (left panel) and after (right panel) chemo-radiotherapy, demonstrating a complete imaging response (also reflected in normalistaion of markers).

Suprasellar IGCT represent 30% of all IGCT, are most commonly GGCT and present with diabetes insipidus, visual failure and hypopituitarism. Indeed, in our own series of 10 cases , diabetes insipidus was invariably present at presentation. Hyperprolactinoma due to stalk disruption was also very common. There was a high incidence of abnormalities of thirst, which, in combination with diabetes insipidus, led to some serious abnormalities in fluid balance in our series - compounded when we used platinum based chemotherapy (for which pre- and post- hydration is mandatory). Weight loss due to anorexia was also a feature in our series.

Diagnosis is nowadays based on MRI demonstrating a midline mass (positive on PET scanning unlike benign tumours of these regions) and either diagnostic levels of CSF/serum markers and/or biopsy.

Curative therapy is with chemo-radiotherapy (Fig 5). Traditionally the results of radiotherapy for this disease have been good but more recently it has been appreciated that there are differing cure rates within the IGCT spectrum. Sano demonstrated from the Japanese data that the prognosis for GGCT was excellent, whereas it was poor for AFP/hCG secreting NGGCT. Furthermore, with the increasing success of chemotherapy for extracranial germ cell tumours, systemic chemotherapy was found to be useful. Our current view is that initial chemotherapy – perhaps four cycles of a cis-platinum bsed regime suitable for germ cell tumours followed by radiotherapy, perhaps dose reduced where there has been a good clinical response (and particularly in GGCT) - is current best medicine. The necessity for whole neuraxis prophylactic radiotherapy to all patients is controversial – the present author is still in the phase of neuraxis radiation dose reduction in patients with GGCT who have achieved a complete response on MRI to the primary chemotherapy regime.

Secondary cancer (metastatic deposits) occur with a preferential disposition in the posterior pituitary and bronchus and breast primary carcinomas are common sites of origin. Radiotherapy is the treatment of choice.

Meningiomas of the sphenoid ridge and cavernous sinus commonly spread across the fossa and clivus chordomas and chondrosarcomas similarly. Both may present with pituitary endocrinopathy. Other conditions include the advanced nasopharyngeal carcinoma that has infiltrated through the skull base and other skull base tumours of which a plasmacytoma has recently been treated in our unit. Radiotherapy has an important role in the management of all these conditions.