Conventional radiation therapy (CRT) has traditionally been used in patients with residual or recurrent secreting and nonfunctioning pituitary adenomas who have failed prior medical management and/or surgery, resulting in a variable long-term tumor control of 87–95 % at 10 years [1–4], and normalization of elevated plasma levels of growth hormone (GH) and adrenocorticotropic hormone (ACTH) in up to 55 %, and 78 % of patients, respectively [5–8]. Hypopituitarism occurs in 30–60 % of patients 5–10 years after irradiation, while other toxicities, including radiation-induced optic neuropathy, cerebrovascular accidents, and secondary tumors have been reported in 0–5 % [9–12].

Stereotactic radiosurgery (SRS) is a sophisticated radiation therapy technique that precisely delivers high dose of irradiation in a single o few (2–5) fractions to well-defined, small-to-moderate brain targets. SRS allows for more precise target localization and accurate dose delivery as compared with CRT, resulting in a reduction of the volume of normal brain tissue irradiated to high radiation doses [13]. The techniques used for the treatment of a pituitary adenoma involve the Gamma Knife (GK) [14], the CyberKnife (CK) robotic radiosurgery system [15, 16], or a modified conventional radiotherapy machine (linear accelerator, LINAC) [17, 18]. Data from literature report a tumor control after SRS up to 97 % at 5 years, with normalization of hormone hypersecretion in more than 50 % of patients [19]. Hypopituitarism is the most commonly reported late complication of treatment, whereas other late radiation-induced complications are low. As high doses are delivered to the tumor with the use of the stereotactic radiosurgical techniques, an accurate delineation of target and surrounding normal brain structures becomes increasingly important to minimize radiation-induced toxicity while maintaining high tumor control.

We aimed to provide a critical review of the different aspects of radiosurgical techniques for pituitary tumors, including the delineation of target and critical organs, technical characteristics of the different types of SRS delivery systems, the optimal dose and fractionation for nonfunctioning and secreting pituitary adenomas, and the long-term efficacy and toxicity.

A literature search was conducted in MEDLINE PubMed that evaluated adults with pituitary adenomas. The search focused on randomized, prospective and retrospective studies published in English. The searches were limited by date from January, 2000 to November, 2015 using a combination of medical subject headings (MeSH) (“pituitary adenomas/radiosurgery” or “nonfunctioning pituitary adenomas” or “acromegaly” or “Cushing disease” or “prolactinomas”) and free text terms (“toxicity” or “hypopituitarism” or “target delineation” or “radiosurgical dose” or “fractionated radiosurgery” or “organs at risk”). Articles were excluded from the review if they: had a non-English abstract, were not available through Pubmed, were pediatric series or case studies involving less than 8 patients, or were duplicated publications. To identify additional articles, the references of articles identified through the formal searches were scanned for additional sources. A total of 984 potentially relevant studies were identified. Finally, 92 studies reporting the clinical outcomes of SRS for either nonfunctioning or secreting pituitary adenomas with a minimum follow-up of 1 year were selected and included in the review.

Defining the optimal target volume for a pituitary adenoma represents a balance between minimizing treatment-related toxicity while maintaining a high tumor control. Current optimal imaging technique for target delineation requires the use of precontrast and postcontrast magnetic resonance imaging (MRI) sequences to improve the accuracy of target identification and delineation. Contrast-enhanced 3D T1-weighted sequences with 1 mm thin slices are extremely useful for accurate target delineation by allowing identification of subtle enhancement patterns in the surrounding neurovascular structures and along the course of the optic nerve [20]. For planning purpose, MRI scan is subsequently fused with thin-slice non-contrast enahnaced CT scan. Although a displacement up to 2.8 mm has been reported for brain soft-tissue based fusion, the magnitude of displacement is considered negligible for lesions of the skull base due to its rigidity and great visibility in all imaging modalities [21]; so far, no additional margins would be required to ensure adequate target coverage during SRS to compensate fusion uncertainties. Since most pituitary adenomas are benign, slow-growing neoplasms, peritumoral edema is generally absent. For this reason, T2-weighted images, which are extremely useful in evaluating the parenchyma of the brain and the perilesional edema, are not generally used for target volume delineation. Preoperative MRI may be helpful to discern postoperative changes from tumor, especially in patients who had undergone several prior surgeries. Similarly, contrast-enhanced T1-weighted images with fat suppression may be used to minimize postoperative changes that might obscure the accuracy of radiosurgical targeting. when MRI is contraindicated, a thin-slice CT imaging through the pituitary regions is performed with and without contrast administration.

The gross tumor volume (GTV) is represented by the lesion visible on MRI/CT. The clinical target volume (CTV) includes microscopic disease. In general, additional margin expansion from GTV to CTV is unnecessary in pituitary adenomas; however, a small margin may be added in the intracavernous portion of aggressive adenomas to encompass potential areas of microscopic tumor infiltration. The planning tumor volume (PTV) should take into account uncertainties of patient setup. Currently, a similar sub-millimteric accuracy of target positioning has been reported for frameless CK and LINAC based systems (Novalis Tx) and frame based GK SRS technology [14–18, 22, 23]. In most centers, a margin of 0–1 mm is generally used for GTV to PTV expansion; however, due to the different commercial SRS systems, each department should audit their setup results and apply the margins on the basis of their own observations.

The sellar and parasellar region is an anatomically complex area including endocrine, nervous, and vascular structures. The pituitary fossa comprises the pituitary gland, which is composed of the adenohypophysis and neurohypophysis. The parasellar region encompasses the cavernous sinuses and the suprasellar cistern structures. The cavernous sinus consists of trabeculated, multilobulated venous channels which are located lateral to the sella turcica and sphenoid sinus. The cavernous sinus contains cranial nerves III (oculomotor), IV (trochlear), V1 (ophthalmic division of the trigeminal nerve), V2 (maxillary division of the trigeminal nerve) and VI (abducens). It also contains the cavernous segment of the internal carotid artery. The suprasellar cistern includes the optic chiasm and nerves, the anterior third ventricle, the hypothalamus, the pituitary infundibulum, the infundibular and suprachiasmatic recesses of the third ventricle.

A careful delineation of all organs at risk (OARs) surrounding the target volume is mandatory. OARs in the skull base region include optic nerves and chiasm, brainstem, pituitary stalk, pituitary gland, and cavernous sinus cranial nerves (an example of GTV and OARs contours is shown in Figs. 1 and 2). Expansion of OARs to create a planning risk volume (PRV) for each OAR may be applied; the margin, as for the GTV, should reflect the accuracy of daily set-up. Overlaps between PRVs and PTV should be considered; however, caution should be used when the reduction of the dose to the OARs may results in inadequate dose coverage of PTV. With regard to dose limits for the OARs, the optic nerves and chiasm are believed to be the most radiation-sensitive structures to SRS. A risk of radiation-induced optic neuropathy up to 2 % has been reported for point doses to the optic pathway of 8–10 Gy [24–31]; however, the risk of optic neuropathy remains low for point doses of 10–12 Gy to small portions of the optic apparatus [25, 27, 29, 30]. In a retrospective series of 222 patients who received GK SRS for benign tumors adjacent to the anterior visual pathway, Leavitt et al. [29] observed no new visual symptoms for patients receiving a maximum dose of 12 Gy to small portions (2–4 mm³) of the optic chiasm after single-fraction SRS. The risk of developing radiation-induced optic neuropathy was 0 for patients receiving a maximum point dose of 8–12 Gy and 10 % for those receiving a maximum point dose of 12–15 Gy to the anterior optic pathway. Hasegawa et al. [27] evaluated 100 patients undergoing GK SRS for craniopharyngiomas. Two patients who received maximum radiation point doses to the optic pathway of 15 and 18 Gy, respectively, developed optic neuropathy, whereas no visual deficits were observed in patients receiving lower doses. While these studies suggest that point doses up to 12 Gy to small portion of the optic pathway are associated with a low risk of optic neuropathy, in clinical practice a maximum point dose of 10 Gy is usually recommended when treating lesions adjacent to the optic pathway.

Little is known about the tolerance of the cranial nerves of the cavernous sinus. Leber et al. [25] reported no cranial nerve injury in patients receiving single-fraction SRS when doses of 5–30 Gy were delivered to the cavernous sinus. In contrast, Tishler et al. [24] reported a 13 % incidence of the third and sixth cranial nerve in 62 patients undergoing GK SRS; however, they could not find a significant relationship between the delivered dose of 10–40 Gy and new or worsening deficits. Although a precise tolerance dose of cranial nerves within the cavernous sinus after single-fraction SRS cannot be defined, doses up to 18 Gy to the cavernous sinus are associated with low incidence of radiation-induced toxicity (0–4 %) [13, 32].

Hypopituitarism is the most common adverse effect after SRS for a pituitary adenoma. Several studies have evaluated the relationship between radiation doses to the normal pituitary gland and distal infundibulum [33–38] and the development of hypopituitarism. Leenstra et al. [35] reported on 82 patients with either nonfunctioning or secreting pituitary adenomas who received GK SRS at the Mayo Clinic. Applying the criteria of a mean dose of 15 Gy to the pituitary gland, they noted new endocrine deficits in 12 of 40 patients (30 %) for doses < 15 Gy compared with 9 of 20 patients (45 %) who received a mean gland dose > 15 Gy. In their analysis they found new anterior deficits in 0 %, 29 %, 39 % and 83 % for mean doses to the pituitary gland ≤ 7.5Gy, 7.6–13.2 Gy, 13.3–19.1 Gy, and > 19.1 Gy, respectively. In another series of 85 patients treated with GK for a pituitary adenoma, Marek et al. [36] reported an incidence of hypopituitarism of 2.2 % for patients irradiated with a mean dose to pituitary < 15 Gy and 72.5 % for those who received a mean dose > 15 Gy. A significant correlation between the mean dose of 15 Gy to the pituitary gland and the development of new pituitary deficits has been reported in other studies [34, 38].

The correlation between the mean dose delivered to the pituitary stalk and the incidence of hypopituitarism has been evaluated in retrospective series [33, 34, 36, 38]. In a series of 130 patients treated with single-fraction SRS, Sicignano et al. [38] reported 5-year actuarial incidence of new pituitary deficits of 8 % for a mean dose to the pituitary stalk < 7.3 Gy and 32 % for a mean dose to the pituitary stalk > 7.3 Gy. Similarly, Feigl et al. [33] observed a significant incidence of new endocrine deficits for doses > 6.5 Gy to the pituitary stalk in a series of 108 patients treated with GK SRS for a pituitary adenoma. In contrast, Vladika et al. [34] found a significant incidence of new pituitary deficits after single-fraction SRS only for patients who received a maximum dose to the pituitary stalk > 17 Gy. Future prospective studies with an appropriate follow-up will be necessary to better identify the maximum safe doses to the pituitary gland and the pituitary stalk. Whenever possible, mean radiation doses to the pituitary gland and stalk should be kept under 12–15 Gy and 7–10 Gy, respectively, with the aim of limiting the development of new pituitary deficits.

Other OARs include the brainstem and hippocampi. For single fraction SRS, maximum brainstem doses of 12–14 Gy are associated with low (<5 %) risk of neurological complications, although this risk significantly increases for doses > 15 Gy given as single fraction [28, 39]. In a recent review of radiation associated brainstem toxicity, Mayo et al. [28] calculated a risk of normal tissue complication probability of 1 %, 13 %, 61 %, and 94 % for partial volume irradiation of one third of the brainstem to doses of 12.5, 14.2, 16, and 17.5 Gy, respectively. A lower risk of complications was observed when the same doses were delivered to a small partial volume (1 %) of the brainstem. Although definitive criteria of dose-volume effects on brainstem dose tolerance after single-fraction SRS remains to be better defined, in clinical practice caution should be used when delivering doses to the brainstem > 12.5 Gy. For tumors located in the parasellar region, hippocampi can be contoured as an effort to reduce the potential negative neurocognitive effect of high radiation doses to the hippocampal region [40]; the principle of this approach is acknowledged but there is currently insufficient evidence to support recommendations on hippocampal sparing during SRS.

There is limited evidence relating tolerance of the optic apparatus and cranial nerves of the cavernous sinus after multi-fraction SRS. Retrospective studies have observed a risk of optic complications of less than 1 % for patients with skull base tumors treated with doses of 21–25 Gy delivered in 3–5 fractions [41–46]. Liao et al. [45] reported the outcome of fractionated SRS delivered with a LINAC system. Thirty-four residual/recurrent pituitary adenomas with a median tumor volume of 4.11 cm³ in close proximity to the optic apparatus (median minimal distance 1 mm, ranging from 0 to 2.5 mm) were treated with a total dose fo 21 Gy in 3 fractions of 7 Gy each. With a median follow-up of 37 months, no patients developed optic neuropathy; the mean single-fraction doses to the optic nerve and chiasm were 5.58 ± 0.98 and 4.86 ± 0.15 Gy, respectively. One patients developed transient diplopia after SRS, which resolved after a short course of dexamethasone. Using doses of 21 Gy in 3 fractions or 25 Gy in 5 fractions delivered with CK, Iwata et al. [44] reported a grade 2 visual disorder in only 1 out of 100 patients at a median follow-up of 33 months; however, no details of doses delivered at optic apparatus were provided in their study. In another study of 34 patients who received a multifraction SRS (5 × 5 Gy) at University of Rome Sapienza for a skull base metastasis involving the anterior optic pathway, at a median follow-up of 13 months no optic neuropathy were observed for doses >25 Gy to less than one-third of optic chiasm and > 27.5 Gy to a small volume of 0.01–0.06 cm³ [46]. With regard to the cavernous sinus cranial nerves tolerance, no deficits have been reported using median doses of 20 Gy delivered in 2 to 5 fractions for perioptic lesions [41–43]. Although these studies indicate that 5 × 5 Gy or 3 × 7 Gy schedules are associated to a low risk of radiation-induced optic neuropathy and cavernous sinus cranial deficits, further studies need to better evaluate the dose-volume relations for OARs during multi-fraction SRS of patients with pituitary tumors.

SRS for pituitary adenomas is typically delivered as single-fraction SRS or, less frequently, as multi-fraction SRS (2–5 fractions). Main used techniques include the use of GK, CK or a modified LINAC [13–18]. In its new version, GK uses 192 radioactive cobalt-60 sources that are spherically arrayed in a single internal collimation system via collimator helmets to focus their beams to a center point. The tungsten collimators are organized into eight sectors of 24 sources each with three different apertures of 4 mm, 8 mm, and 16 mm, respectively. A highly conformal but inhomogeneous dose distribution and high central tumor dose can be achieved through the optimal combinations of the number, the aperture and the position of the collimators [14, 15, 22]. Traditionally, patients are placed in a rigid stereotactic frame achieving submilimeter accuracy in dose delivery. The dose is typically prescribed at the 50 % isodose to obtain the maximum dose at the center of each pinpointed target and the prescribed dose at target edge.

CK (Accuray, Sunnyvale, CA) is a relatively new technological device that combines a mobile linear accelerator mounted on a robotic arm with an image-guided robotic system [15, 16, 23, 47]. Patients are fixed in a thermoplastic mask and the treatment can be delivered as single-fraction or multi-fraction SRS. A variable number of overlapping beams (up to 200) are delivered non-isocentrically to the target, resulting in excellent dose coverage to the target and conformity. The set of beam directions and analysis of dose distribution are chosen through an inverse planning process. During the treatment, acquired oblique digital X-ray images of the patients are compared with digitally reconstructed radiographs (DRRs), which are obtained from planning CT images, and positioning errors corrected by translating and rotating the treatment table with an accuracy of less than 1 mm [15, 16].

LINAC is the most frequently used device for delivery SRS in the world and uses multiple fixed fields or arcs shaped using a multileaf collimator with a leaf width of between 2.5 and 5 mm [17, 18, 48–51]. Dose conformity can be improved by the use of intensity modulation of the beams (IMRS) or volumetric modulated arc therapy (VMAT), resulting similar to that achieved with the GK and the CK. Patients are usually immobilized in a high precision frameless stereotactic mask fixation system with a reported accuracy of 1–2 mm [48]; however, technically most advanced LINACs offer improved accuracy of patient repositioning with the use of on-board imaging systems with either orthogonal x-rays or cone beam CT (CBCT) that achieves an accuracy of less than 0.5–1 mm [17, 18, 50, 51]. The ExacTrac®X.ray 6D system uses a combination of two main subsystems: an infrared-based system for initial patient setup and precise control of either translational or rotational couch movements, and a radiographic kV X-ray imaging system for position verification and readjustment based on internal anatomy. A CBCT system utilizes either the megavoltage radiation beam delivered from the LINAC or a kilovoltage beam delivered using an additional x-ray tube mounted on the LINAC. During a single 360° scan rotation, the system produces a series of two-dimensional images of the entire volume of interest from multiple projection angles, which can be reconstructed in a three-dimensional data that can be directly compared with the CT planning study.

The superiority in terms of dose delivery and distribution for each of these techniques remains matter of debate. Despite several differences in treatments-related parameters among GK, CK and LINAC, there are no comparative studies demonstrating the clinical superiority of a technique over the others in terms of local control and radiation-induced toxicity for patients with brain tumors. Regardless of the technology used, a robust quality assurance (QA) program, encompassing all clinical, technical, and patient-specific treatment aspects, is mandatory to ensure the accuracy and safety of cranial SRS [52]. As stated by The World Health Organization, proper QA measures are imperative to reduce the likelihood of accidents and errors and increase the probability that the errors will be recognized and rectified if they do occur [52]. For brain SRS, detailed equipment specifications and tolerances, as well procedures that minimize the risk of errors and incidents have been reported by several professional organizations [52–57].

SRS is an effective treatment modality for patients with pituitary adenomas after unsuccessful surgery and/or resistant to medical therapy. Doses of 13–16 Gy are usually employed for nonfunctioning pituitary adenomas with a reported tumor control of 85–95 % at 5–10 years, whereas higher doses are commonly used for hormonally active pituitary adenomas. For secreting adenomas, normalization of hormone hypersecretion is reported in more than 50 % of patients at 5 years, being similar for doses of 20–25 Gy or > 25 Gy. Currently, the optimal dose to achieve biochemical remission of hormone-secreting adenomas remains to be determined. The majority of studies report on the use of GK SRS in patients with either nonfunctioning or secreting pituitary adenomas, whereas only few retrospective series show the results of LINAC SRS. In the respect of few series, the reported tumor control, biochemical remission of disease, and toxicity so far are broadly equivalent.

Hypopituitarism represents the most commonly reported late complication of treatment, whereas the incidence of other late effect radiation complications are low. In this regard, an accurate delineation of the target and surrounding structures is mandatory during the radiosurgical process; future studies need to incorporate precise dosimetric information of doses delivered to OARs to better understand the relationship between doses to OARs and development of hypopituitarism.

A few series suggest that multi-fraction SRS may be an appropriate treatment in patients with tumors in close proximity to the optic apparatus; however, the advantages of hypofractionated schedules in terms of local control and risk of radiation-induced toxicity as compared to single-fraction SRS remains to be proved. For large pituitary adenomas involving the optic apparatus, the use of fractionated stereotactic radiotherapy using a conventional fractionation (45–54 Gy in 25–30 daily fractions) is recommended. Several studies have shown a tumor control of 90–95 % for pituitary tumors of any size, including large or giant tumors, and hormone hypersecretion normalization of 50 % at 5 years [132–142].

In clinical practice, single fraction SRS is recommended for small-to-moderate sized pituitary adenomas (< 2.5–3 cm) even when the adenoma is close to the optic apparatus as long as the dose to the optic apparatus is kept below 8–10 Gy. Fractionated SRS, usually 25 Gy in 5 fractions, may represent a better treatment option when a single fraction dose carries an unacceptable risk of optic neuropathy (as for tumors adiacent the optic chiasm); however, studies with more patients and longer follow-up are required to draw definite conclusions. Fractionated stereotactic radiotherapy would be the recommended radiation treatment modality for lesions > 3 cm in size and/or compressing the anterior visual pathway.