Introduction

Posterior pituitary tumors (PPTs) are very rare neoplasms, with less than 300 cases reported to date [1, 2]. They are low-grade nonneuroendocrine sellar/suprasellar tumors originating from pituicytes, specialized glial cells of the neurohypophysis [3]. Four pathological types have been reported: pituicytoma (Pi), granular cell tumor (GCT), spindle cell oncocytoma (SCO), and sellar ependymoma (SE). These tumors are immunohistochemically characterized by expression of thyroid transcription factor-1 (TTF-1) as described in fetal and adult pituicytes but not in neurosecretory cells of the adenohypophysis [4]. As evidenced by a negative immunohistochemical staining for all adenohypophyseal hormones and ultrastructural microscopic characteristics, pituicytes do not have the ability to synthesize and secrete hormones.

PPTs are usually diagnosed in the 5th decade of life without sex predilection. Pi is the most common histological type (approximately half of the cases), followed by GCT (~25%), SCO (18%), and SE (3%) [1]. They usually manifest as visual disturbances, headaches with or without hypopituitarism. The prevalence of diabetes insipidus is very low at clinical presentation [1, 5]. However, a nonnegligible percentage of patients have been associated with hypercortisolism [5,6,7,8,9,10,11,12,13,14,15,16,17] or acromegaly [5, 11, 12, 17, 18]. This association is relatively high given the low incidence of both diseases [1].

The objective of this article is to review and describe in detail the cases of Cushing’s disease (CD) and acromegaly associated with PPT until now. Selection criteria and data analysis were performed as previously published [1]. All cases that met WHO criteria for PPT diagnosis and clinical and hormonal data for CD and acromegaly were selected. In addition, cases reported in PubMed’s database in the last 2 years until March 2020 were also included. Further, we analyze the possible different pathogenic mechanisms of this association.

Cushing’s disease and posterior pituitary tumors

Only 15 cases of hypercortisolism associated with PPTs have been reported so far, all of them in the last 8 years [5,6,7,8,9,10,11,12,13,14,15,16,17] (Table 1). This figure indicates a prevalence of 5.6% of all PPTs [1]. Most of them (n = 13, 86.7%) were Pi and the rest GCT. To date, no clinical reports of CD associated with SCO or SE have been reported.

Table 1 Summary of reported clinical cases of ACTH-dependent hypercortisolism and Cushing’s disease associated with posterior pituitary tumors
Full size table

CD is a very rare disease with an incidence of 6.2–7.6 per million person years [19]. For this reason, a prevalence of CD of 5–6% in PPT patients is very high, allowing us to consider that their relationship is probably not just coincidental.

The first case was reported by Schmalisch et al. [6]. He was a 48-year-old man with CD and a 4 mm microadenoma image in pituitary magnetic resonance imaging (MRI). Pathological study after transsphenoidal surgery (TS) showed an incidental Pi. After surgery, hypercortisolism persisted and a right hemi-hypophysectomy was performed. The patient developed adrenal insufficiency, indicating the remission of CD. A corticotropic adenoma (CA) could not be demonstrated in the pathological study.

Between 2013 and 2018, another nine patients (eight women and one man) with clinical and biochemical characteristics of ACTH-dependent Cushing’s syndrome (CS) were reported. Seven of which had pituitary adenoma (PA) on pituitary MRI [7,8,9,10,11,12,13]. In all cases, the pathological study showed an incidental Pi. Six of the nine patients achieved remission of CD after surgery, although only two of them presented histopathological findings compatible with CA. In the remaining three patients, hypercortisolism persisted after surgery, and were treated with radiotherapy (n = 2; one of them with corticotropic hyperplasia, CH) and surgery in another patient. In the latter, the presence of a CA could be demonstrated in the pathological specimen.

The association of GCT with CD was reported for the first time in 2018 by Zhang et al. [14] in a 32-year-old man with CD and normal pituitary MRI. Pathological study after TS showed an incidental GCT without evidence of PA or corticotroph hyperplasia (CH). After surgery, the patient showed a normal serum cortisol rhythm and urinary free cortisol levels without tumor recurrence.

In 2019, the remaining four patients (three women and one man) with CD associated with PPT were reported [5, 15,16,17]. Pituitary surgery demonstrated an incidental Pi in three of the patients and a GCT, also incidental, in the other patient. In the three patients with Pi, histopathological lesion of CA or CH was not demonstrated, while the patient with GCT showed CH. All patients with Pi went into CD remission after surgery. Whereas, GCT patient showed CD persistence and ketoconazole was began to control hypercortisolism.

A more detailed analysis of this series of 15 patients indicates that all PPT patients showed a hormonal study compatible with ACTH-dependent hypercortisolism (Table 1).

Histological lesions associated with ACTH hypersecretion could not be demonstrated in more than half of the patients (n = 9, 60%). However, five patients (33.3%) showed evident histological lesions [three CA [8, 11, 13], one was discovered in the second TS [8], and two CH [5, 9]]. Another patient (patient #1, Table 1) (6.7%) presented a nontumorous specimen of the adenohypophysis cells with signs of Crooke’s hyalinization compatible with CD [6]. The majority of patients (n = 9, 60%) presented remission of CD or hypercortisolism symptoms after surgery, regardless of whether (n = 2) or not (n = 8) histological lesion was found. In the remaining five patients with CD persistence, two of them remitted (one after a second surgery and the other after pituitary radiotherapy), while in another two patients hypercortisolism was controlled with bilateral adrenalectomy in one of them and with ketoconazole in the other. The remaining patient was treated with radiosurgery without follow-up data.

Present data suggest that ACTH-dependent hypercortisolism due to CD, is associated with relative frequency to PPTs. As occurs with sporadic CD, it is more common in women, but in this case with a lower female:male ratio (2.5:1 vs 5–10:1). Mean age at diagnosis was 41.3 ± 15.1 year (range, 7–62 year). In this series men were slightly younger than women (35.2 ± 15.1 vs 43.5 ± 16.5 year, ns). No lesion related to CA or CH was detected on imaging in any of the PPT patients, which occurs in about 40% of sporadic CD [20]. After the first surgery, no histological lesion was observed in more than half of the cases. When present, the most common demonstrated histological lesion is CA, followed by CH. The absence of histological lesion in the surgical specimen is relatively frequent in patients undergoing CD surgery. In fact, negative histology for PA has been reported in 12.5% of patients with CD that underwent TS, those of which 69.4% were in remission after surgery [21]. CD remission in the cases of negative histological study could be explained by a suction of the ACTH-secreting pituitary microadenoma during the surgical procedure.

There are some distinguishing features between PPTs associated with or without CD. In contrast to sporadic PPTs (22 ± 14.2 mm) [1], those associated with CD are usually micro-tumors (<10 mm), all of which have been incidentally discovered in the histological study. Moreover, sporadic PPTs have a higher Ki67 index (1–7% vs 1–2%) [5]. The reason for these differences is unknown, and therefore needs further investigation. Lastly, all PPTs associated with CD went into remission after surgery without recurrence, differently to sporadic PPTs in which the recurrence rate is around 5% [1]. Clinical features of PPT associated with CD are shown in Table 2.

Table 2 Clinical features of posterior pituitary tumors (PPT) associated with Cushing’s disease (CD) or acromegaly
Full size table

Acromegaly and posterior pituitary tumors

Although less frequently observed than with CD, acromegaly has also been described in association with PPTs (Table 3). Until now, only six acromegalic patients have been reported, all of them in the last 10 years [5, 11, 12, 17, 18]. Therefore, and according to the published series, the estimated prevalence of acromegaly associated with PPT is 2.3% [1]. Most of them (n = 4, 66.7%) were Pi and the rest GCT. To date, there is no report of acromegaly associated with SCO or SE. Although this prevalence may appear to be low, as it occurred with CD, the low prevalence of acromegaly in the general population (2.8–13.7 cases per 100,000 people) [22], together with the small number of patients described with PPT (<300 cases) [1] lead us to consider it as a nonincidental association. The cases described to date are summarized below (Table 3).

Table 3 Summary of reported clinical cases of acromegaly associated with posterior pituitary tumors
Full size table

The first clinical case was reported by Losa et al. [18]. She was a 48-year-old woman with 10 years of GH excess symptoms. Endocrinological evaluation confirmed the diagnosis of acromegaly and MRI revealed an 11 × 11 × 9 mm pituitary lesion. She underwent TS, achieving a subtotal resection. The histological study showed a GCT without somatotroph adenoma (SA) or somatotroph hyperplasia (SH). After surgery, acromegaly persisted and she underwent TS reoperation obtaining the same histological result as previous surgery. At follow-up, remission of acromegaly was demonstrated.

Eighteen years later, Chang et al. [11] reported a 46-year-old male patient who complained of acromegaly symptoms. The diagnosis was confirmed after hormonal study. The MRI revealed an 8.7 × 11.9 × 7.5 mm pituitary lesion. He was operated via TE approach with total resection of the lesion. The histological study showed a Pi with no evidence of SA or SH, achieving the Pi cure and acromegaly remission.

In that same year, Feng et al. [12] reported two other women (56 and 65 years of age), diagnosed with acromegaly and macroadenoma on the pituitary MRI. Both underwent TS. In the first patient, the histological study was Pi without lesions compatible with PA or SH, and acromegaly persisted after surgery. In the second patient, a subtotal resection of the lesion was performed by TS. The histological result was a GCT without SA or SH data. After surgery, acromegaly persisted and a somatostatin analog (SSA) was prescribed.

The last two patients were reported in 2019 [5, 17]. The first of them was a 29-year-old acromegalic woman with pituitary macroadenoma undergoing endoscopic endonasal transsphenoidal surgery with subtotal resection. The pathological result was Pi and SA. After surgery, acromegaly persisted, and patient was treated with radiosurgery [17]. The second patient was a 70-year-old acromegalic woman with a pituitary MRI compatible with SA. She underwent TS with complete tumor resection. Pathological study showed an incidental Pi without SA or SH. Postoperative course revealed acromegaly remission.

This series of acromegalic patients with PPT suggests that this association is more frequent in women (female: male ratio, 5:1), unlike what occurs in cases of sporadic acromegaly, in which the distribution by sex is similar [22]. Mean age at diagnosis was 51.3 ± 15.3 year (range, 29–79) similar to sporadic cases [22]. All but one of the six patients showed macroadenomas (≥10 mm) in pituitary MRI. However, these lesions corresponded histologically with PPTs and not with SA. Only one patient (#5, Table 3) of the six cases showed a PA in the histological study. There was no case described as SH. Acromegaly persisted in four patients (#1, #3, #4, and #5) after the first surgery. One of whom (#1) was cured after a second surgery, although the histological study did not show SA or SH. Another patient (#4) was controlled with SSA. Surprisingly, two of the six patients (#2 and #6) without histological lesions compatible with SA or SH, remitted after the first surgery. This indicates a possible pathogenic role of PPT in the development or maintenance of GH hypersecretion.

Interestingly, and as it occurs with CD, PPTs associated with acromegaly are smaller (14.4 ± 3.5 mm) than sporadic PPTs (22.0 ± 14.2 mm) all of which were incidentally discovered [5]. The most common PPT associated to acromegaly was Pi. However, GCT was higher in patients with acromegaly (33.3%) compared to sporadic PPT group (25.6%). In this population, all patients with PPT were cured after the first surgery, except for one patient who required second TS. Clinical features of PPT associated with acromegaly are shown in Table 2.

Hyperprolactinemia and posterior pituitary tumor

Hyperprolactinemia is the most frequently anterior pituitary hormone disorder associated to PPTs. Hyperprolactinemia appears in about 40% of patients. It is generally mild but some cases with galactorrhea have been reported [1, 5].

To our knowledge, no histologically proven cases of prolactinoma associated with PPT have been reported so far. In 2015, a case of a giant prolactinoma was described in a 25-year-old woman who underwent surgery. Pathological study revealed a prolactin-secreting PA with atypical spindle cell morphology. However, the immunohistochemical study was positive for prolactin, chromogranin A, synaptophysin; and negative for glial fibrillary acidic protein, S-100 protein, epithelial membrane antigen, and vimentin [23].

The lack of prolactinoma associated with PPT may be related to the fact that surgery is rarely performed in mild cases of hyperprolactinemia and PA, as these patients are easily controlled with dopamine agonists.

Pathophysiological mechanisms

Different mechanisms might explain the association of adenohypophyseal hypersecretion syndromes with PPTs.

In relation to hyperprolactinemia, the responsible mechanism seems to be the compression of the pituitary stalk and the consequent reduction of inhibitory dopaminergic tone on anterior pituitary lactotroph cells.

Regarding the association of CD and acromegaly with PPT the explanation is more difficult and complex. The association of PPTs with CD is more common than with acromegaly. However, in the general population the incidence and prevalence of CD is lower than acromegaly. Different pathogenic mechanisms could be considered (Table 4).

Table 4 Possible pathogenic mechanisms related to the development of adenohypophyseal hyperfunction syndromes in posterior pituitary tumors
Full size table

PPT cells may have hormonal secretory capacity acting as neuroendocrine cells with hypersecretion of ACTH or GH in some patients. This hypothesis should be completely ruled out since it has been recently shown that PPTs derive from pituicytes, specialized cells of the glia of the posterior pituitary gland [3]. Although morphological features of endocrine differentiation (secretory granules) have been reported in a case of Pi [24], the expression of adenohypophyseal hormones in cells from PPTs have not been demonstrated [1, 4, 6, 10,11,12, 16, 18].

Cells from PPTs might induce stimulation signals (cytokines and/or growth factors) and cell proliferation on adjacent adenohypophyseal neurosecretory cells facilitating the development of hypersecretion syndromes with [8, 13, 17] or without PA [1, 6, 7, 10,11,12, 14,15,16,17,18]. In support of this argument, some patients with PPT hypersecretion syndromes, PPT were found in the location where the different neuroendocrine neurohypophyseal cells (ACTH- or GH-secreting cell region) are commonly placed [11]. This action could be due to a permissive effect of hypothalamic adenohypophyseal hormone-releasing hormones or through proliferative paracrine signals that could change the microenvironment of the anterior pituitary (local irritation) facilitating the development of hyperactivity adenohypophyseal cells [18]. If this hypothesis was to be true, the glia cells of the posterior pituitary lobe might play an important role in the development of adenohypophyseal hyperplasia [9] or oncogenic differentiation of the anterior pituitary gland [8, 12, 13, 17]. However, to date, no potential stimulatory signal or factors for adenohypophyseal hormone secretion from PPT cells have been identified.

Other authors suggest that Pi can regulate hypothalamic releasing hormones (CRH or GHRH) or different ACTH- or GH-releasing factors [18, 25, 26]. In this setting, some studies have reported that most PPTs patients with neuroendocrine secretion show immunohistochemical (IHC) staining findings that indicate specific hormone change in their normal pituitary glands [11]. This would explain, at least in part, the remission of CD and acromegaly after PPT removal in which no PA was detected. On the other hand, the IHC study for GHRH performed in a patient with acromegaly and GCT was negative [18], indicating that the pathogenic role of hypothalamic releasing hormone hypersecretion in PPT-associated adenohypophyseal hypersecretion syndromes is not clearly defined.

Another possibility could be the coexistence of a functioning PA with PPT. As discussed above, given the rarity of the association of both tumors, it seems unlikely to be coincidental [8]. Therefore, it could be possible that the glia cells of the posterior pituitary might play a pathogenic role in the tumorigenesis of the anterior pituitary. However, there is no clear explanation that supports that PPT can develop a functioning PA, although this possibility cannot be totally ruled out. The explanation of a possible common origin of both tumors from the same precursor cell has been suggested by some authors [8]. Other authors have described ultrastructural features intermediate between a Pi and a PA, suggesting that Pi might also arise from the specialized stromal folliculo-stellate cells (FSCs) of the adenohypophysis, which would be able to differentiate into endocrine cells [24]. FSCs could act as stem cells, by differentiating into endocrine cells and expressing the oncoprotein B-cell lymphoma-2 that could contribute to the progression of various tumor types, including adenomas and PPTs [27]. However, this explanation seems to be unlikely due to the different biomolecular profile recently reported in both tumors [10, 28]. In fact, TTF-1 expression is a feature of PPTs and not FSCc [4, 29, 30]. Lastly, we should consider, in cases of PPT and adenohypophyseal hyperfunction syndrome that remits after PPT surgery without histological demonstration of PA, the possibility that the PA can be aspirated during surgery, a situation that frequently happens in the EC.

Another possible explanation could be the development of an ectopic functioning PA located in the posterior pituitary, as it has been occasionally reported [31]. However, this situation is not plausible in the cases reported here, because PPT was confirmed histologically in all of them.

Lastly, a stimulation of the corticotroph cells of the intermediate lobe by PPTs could explain the development of CD in some PPT patients [32].

Interestingly, unlike Pi and GCT, SCO have never been associated with CD or acromegaly. The reason for this lack of association is unclear given the common origin with the other PPTs. It is true that the number of SCOs published until now is very low (n = 47; 17%) [1]. We cannot rule out that with an increasing number of SCO, new cases of CD or acromegaly could emerge, although this is merely speculative.

Conclusions and future perspectives

In summary, PPTs are rare neoplasms originating from the glia cells of the neurohypophysis. Although cases reported to date have been few, the association of these tumors with hormone hypersecretion syndromes is higher than what would be expected by coincidence. Furthermore, in reported associations some peculiarities stand out, such as the preference for the female gender, the predominance of Pi, the smaller size of PPTs compared to sporadic ones, and the incidental nature of all PPTs. Cure rates of PPTs after surgery has been reported to be high, although no pathological imaging of PA was found in most patients. Nonetheless, hormone hypersecretion remission was not always achieved, especially in the cases of patients with acromegaly.

The structural and functional relationship of pituicytes with adenohypophysis cells and the intermediate pituitary lobe should be further studied. In this context, it would be helpful to carry out cytological and IHC studies in cells from PPT to analyze the presence of substances such as cytokines, different growth factors, ACTH- or GH-releasing factors that could stimulate the secretion and growth of corticotropic and somatotropic cells. This could demonstrate that neoplastic glial cells of the neurohypophysis play a role in hormonal secretion and oncogenic differentiation of some types of adenohypophyseal cells. Therefore, the focus should be on investigating the effect of neurohypophyseal glial cells on the secretory and proliferative function of anterior pituitary cells.

Our current knowledge is limited and further molecular, translational and clinical investigation on PPTs is warranted.