Pituitary apoplexy: re-evaluation of risk factors for bleeding into pituitary adenomas and impact on outcome

Introduction

Pituitary apoplexy (PA) is defined as an acute clinical syndrome characterised by sudden onset of headaches, visual impairment and ophthalmoplegia due to haemorrhage with enlargement of a pituitary adenoma (1). The majority of reports on PA consist of single cases or small case series describing this clinically impressive condition; most of these reports, however, do not allow more rigorous data analysis including statistical evaluation. We found 11 retrospective studies in the last 29 years with a large number of patients with PA (≥16) (1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11). The results and essential information of these studies are summarised in Table 1. To the best of our knowledge, there are no previous case–control studies comparing potential risk factors, symptoms and outcome of patients with PA and pituitary adenoma patients without PA.

The aims of our study were to assess frequency, symptoms and outcome of PA among pituitary adenoma patients and to analyse potential risk factors for developing PA by means of a matched control group. Knowledge and information about potential risk factors could help to prevent PA in pituitary adenoma (and incidentaloma) patients with risk factors by providing an early indication for surgery.

Subjects and methods

We performed a retrospective chart review of patients with pituitary adenoma presenting to the Division of Endocrinology, Department of Internal Medicine, University Hospital of Zurich. Patients were identified by a search in the electronic database of our division with the following keywords: pituitary adenoma, pituitary tumour, hypopituitarism, prolactinoma, acromegaly and Cushing's disease.

Charts of 574 pituitary adenoma patients from 1980 to November 2007 were reviewed. Patients who suffered an apoplexy in their pituitary adenoma were identified. We diagnosed PA according to the clinical definition of Randeva et al. (5): presence of a pituitary adenoma together with at least one of the following criteria: episode of acute severe headaches and sudden visual disturbance such as visual field defects or ophthalmoplegia.

Patients with PA were compared with the total cohort of patients with pituitary adenoma during the same period. To analyse potential risk factors or associated factors for PA and to estimate the sequelae of PA, we compared PA patients with a control group (matched case–control study design). The control group was formed as follows: we assigned to each PA patient two pituitary adenoma patients without PA, who were matched for sex, tumour type, tumour size and age (downward order of relevance). For additional risk factors (i.e. oestrogens, antithrombotic therapy, diabetes mellitus, arterial hypertension, bilateral adrenalectomy, cardiac surgery and therapy with dopamine agonists) and outcome, we compared the PA patients with the control group (5, 12, 13).

All patients underwent computed tomography (CT) scan and/or magnetic resonance imaging (MRI) of the sellar region. In the PA group, 11 patients (26%) had a CT, 23 patients (55%) had an MRI, and 8 patients (19%) had both. In the control group, 26 patients (31%) had a CT, 43 patients (51%) had an MRI, and 15 patients (18%) had both. Tumours measuring <1 cm in diameter (as determined by neuroimaging) were classified as microadenomas and larger tumours as macroadenomas. Tumour type was mostly determined on histology, but sometimes histological analysis only showed necrosis. In these cases, tumour type was defined by clinical presentation and endocrine studies at diagnosis. Clinical presentation of gonadotrophin-secreting tumours is similar to that of clinically non-functioning tumours. Therefore, these tumours were included under the term of clinically non-functioning tumours. Tumours with no further characteristics or endocrine activity were assigned to that group, too. Endocrine function was evaluated at the time of admittance to the hospital (before medical and surgical treatment), as well as 12 months after surgery. Secondary adrenal failure was defined by the perceived need for glucocorticoid replacement therapy with concomitant low (in the majority <200 nmol/l) serum cortisol levels in the early morning. Secondary hypothyroidism was defined by repeatedly low (usually <10 pmol/l) serum-free thyroxine levels recorded previously. We defined male hypogonadism as having low serum testosterone (usually <12 nmol/l) and gonadotrophin levels (usually FSH and LH both <2 mU/ml), and female hypogonadism as having amenorrhoea and low oestradiol (usually <70 pmol/l) without rising gonadotrophin levels (usually FSH and LH both <5 mU/ml).

Median and range were used for descriptive statistics. Fisher's exact test was used to analyse categorical variables and groups. The McNemar test was used to analyse differences in categorical variables within the group (14). Multivariate logistic regression analyses adjusting for age, sex and tumour size were conducted with the whole group. A P value of <0.05 was considered statistically significant. All statistical analyses were performed using SPSS version 16.0 (SPSS Inc., Chicago, IL, USA).

Results

Of the 574 patients, 42 (7.3%) with pituitary adenoma fulfilled the diagnostic criteria for classical PA as defined above. In 41 of the 42 patients with classical PA, pituitary adenoma had been unknown before PA occurred. The median age of the PA patients was 53.5 (range 21–85) years. The median age of the control group was 50.0 (range 23–84) years.

Discussion

In our retrospective study, 7.3% of patients with pituitary adenoma presented with PA. In previous studies, the incidence of PA in patients with all kinds of pituitary adenoma amounted to 1.6–12.8% (1, 3, 4, 5, 10, 11).

In contrast to some case reports demonstrating that PA may also occur in small pituitary tumours, we postulate that a large tumour size is associated with a significantly increased risk for PA, because all our patients with PA had a pituitary macroadenoma (13, 15). This finding is in line with the study of da Motta et al. (4), in which all patients had a pituitary tumour with important suprasellar extension. An additional major risk factor in our study was male sex. Men suffered significantly more frequently from PA than women, irrespective of tumour size. Some, but not all studies, report that male sex is an independent risk factor for PA (1, 4, 5, 8, 11, 12, 13, 16, 17). Yet another major risk factor represents tumour type. In agreement with some but not all previously published studies, the incidence of PA was significantly higher in our patients with clinically non-functioning tumours (1, 2, 4, 5, 6, 8, 11, 12). Wakai et al. (1) did not find a significant difference in the incidence of PA among different tumour types in their study of 560 pituitary adenoma patients and 51 PA patients. However, da Motta et al., Randeva et al., Sibal et al. and Semple et al. (4, 5, 6, 8) reported an increased risk for PA in clinically non-functioning tumours (between 60 and 77%).

Mechanisms for the development of PA are not fully understood. As discussed by Bjerre et al. (18), blood supply to the anterior pituitary lobe is provided by portal vessels through the infundibulum. As the perfusion pressure is very low in these portal vessels, pituitary adenomas are particularly susceptible to even minor increments in intrasellar pressure caused by a tumour (18). Beside an elevated intrasellar pressure, a further important factor for developing PA could be the degree and type of vascularisation of the tumour. Some authors suggest that pituitary adenomas can outgrow their blood supply resulting in ischaemic necrosis followed by haemorrhage (5, 8). Cardoso & Petersen (17) have postulated that an intrinsic vasculopathy in pituitary adenomas renders them more susceptible to infarction and haemorrhage. McCabe et al. (19) found a markedly raised vascular endothelial growth factor (VEGF) mRNA expression in the tissue of clinically non-functioning tumours compared with other types of pituitary tumours and normal pituitaries, which could indicate that angiogenesis may be different, and that properties of tumour vessels could contribute to an increased risk for PA in clinically non-functioning tumours. The results of these studies, including ours, are consistent with proposed theories on the pathogenesis of PA: beside male sex, large tumours and non-functioning tumour type are major risk factors for development of PA.

In addition to these three major risk factors, other potential risk factors have been discussed in the literature: arterial hypertension, sudden changes in arterial blood pressure, diabetes mellitus, head trauma, transient elevations of intracranial pressure, cardiac surgery, dynamic tests of pituitary function with releasing factors, antithrombotic therapy, oestrogens, dopamine agonists, somatostatin analogues and radiotherapy (5, 12, 13). Of these risk factors, we could confirm antithrombotic therapy (vitamin K antagonist or platelet inhibitors) as significant in our study.

Suffering a head trauma may also be associated with an elevated risk for PA, but because of its rare occurrence (three in the PA group and none in the control group), statistical analysis was not undertaken. The mechanism of post-traumatic PA is not yet fully understood. Bao et al. (20) speculate that the change in blood flow in pituitary adenomas due to fluctuations of intracranial pressure and blood pressure following severe head injury leads to the apoplectic event in a pituitary adenoma.

Similarly, statistical analysis was not possible for oestrogen treatment in female patients with prolactinoma due to the small sample size. Oestrogens may enhance growth and activity of prolactinoma cells, may increase perfusion demand and stimulate tumour vascularisation. Furthermore, oestrogen treatment was reported to stimulate the production of VEGF in cells isolated from prolactinoma (21).

Treatment with dopamine agonists in prolactinoma patients, by contrast, did not appear to be harmful in our study. Given that these agents decrease growth and activity of prolactinoma cells, we suggest that dopamine agonist treatment should no longer be included in the list of risk factors for PA, despite previous case reports suggesting the opposite (3, 12).

Diabetes mellitus and arterial hypertension were not found to be significant risk factors for PA in our study suggesting that the overall cardiovascular risk profile (which predicts stroke) does not predict bleeding into a pituitary adenoma.

Few studies address the long-term outcome of patients with PA. In our study, most patients recovered from ophthalmoplegia, whereas outcome in patients with visual field defects was worse. Since the three ocular motor cranial nerves (III, IV and VI) are peripheral nerves, they can undergo regeneration. The optic nerve as part of the white matter cannot recover after axonal disruption. In the studies by Bills et al. and Onesti et al. (2, 22), visual outcome was also better in patients with ophthalmoplegia than in those with visual field defects.

Concerning the endocrine outcome, an event of PA may lead to irreversible loss of pituitary cells, thus leaving the majority of our PA patients with (at least partial) pituitary insufficiency. These observations are in line with some other studies (10, 23). Zayour et al. (23) speculate that the rapid increase in intrasellar contents after haemorrhagic infarction of a pituitary adenoma may lead to a sudden increase in intrasellar pressure, resulting in ischaemic necrosis of the anterior pituitary and limiting the potential for functional recovery after decompression. However, Liu et al. (11) have found that some recovery of hormone deficiencies in their PA patients occurred following surgery.

With increasing use of CT scans and MRI studies, incidentally discovered pituitary adenomas become more frequent. Most pituitary adenoma patients are assigned for treatment, especially when a clinically active adenoma or threatening visual field defects by a large macroadenoma is found. Based on our findings, we cannot comment on the best management of such patients, but our observations may be helpful in management decisions for patients with pituitary macroincidentalomas. According to our results, the threat of bleeding is highest in large, clinically non-functioning tumours in patients who are taking antithrombotic therapy for prevention or treatment of atherosclerosis and its complications. Therefore, among patients with pituitary incidentaloma, we would propose to treat those with macroadenoma and atherosclerosis by surgery even if there are no tumour-related signs and symptoms, to prevent PA.

Pituitary adenomas are rarely known at the time of presentation of PA; it is therefore not surprising that all epidemiological data on PA derive from retrospective studies with their obvious drawbacks. Apart from being retrospective, another limitation of our study is its long period of time. The quality of documentation of some elder cases was variable, and although neuroimaging by MRI or CT scan was performed in all patients, the exact tumour size and tumour infiltration were not always described in sufficient detail. Some tumours were possibly overestimated in size because of bleeding. However, in 36 of our 42 PA patients (86%), suprasellar and/or parasellar extensions were documented by brain imaging studies. Among the remaining six PA patients, four had a tumour diameter of 2 cm and only two had too fragmentary data about tumour dimensions. Therefore, we feel confident that the vast majority of our PA patients had macroadenomas before the event of PA. Without precise assessment of the tumour size in all patients, it was not possible to prove whether large tumour size and non-functioning tumour type were independent risk factors for PA. It can be argued that non-functioning tumours may be diagnosed at a later stage of disease due to late compression symptoms caused only by large tumours. Unfortunately, as long as we have only crude figures about tumour size (i.e. tumour size <1 cm or more than 1 cm) in some of the patients, we are unable to further determine any independency between these two risk factors. Moreover, concerning the endocrine status of the patients, we had only limited data on insulin hypoglycaemia stress tests, considered as gold standard for the determination of secondary adrenal failure. However, serum basal cortisol levels in the early morning may be used as first-line test in the assessment of the hypothalamic–pituitary–adrenal axis both preoperatively and postoperatively, as well (24).

The strengths of our study are both the large number of included patients and the matched case–control study design comparing patients suffering from PA with a group of patients having comparable tumours without PA from the same institution and time period.

We conclude that the risk for PA depends mainly on properties of the tumour itself (tumour size and type) and on the patient's sex. Antithrombotic therapy and possibly also head trauma and oestrogen treatment in patients with prolactinoma may enhance or trigger such an event. Over the following months, most patients recovered from ophthalmoplegia but few from pituitary failure. Early diagnosis of partial pituitary failure by careful history and examination may help to discover pituitary adenoma in some – especially male – patients at an earlier stage and thus reduce the risk for PA.

Declaration of interest

The authors declare that there is no conflict of interest that could be perceived as prejudicing the impartiality of the research reported.

Funding

This research did not receive any specific grant from any funding agency in the public, commercial or not-for-profit sector.

Author contribution statement

D L Möller-Goede wrote the manuscript and was responsible for data collection. C Schmid, M Brändle and K Landau revised the manuscript. M Brändle and D L Möller-Goede calculated the statistics. K Landau made most of the ophthalmological examinations and R L Bernays performed most of the surgeries. All the authors contributed to the quality of this manuscript by helpful critical review of the results and the discussion.