GPA is a rare autoimmune disease characterized by necrotising granulomatous vasculitis that mainly involves the upper and the lower respiratory tracts and the kidneys, although any organ may be affected [1, 12]. One-third of patients with GPA have a neurological engagement, primarily in the form of peripheral neuropathy due to small-vessel vasculitis. Central nervous system involvement occurs in 7–18% of all cases . First described in 1953 , PD in GPA is rare and present in about 1% of all GPA patients . In our catching area (western Sweden) approximately 250 to 300 patients have GPA. To our knowledge, of those, only the patients presented in the current report have developed PD, i.e. the prevalence is in agreement with previous estimates .
Inflammatory and granulomatous lesions in the pituitary gland are rare causes of PD [14, 15]. Of these, lymphocytic hypophysitis is the most common. Pituitary involvement in GPA is a form of a so called “secondary hypophysitis”, a term first introduced for more than 20 years ago to indicate a pituitary inflammation resulting from systemic diseases or pituitary near-lying lesions . Other causes of granulomatous pituitary lesions are sarcoidosis, tuberculosis, idiopathic giant cell granulomatosis, Crohn´s disease and Takayasu´s vasculitis . To date, only two case series and 34 cases have been published describing GPA-related PD, resulting in 51 patients reported in the English medical literature.
The pathophysiology behind PD in GPA is not entirely known. Three main pathogenic mechanisms are thought to be involved [5, 17]. First, and the most commonly implicated mechanism, is a direct intracranial extension of the granulomatous process from the nasal or paranasal granuloma to the pituitary gland. The second one is a vasculitis of the pituitary vessels that results in ischemic and/or hemorrhagic infarctions. The third possible mechanism is a development of new granulomas in the pituitary gland itself.
Clinical features of PD are rarely present at onset of GPA. In fact, in a recent case series, only one of nine patients had PD as a presenting symptom . Instead, the diagnosis of PD more commonly occurs subsequently, between 2 months to 15 years after the diagnosis of GPA. Worth to note is that the symptoms of PD in GPA may be nonspecific including weakness, headache, or fatigue, which may lead to underdiagnosing or a delay in the diagnosis.
The three cases reported here, all women aged between 32 and 37 years, share a similar clinical profile with patients previously reported in the literature (Table 1). PD was the presenting feature in one (Case 2) and two developed PD in the course of the disease (Case 1 and 3). Yong et al. reviewed 22 cases published in the English medical literature between 1966 and 2006 . The mean age at diagnosis of PD in GPA was 38 years (range 13–71) and 74% of patients were female. Two recent case series described patients with a somewhat different profile, though. In the French Vasculitis Study Group cohort the mean age at the onset of PD in nine patients with GPA was 51 years (range 24–77), five were women and four men . Kapoor et al. described eight cases with a mean age of 48 years (range 28–68), with a 1:1 male to female ratio .
One or more pituitary axes functions can be affected in PD caused by GPA. Consequently, every pituitary hormone function should be investigated. One of the most common clinical entities is DI, isolated or associated with anterior PD [6,7,8,9,10]. Conversely, anterior PD, without coexisting DI is uncommon. De Parisot et al. recently reported that secondary hypogonadism is the most common endocrinopathy in pituitary GPA, closely followed by DI (detected in 78 and 71% of all cases, respectively) . This was in agreement with a previous case series, showing hypogonadism in 88% and DI in 75% . The pathogenic mechanism behind hypogonadism remains unclear and is presumably multifactorial. Several factors can affect the hypothalamic-pituitary-gonadal axis, including hyperprolactinemia, the chronic disease itself and pharmacologic treatment. It is well known that treatment with both glucocorticoids and cyclophosphamide can provoke hypothalamic-pituitary-gonadal axis suppression . The high rate of hypogonadism due to GPA itself may therefore be overestimated. It is unclear whether hypogonadism reported in the previous case series was diagnosed at onset of GPA or developed after the treatment was initiated. All our patients had DI and only one developed hypogonadism (Case 1) and we cannot exclude that it was related to treatment with high dose of steroids and cyclophosphamide.
Other hormone deficiencies in GPA-related pituitary dysfunction are not as common as DI and hypogonadism. Low levels of free thyroxine, consistent with diagnosis of central hypothyroidism, have been reported in 54% of patients. Secondary adrenal insufficiency occurs in 39% of cases and growth hormone deficiency (GHD) in 20% . Finally, hyperprolactinemia is reported in 37% of patients, probably due to pituitary stalk compression. The prevalence of GHD in GPA-associated PD is probably underestimated. In fact, the diagnosis of GHD in the previous reported cases was based on abnormally low IGF-1 and not on dynamic testing. For the first time we report a patient who was diagnosed with GHD through an insulin tolerance test, the gold standard for diagnosing GHD . GHD in this case developed 2 years after being diagnosed with GPA despite remission of the systematic disease. In granulomatous pituitary lesions of other causes than GPA, GHD is reported to be the most common endocrinopathy followed by hypogonadism, central hypothyroidism and secondary adrenal insufficiency . On the contrary, the prevalence of GHD in pituitary GPA is reported to be seen in only 20% of the patients. However, none of the patients described in the medical literature had any dynamic testing. Therefore, we recommend performing dynamic stimulation tests for evaluating GHD, if clinical and biochemical findings are suspicious for this hormone deficit.
Visual field test is an important tool in order to complete the diagnostic work-up in pituitary GPA. Thus, all patients with a sellar mass should undergo visual field evaluation. Kapoor et al.  disclosed visual deficits in 40% of their patients, compared to 17% reported in previous literature . In contrast, none of our patients presented with abnormalities on visual field testing.
PD is frequently diagnosed on the basis of head imaging, commonly performed to investigate headache and/or sinusitis in patients with GPA. The most frequent finding on MRI in patients with pituitary GPA is a sellar mass with peripheral enhancement and a central cystic hypointense lesion. Absence of the posterior hyperintense signal on T1-weighed images and a thickening of the pituitary stalk are also common [18, 21]. The most frequent pituitary diseases in general are pituitary adenomas. Although pituitary adenomas are usually isointense on MRI, some have a similar appearance as patients with GPA, i.e. peripheral enhancement around a hypointense lesion. However, pituitary adenomas only very rarely present with DI and other diseases such as GPA, hypophysitis, sarcoidosis, Langerhans cell histiocytosis and metastasis should be carefully considered in patients with new onset polyuria and polydipsia .
Two of our patients (Case 1 and 2) presented with a typical pituitary cystic lesion with peripheral enhancement. One (Case 3) had a sellar mass with homogeneous contrast enhancement. The correct diagnosis of pituitary lesions is essential to avoid unnecessary biopsy or surgical treatment, which is not primarily recommended in pituitary GPA, given the excellent response to treatment with immunosuppressive therapy.
Treatment and outcome
Conventional remission-induction treatment for PD in GPA consists of high-dose glucocorticoids combined with oral or intravenous cyclophosphamide, with a reported remission in two-third of patients [5, 11]. De Parisot et al. reviewed 51 cases in English and French medical literature, showing that 69% of all patients were treated with conventional treatment. After a median follow-up of 5 years, 11% had a relapse of systemic disease . Induction treatment without cyclophosphamide is associated with a relapse in 50% of the patients after a median follow-up of 4.5 months . Thus, cyclophosphamide seems to be more effective in pituitary GPA than other immunosuppressive treatments. Patients that are refractory to the conventional therapy can be successfully treated with rituximab, approved in severe ANCA-associated vasculitis . Recent trials have shown that GPA patients treated with rituximab achieve complete remission more frequently than those who are treated with cyclophosphamide . These findings suggest that clinicians may consider the use of rituximab as first-line therapy . However, treatment experience with rituximab for PD in GPA is limited and further studies are needed [3, 5].
All our patients were treated with glucocorticoids and cyclophosphamide. Two patients had a complete resolution of the pituitary mass (Case 1 and 2) and one had a partial resolution. However, all our patients had a persistent PD. It has been reported that 65% of all cases have a complete or partial resolution of the pituitary lesions on MRI assessment, after the treatment . Interestingly, there is no relationship between hormonal, radiologic and systemic outcomes, thus follow-up should include both imaging and pituitary function assessment. Despite a high rate of systemic disease remission and regression of radiological abnormalities, the outcome of pituitary function is less favorable. In fact, De Parisot et al. reported that 86% of the patients have permanent deficiency of one or more of the pituitary hormones, probably due to a permanent pituitary damage related to necrotizing granulomatous inflammation of the gland .
No data are available concerning long-term outcome in pituitary GPA. Patients with hypopituitarism, irrespective of underlying aetiology, have increased morbidity and mortality rates compared to the general population [25, 26]. Whether patients presenting with PD in the context of GPA have a more aggressive disease burden per se, and to what extend this could affect the overall prognosis in these patients is unknown.