Introduction

Granulomatosis with polyangiitis (GPA) is characterized by necrotizing granulomatous vasculitis affecting small to medium vessels. The typical manifestation is a triad syndrome, including lesions of ear, nose and throat (ENT), pulmonary vasculitis and glomerulonephritis. Central nervous system (CNS) involvement occurs in 7–11% of patients [1]. Patients usually present with hypertrophic cranial pachymeningitis (HCP), stroke, encephalitis, and epilepsy.

Pituitary involvement is a rare manifestation. It was first described by Ahlstrom in 1953 [2]. From 1983 to 2017, 26 case reports [3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28] and 2 retrospective cohort studies [29, 30] were published on PubMed. Central diabetes insipidus (CDI) and hypogonadotropic hypogonadism were the prominent endocrine features in patients with pituitary involvement [29,30,31]. A correct diagnosis is difficult to determine without positive-ANCA or a typical triad syndrome in GPA patients with pituitary involvement. Furthermore, many cases are inclined to be misdiagnosed as lymphocytic hypophysitis, pituitary abscess or tuberculosis. In this retrospective clinical study, clinical features of 12 cases of patients with GPA and pituitary involvement were reviewed. Additionally, a literature review was conducted to compare previous findings with data from our study and to ascertain clinical features of pituitary involvement in GPA patients.

Methods

Patient cohort and inclusion criteria

Patients with GPA and pituitary involvement, admitted to Peking Union Medical College Hospital (PUMCH) between January 2000 and December 2017, were evaluated in this retrospective study. The query terms used to search pituitary involvement in GPA were “pituitary involvement”, “hypopituitarism”, “panhypopituitarism”, “pituitary neoplasm”, and “diabetes insipidus”. The included patients met all of the following criteria: (1) patients were diagnosed with GPA according to the 2017 ACR/EULAR classification criteria [32]; (2) pituitary involvement, including CDI, and/or hypopituitarism, and/or abnormalities in pituitary magnetic resonance images (MRIs); (3) other causes of pituitary dysfunction could be excluded, such as infection and tumor. Two authors (Gu Y and Sun XF) independently investigated the medical records of patients who appeared to meet the inclusion criteria and they filled out data collection forms. Any discrepancy between the two authors was resolved by discussion with a third author (Mao JF).

Laboratory examination and hormonal assessment

The erythrocyte sedimentation rate (ESR) and C-reactive protein (CRP) were measured on admission. Antineutrophil cytoplasmic antibodies (ANCA) were measured by indirect immunofluorescence as well as proteinase 3 (PR3) and myeloperoxidase (MPO) antigen-specific immunometric assays. Lumbar puncture was conducted and cerebrospinal fluid was collected for routine and biochemistry tests.

CDI was diagnosed by polyuria (24 h urine volume > 3 l) with a good response to desmopressin therapy. Hypogonadotropic hypogonadism was defined as low level of serum testosterone or estradiol accompanied with inappropriately low or normal luteinizing hormone (LH) and follicle-stimulating hormone (FSH) levels. Thyroid-stimulating hormone (TSH) deficiency was defined as a low serum-free thyroxin level (< 1.04 ng/dl) associated with an inappropriately low or normal serum TSH level. Adrenocorticotropic hormone (ACTH) deficiency was defined as a morning cortisol value of less than 5 µg/dl (8am) associated with an inappropriately low or normal ACTH level. Growth hormone (GH) deficiency was defined as insulin-like growth factor (IGF-1) level lower than age-matched normal range. No one underwent stimulation tests.

Radiological assessment

Magnetic resonance images were used to evaluate the pituitary gland and sella turcica. The abnormalities of MRIs were defined by the reports from radiologists. The dimensions of the pituitary gland, gadolinium enhancement, posterior hyperintense signal of pituitary on T1-weighed images (T1WI) with or without HCP, and involvement of adjacent tissue were summarized.

Activity of GPA

The Birmingham vasculitis activity score (BVAS) was calculated at the onset of pituitary dysfunction and at the conclusion of follow-up. Two authors (Gu Y and Sun XF) independently evaluated BVAS of patients according to BVAS version 3.0 [33]. Disagreements were resolved by consensus.

Treatment and prognosis

Immunosuppressive therapy and hormone replacement therapy were administrated. The follow-up times (from discharge to the last out-patient visit) and activity of GPA (BVAS) at the conclusion of follow-up were recorded. Pituitary hormones and pituitary MRIs of some patients were repeatedly measured during follow-up.

Statistical analysis

The descriptive results were presented as percentages and ranges if necessary.

Search strategy for literature review

We used the terms “granulomatosis with polyangiitis” or “Wegener’s granulomatosis” and “diabetes insipidus” or “hypopituitarism” or “pituitary abnormality” to identify the relevant articles between 1983 and 2017 on PubMed. The language of evaluated articles was limited to English. Inclusion criteria were defined as follows: (1) diagnosis of GPA was based upon the 2017 ACR/EULAR classification criteria; (2) pituitaries with abnormalities in function and/or morphology; (3) cases with well-documented clinical information. Patients with secondary vasculitis or other autoimmune diseases, such as drug-induced vasculitis or autoimmune thyroiditis, were excluded. The applicable cases were selected by two investigators (Zhang T and Peng M) independently. Discrepancies were resolved by discussing with an additional author (Shi JH).

Results

Patient characteristics

Among 304 patients diagnosed with GPA between January 2000 and December 2017 in the PUMCH administration database, 12 (3.9%) inpatients were identified as having pituitary dysfunction related to GPA. Half of them were female. The average age at the onset of GPA was 48 years (range 19–64 years).

Clinical manifestations

Two patients presented with polydipsia and polyuria as their first symptoms. Other patients presented with initial symptoms of headache (n = 4), hearing loss (n = 1), blurred vision (n = 1), conjunctivitis (n = 1), ophthalmodynia (n = 1), fever and expectoration (n = 1) and dry cough (n = 1). The lag time between occurrence of initial symptom presentation and pituitary dysfunction ranged from 0 to 103 months. The BVAS, calculated when pituitary symptoms emerged, ranged between 0 and 27. During the course of GPA, ENT manifestations appeared in all patients. Other organ involvement included ocular impairment in ten patients (83.3%), CNS (other structures of CNS except pituitary) involvement in seven patients (58.3%, including 4 patients with HCP), lower respiratory tract involvement in five patients (41.7%), albuminuria in two patients (16.7%), and skin ulceration in two patients (16.7%). Demographics and clinical characteristics are summarized in Table 1.

Table 1 Clinical characteristics in GPA patients with pituitary involvement
Full size table

Laboratory findings

Inflammatory markers, including erythrocyte sedimentation rate and C-reactive protein, increased in 11 patients (91.7%). Seven patients were ANCA-positive (Table 1). All patients with lung or kidney involvement were ANCA-positive (patients 1, 2, 3, 5, 7, and 12). All patients without lung and kidney involvement were ANCA-negative, except Patient 8 who had a low titer of p-ANCA 1:10. Six patients presented with increased cerebrospinal fluid (CSF) pressure (> 180 mmH2O). In general, CSF biochemical markers were normal aside from slightly increased protein (n = 6) and glucose levels (n = 2).

Hormonal evaluation

Central diabetes insipidus presented in 11 patients (91.7%) and all of them responded well to desmopressin. Hypogonadotropic hypogonadism was confirmed in 6/11 patients (54.5%). TSH deficiency was noted in 5/11 patients (45.5%). ACTH deficiency was identified in 4/7 patients (57.1%) before glucocorticoid treatment started. GH deficiency was found in 3/9 patients (33.3%). Prolactin was slightly elevated in 1/11 patients. The hormonal evaluation data are outlined in Supplemental Table 1.

Radiological evaluation

Pituitary MRIs were conducted in all patients. Six patients had enlarged pituitary glands with heights exceeding 8 mm. Six pituitaries revealed homogenous enhancement and two showed peripheral enhancement and central cystic hypo-intensive images. One patient presented with heterogeneous enhancement in the pituitary gland. The posterior hyperintense signal was absent in 11/12 patients. Only Patient 6 had a normal posterior hyperintense signal and normal urine volume during the follow-up. HCP was found in four patients (Fig. 1). The pituitary lesions could invade cavernous sinus (n = 5) and compress the optic chiasma (n = 1). The main features of pituitary MRIs are summarized in Table 2.

Fig. 1
figure 1

HCP and pituitary enlargement in Patient 4 and Patient 6. Patient 4 (a–c): sagittal T1WI scan (a) shows an enlarged pituitary gland (1.99 × 0.93 × 1.13 cm) with heterogeneous enhancement after gadolinium administration (b). The thickened meninges (white arrow), invading bilateral cavernous sinus, indicate HCP (c). Patient 6 (d–f): sagittal T1WI scan (d), enhancement sagittal (e) and coronal T1WI (f) show thickened meninges occupying the sella turcica region completely (white arrow)

Full size image
Table 2 Pituitary MRIs in GPA patients with pituitary involvement
Full size table

Pathology of pituitary

Seven patients underwent biopsy, including lung (n = 2, Patient 1, 2), nasal sinus (n = 2, Patient 8, 11), mastoid process (n = 1, Patient 10), pituitary (n = 1, Patient 6), skin (n = 1, Patient 1), and bronchial mucosa (n = 1, Patient 12). Pituitary biopsy in Patient 6 revealed polyclonal B and T lymphocytes infiltration (Fig. 2), supporting the diagnosis of lymphocytic hypophysitis.

Fig. 2
figure 2

Pituitary pathology of Patient 6. Histology shows diffused lymphocyte infiltration under × 10 (a) and × 20 (b) power using the haematoxylin and eosin (H&E) stain. Photomicrographs illustrating lymphocytes are CD3- (c) or CD20- (d) positive, indicating they are T and B lymphocytes, respectively

Full size image

Treatment and follow-up

All patients were treated with corticosteroids (Table 3). Four patients received pulse therapy (methylprednisolone 500–1000 mg/day, intravenously administrated for 3 days, followed by high-dose corticosteroids). The remaining eight patients received high-dose corticosteroids (equivalent to prednisone 0.8–1 mg/kg/day). Cyclophosphamide and other immunosuppressive agents were administrated for all patients, such as ciclosporin (n = 2), tacrolimus (n = 2), methotrexate (n = 1), mycophenolate mofetil (n = 1) and leflunomide (n = 1). Patient 1 received rituximab therapy due to his poor response to the conventional therapy. The follow-up periods ranged from 1 to 133 months. Nine patients remained in remission (BVAS = 0–2) at the end of follow-up under low-dose corticosteroids with or without immunosuppressive agents. Patient 8 died 1 month following discharge due to severe intracranial hypertension. Patient 12 was diagnosed with myelodysplastic syndrome related to long-term use of cyclophosphamide after 133 months. Patient 10 was unable to be contacted for follow-up.

Table 3 Treatment and follow-up in GPA patients with pituitary involvement
Full size table

Desmopressin was given to all patients with CDI. Patient 2 received levothyroxine for replacement therapy. No growth or sex hormone replacement was given. During the follow-up, all patients with CDI were still on desmopressin treatment. Pituitary hormones, retested in two patients, did not recover to normal range after immunosuppressive treatment. Pituitary MRIs were reevaluated in four patients, and the size of lesions decreased in three patients but increased in one patient.

Literature review

Between 1983 and 2017, 51 cases (2 case series and 26 case reports) of GPA with pituitary involvement have been published on PubMed (Supplemental Tables 2 and 3). The average age at the diagnosis of pituitary dysfunction was 42 years (range 16–77 years) with female predilection (74.5%). The pituitary dysfunction preceded additional organ involvement in 10 patients (15.7%) [4, 5, 7, 15, 18,19,20, 23, 30], was concomitant in 16 patients (31.3%) [5, 8, 12,13,14, 16, 18, 24, 26, 28,29,30], and appeared subsequently in 25 patients (49.0%) [3, 4, 6, 9,10,11, 17, 21, 22, 25,26,27, 29, 30].

The ANCA-positive rate was 91.3% in the literature review. However, ANCA assessment was performed at different times over the course of GPA. ENT (84.3%) were the most common involved organs, followed by lung (51.0%), kidney (39.2%), oculi (31.5%), peripheral nerve system (19.6%), skin (15.7%), and gastrointestinal tract (3.9%). CDI (86%, 43/50) was the most frequent endocrine abnormality and one patient was undergoing dialysis at the beginning of the study. Other pituitary dysfunction included, listed in decreasing frequency, hypogonadotropic hypogonadism (68.3%, 28/41), TSH deficiency (51.2%, 22/43), ACTH deficiency (48.1%, 13/27), hyperprolactinemia (42.5%, 17/40), and GH deficiency (28.1%, 9/32). Due to cyclophosphamide administration, sex hormone was not measured in two patients. ACTH of 16 patients could not be evaluated due to the administration of glucocorticoid in these patients. Pituitary MRIs of 48 patients were available. The common lesions included enlarged pituitary glands (91.1%, 41/45), pituitary glands enhanced after gadolinium injection on T1WI (93.8%, 30/32), and absence of posterior hyperintense signal of pituitary on T1WI (84.6%, 22/26). The clinical manifestations, hormone evaluation and pituitary MRIs are shown in Supplemental Table 2.

The majority of patients received glucocorticoid therapy (94.1%). Cyclophosphamide (n = 32) was the most widely used immunosuppressive agent, followed by methotrexate (n = 6), mycophenolate mofetil (n = 3), and acetazolamide (n = 3). In addition, other therapeutic strategies included rituximab (n = 7), infliximab (n = 5), intravenous immunoglobulin (n = 2), plasma exchange (n = 1), and 15-deoxyspergualin (n = 1). Following treatment, the systemic manifestations of GPA were alleviated in 27 patients and pituitary size decreased in 26 patients. However, endocrine abnormalities were improved in only a minor subset of patients, including decrease of urine volume (n = 9), improvement of hypogonadotropic hypogonadism (n = 5), increase in ACTH (n = 1, most patients were administrated with glucocorticoid), increase in TSH (n = 3), and decrease in prolactin (n = 5). The treatment and follow-up results are summarized in Supplemental Table 3.

Discussion

This study found that pituitary involvement occurred in 3.9% (12/304) patients with GPA. To our knowledge, this is the largest sample size from a tertiary academic hospital. Among these patients, the positive rate of ANCA was remarkably higher in patients with lung and/or kidney involvement compared to patients without. CDI and hypogonadotropic hypogonadism were the dominant endocrine abnormalities. Enlarged pituitary glands and the absence of posterior hyperintense signal on T1WI were common features on pituitary MRIs. After immunosuppressive treatment, the pituitary could decrease in dimensions, while CDI and hypopituitarism could not recover in most cases.

The incidence rate of pituitary involvement in GPA seemed to be underestimated due to the symptoms of hypopituitarism being closely associated with GPA systemic manifestations. The incidence rate of 3.9% was slightly higher than 1.1–1.3% previously reported in two case series reports [29, 30]. Pituitary involvement affected males and females equally in our center, but the majority of patients assessed in the literature review were female. The average age at the onset of GPA in our cohort was 48 years, which was similar to the studies from Mayo Clinic [30], French Vasculitis Study Group [29] and the literature review (48, 46, and 42 years, respectively).

The pathogenesis of pituitary involvement was speculated to be related to granuloma invasion from the adjacent organs (ENT, oculi, meninges), vasculitis of the pituitary gland itself, or granulomatous formation in situ [34]. Involvement rates of ENT, oculi and meninges (presented as HCP) were 100%, 83.3%, and 25%, respectively, in our participants, which were higher than the general GPA population (ENT 70–100% [35], oculi 50–58% [36], meninges/HCP 0.6–13.8% [37, 38]). In the literature review, we found that ENT impairment was common as well. Previous studies showed that ENT lesions were positively associated with CNS involvement [34, 37]. The above evidence supports granuloma extension from ENT, orbits and meninges as the main mechanism of pituitary involvement. Surprisingly, the only pituitary biopsy obtained from Patient 6 merely revealed lymphocyte infiltration. GPA was diagnosed according to clinical manifestations and positive-ANCA detection occurred subsequently, and lymphocytic hypophysitis was considered as a secondary response to GPA. This result expanded our knowledge on the pathology of pituitary involvement in GPA.

Pituitary involvement can occur at any time, before or after other organs are involved. Our literature review revealed that most of pituitary dysfunction occurs concurrently or following additional organ involvement (prior, concurrent, and post organ involvement percentages are 15.7%, 31.3%, and 49.0%, respectively). Nevertheless, pituitary impairment can be the only initial abnormality as well, which deserves more attention in the field. Polydipsia and polyuria were the first symptoms in two patients in our study. Patient 3 presented with polyuria and polydipsia, while her ANCA assessment was initially negative. Two months later, the patient visited the hospital for expectoration and exertional dyspnea. Computed tomography of chest revealed pulmonary opacities distributed along bronchovascular bundles (Fig. 3). Her ANCA reassessment resulted in a positive reading (p-ANCA 1:80 and MPO-ANCA 78 RU/ml), leading to the diagnosis of GPA. After treatment with glucocorticoids and cyclophosphamide, lesions in pituitary and lung resolved simultaneously, accompanied with negative-converted ANCA. For this reason, the diagnosis of GPA cannot be excluded in patients that only present with initial pituitary dysfunction. Conversely, the onset of pituitary dysfunction can lag behind the impairment of other organs for up to 103 months (Patient 7). A previous case report showed that pituitary abnormality could occur even when the disease was stable [39]. According to these results, pituitary involvement does not synchronize with additional organ involvement in GPA.

Fig. 3
figure 3

Changes of pituitary and lung images in Patient 3. Polyuria, polydipsia accompanied with a mass in neurohypophysis (a) lead to a diagnosis of CDI. Patient 3 had a normal lung (b) and initial negative-ANCA. Two months later, pituitary was still enlarged (c). And expectoration, lung opacity (d) and positive serum ANCA (p-ANCA 1:80 and MPO-ANCA 78 RU/ml) resulted in diagnosis of GPA. After treatment, pituitary mass, lung opacity and positive ANCA disappeared (e, f). GPA granulomatosis with polyangiitis, CDI central diabetes insipidus

Full size image

Positive ANCA seemed to be more closely related to lung or kidney rather than pituitary involvement. In our study, the diagnosis of six patients, whose lesions were limited to ENT, oculi and CNS, was difficult with negative-ANCA. But according to 2017 ACR/EULAR classification criteria, three patients (Patients 4, 6, and 9) were diagnosed with GPA due to severe nasal symptoms, hearing loss and oculi damages. Increased inflammation makers, HCP, response to immunosuppressive therapy and subsequent acquisition of positive ANCA supported the diagnosis as well. Another three patients were diagnosed with GPA because granulomatous lesions were discovered by biopsy at the nasal sinus (Patients 8, 11) and mastoid process (Patient 10). Our cohort study showed that all patients with lung and/or kidney involvement (6/12) were ANCA-positive, while patients with lesions limited to head and upper respiratory tract were almost ANCA-negative. GPA is traditionally classified into localized and systemic forms, according to the involved organs [40]. The former is localized to the upper respiratory tract with an ANCA-positive rate 50–80%. The latter involves kidney and/or lung with ANCA-positive rates above 90%. These findings provide further evidence that negative-ANCA does not exclude the diagnosis of GPA in patients with pituitary mass who do not have lung or kidney impairment.

Consistent with the literature review, CDI and hypogonadotropic hypogonadism were the most common endocrine presentations. Patients presented with CDI in almost all patients in our study. Life-threatening conditions such as dehydration and electrolyte disturbance can be caused by CDI. The pituitary–gonadal axis was mostly involved in patients with hypopituitarism. However, for patients with chronic disease, hypogonadotropic hypogonadism is difficult to identify based on symptoms, such as amenorrhea or low libido, since administration of cyclophosphamide directly impairs gonadal function as well. The pituitary–adrenal axis is also difficult to evaluate because levels of ACTH and cortisol are inevitably influenced by dosing of corticosteroids that surpass physiological levels. 4-Hydroxycyclophosphamide is a bioactive metabolite of cyclophosphamide. Studies have confirmed that severe hypothyroidism can reduce the concentration of 4-hydroxycyclophosphamide, potentially resulting in treatment failure [41]. Therefore, thyroid hormone replacement therapy is necessary. In our study, levothyroxine was prescribed to one patient who had a very low level of free thyroxine (FT4 = 0.547 ng/dl).

The typical MRIs of pituitary involvement in GPA showed enlarged pituitary gland, absence of posterior hyperintense signal on T1WI, HCP and lesions invading cavernous sinus. Most pituitary diseases present as pituitary enlargement, including adenoma and lymphocytic hypophysitis, and the diagnosis is determined by pituitary biopsy. HCP is considered as granulomatous inflammation of meninges in GPA [42]. It usually manifests with symptoms of headache and cranial nerves paralysis. HCP frequently invades the falx cerebri and tentorium cerebelli but is seldom around cavernous sinus [43]. When cavernous sinus is involved on imaging, pituitary impairment should be suspected (Fig. 1). Furthermore, sinusitis, mastoiditis and orbit mass in MRIs likely indicate a diagnosis of GPA. Enlarged pituitary glands can shrink after immunosuppressive treatment, which was confirmed in our study and literature review. However, for some patients with progressive disease, pituitary resection surgeries were needed to relieve the oppression of granulomatosis [7, 26, 30].

Most patients remained in remission from GPA after immunosuppressive therapy, while hypopituitarism and CDI were not alleviated. The BVAS and its modifications were the most widely used tools for disease activity assessment in ANCA-associated vasculitis. The BVAS of seven patients decreased dramatically after treatment, indicating relief of GPA. Patient 1 experienced remission after rituximab infusion, while hypogonadotropic hypogonadism and TSH deficiency still remained after treatment. Irreversible hypopituitarism may be caused by permanent damage and fibrous tissue replacement in the pituitary. The damage to neurohypophysis is also irreversible, and desmopressin is needed for the foreseeable future. In the literature review, the improvement of CDI and hypopituitarism was rare as well. On the contrary, a case series revealed that 83% (5/6) of CDI patients recovered from the disease [30]. They concluded that early diagnosis and treatment potentially made a contribution to the resolution of CDI. Therefore, hormone replacement therapy for the duration of the patients’ lives would be necessary for most patients.

Several limitations of this study should be addressed. First, in this retrospective study, pituitary images were only reevaluated in four patients and two of them had repeated pituitary function during the follow-up. Second, only one patient received pituitary biopsy and the result showed polyclonal lymphocyte infiltration.

In conclusion, pituitary involvement in GPA can occur before or after other systemic impairment and even as the initial presentation. Almost all patients without lung or kidney impairment were ANCA-negative, while all patients with lung or/kidney impairment were ANCA-positive. These findings indicate that negative-ANCA cannot eliminate the diagnosis of GPA especially for patients without lung or kidney involvement. CDI and hypogonadotropic hypogonadism were the most common endocrine manifestations. Immunosuppressive treatment could notably improve systemic disease and pituitary imaging, while the reversal of anterior and posterior pituitary function was seldom obtained.