Introduction

Pituitary gigantism (PG) is a very rare disease caused by chronic growth hormone (GH) and insulin-like growth factor 1 (IGF-1) hypersecretion occurring before complete fusion of the epiphyseal growth plates. Commonly, GH overproduction in PG derives from a pituitary somatotropinomas [1, 2]. The large majority of PG occurs as a sporadic disease while a small number occurs in the context of genetic syndromic disorders, such as McCune Albright syndrome (MAS), Carney complex (CNC), multiple endocrine neoplasia types 1 and 4 (MEN 1 and MEN 4) and the paraganglioma, pheochromocytoma and pituitary adenoma association (3PA) [1, 3, 4].

The genetic background of non-syndromic PG includes inactivating germline mutations in the aryl hydrocarbon receptor-interacting protein (AIP) gene. These mutations were found in about 29% of gigantism cases either sporadically or in the setting of Familial Isolated Pituitary Adenoma (FIPA) [5, 6]. AIP mutations are more frequent in young, predominantly males, patients and have been associated with large and invasive tumors and pituitary apoplexy [5, 7,8,9]. Generally, these patients more often had GH excess, were more often resistant to treatment with somatostatin receptor ligands (SRL) and underwent more surgical interventions, requiring multimodal therapy [5, 7,8,9].

Recently, an additional cause of gigantism has been linked to microduplications of G protein-coupled receptor 101 (GPR101) in Xq26.3 and termed X-linked acrogigantism (X-LAG) [10].GPR101 microduplications were found in both sporadic and familial cases of patients with somatotropinomas [10]. Patients with X-LAG have a distinct phenotype characterized by extraordinarily early gigantism with a median age of onset of 12 months [10,11,12].

Here, our main objectives were to evaluate (a) the relationship between clinical characteristics, genetic abnormalities, and long-term outcomes of patients with non-syndromic PG followed at a single pituitary center in Sao Paulo, Brazil, and (b) the different therapeutic strategies used to induce disease remission. These results add new case reports of this very rare disease increasing knowledge of pituitary gigantism.

Materials and methods

Subjects

This study included 18 patients with non-syndromic pituitary gigantism followed at a quaternary referral Neuroendocrine Unit, Division of Endocrinology and Metabolism, Clinical Hospital—FMUSP, between 1990 to 2016. All patients were from Sao Paulo state. PG encompassed patients with pituitary lesions leading to GH/IGF-1 hypersecretion that presented: (a) a final height standard deviation score (Z-score) above + 2 or (b) elevated rates of height growth or (c) height above + 2 SD from the mid-parental height [2, 13]. This study was approved by the local Ethics Committee and an informed consent form was obtained from all patients or their legal guardian.

Study design

This study included a retrospective review and a cross-sectional evaluation of PG, presenting our experience in the diagnosis, management, and follow-up of these patients. The data were obtained from medical records at the time of first symptoms, at diagnosis, and at last follow-up. Anthropometric data were established at medical appointment, when height was measured using a stadiometer. Whenever possible, patient’s parents underwent clinical evaluation for confirmation of their height.

Clinical and hormonal data

Height was measured using the stadiometer and expressed in centimeters and as sex and age specific Z-scores [14].Target height was based on sex adjusted mid-parental heights, − 6.5 cm for girls and + 6.5 cm for boys [15]. Body mass index (BMI) was calculated as kilograms per square meter (kg/m2). Pubertal development was determined according to the classification proposed by Marshall and Tanner [16, 17]. Bone age was estimated based on non-dominant-hand/wrist radiographs using the atlas of Greulich and Pyle [18]. Familial tall stature and signs of obesity, precocious or delayed puberty signs and dysmorphisms, and stigmata of specific disorders known to be associated with tall stature were also evaluated.

Serum GH concentration was measured with immunoradiometric assay (IRMA) or by immunofluorometric assay (IFMA) (AutoDELFIA, Wallac, Turku, Finland) with monoclonal antibodies. IGF-1 was measured by RIA after ethanol extraction (Diagnostic Systems Laboratories, Webster, TX) or by chemiluminescence assays (CLIA) (IMMULITE; Diagnostic Products Corp., Los Angeles,CA). IGF-1 level was standardized for age and sex, according to reference values provided by the manufacturer’s.

GH hypersecretion was diagnosed by lack of suppression of GH levels during oral glucose tolerance test (OGTT) and IGF-1 level above the age-adjusted normal range. In patients undergoing surgical approach, the status of the disease was also evaluated by OGTT and IGF-I levels performed four months after surgery. In patients on medical therapy, the status of the disease was defined by random GH and IGF-I levels in the age-adjusted normal range. The nadir GH cutoff used to distinguish active disease from control/remission was method-specific, depending on the assay available at the time of diagnostic. IGF-1 levels were expressed as a multiple above the upper upper limit of normal reference range for age (x ULNR IGF-I; normal = x ULNR IGF-I < 1). Based on ULNR-IGF-1 the disease was defined as controlled (IGF-1 ≤ ULNR) or active (IGF-1 > ULNR) [19].

MRI and histopathological studies

The maximum tumor diameter was measured preoperatively by computed tomography (CT) or magnetic resonance imaging (MRI) T1 weighted coronal view evaluated by a single neuroradiologist. Concerning histopathological analysis, tumor specimens were evaluated by routine eosin-hematoxylin stain and immunohistochemical staining for GH, PRL, ACTH, LH, and FSH hormones. Somatotropinomas were confirmed based on positive staining for GH. Tumor T2-weighted signal intensity was assessed by visual inspection on coronal plane and adenomas are classified as hypo-, iso- or hyper-intense in relation to the healthy pituitary gland or the temporal grey matter as reference tissue.

Genetic screening

Blood samples for DNA extraction (supplementary methods) was available for 16 patients. For these, the entire coding region of AIP (ENST00000279146), GPR101 (ENST00000298110), MEN1 (ENST00000312049), CDKN1B (ENST00000228872.9) and GNAS hotspots (exons 8 and 9; ENST00000371100) were amplified by polymerase chain reaction (PCR) and Sanger sequenced (supplementary methods). All variants identified were confirmed in two independent PCR products and by sequencing both DNA strands. The new genetics variants were categorized in different classes of pathogenicity according to the American College of Medical Genetics and Genomics (ACMG) guidelines [20].

Droplet digital PCR (ddPCR) was used to assess copy number variants (CNVs) at the GPR101 and AIP genes using blood-derived DNA [21]. GPR101 CNV analysis was also performed using DNA extracted from tumor or buccal cells, when available. Detection of a GPR101 microduplication by ddPCR was also confirmed by array comparative genomic hybridization (aCGH), as previously described [10]. The term “non mutated patients” was used throughout this manuscript to define the patients without pathogenic variants in known genes.

Statistical analysis

All statistical analyses were performed using the Stata/SE 14.2 software (StataCorp LLC, Texas, USA). All data were expressed as median and lower–upper quartile and compared with two-samples Wilcoxon rank-sum (Mann–Whitney) non-parametric test. Categorical variables are presented as absolute values or percentages and were tested using the Fisher exact test. P values < 0.05 were considered statistically significant.

Results

Clinical and biochemical characteristics of patients

Table 1 presents the main clinical and biochemical characteristics of each patient. Fifteen patients were males (83%) and three were females (17%). Three cases (17%) had a familial history of pituitary adenomas (FIPA) but not of tall stature. The median age at time of diagnosis of PG was 17 (15–20) years with first signs and symptoms noticed at 14 (9–16) years. Therefore, the delay in the diagnosis of gigantism was 3.5 (2–8) years. The median of height Z-score was 3.6 (2.9–5), with adult height median of 198 (195–203) cm. The most common complaint was accelerated growth and tall stature in twelve patients (67%), followed by loss of libido in three (17%), diabetes mellitus in two (11%), and paresthesia in one case (5%).

Table 1 Clinical, hormonal and tumor features, outcomes, and germline genetic abnormalities identified in PG patients
Full size table

Headache was present in fourteen (78%) and visual disturbance in thirteen (72%) patients, hyperhidrosis in fourteen (78%) patients, enlarged hands and feet in eleven patients (61%), arterial hypertension in nine patients (50%), arthralgias in eight (44%) patients, paresthesia in four (22%) patients, fatigue in nine (50%) patients. None of the patients had galactorrhea and in three cases (17%) diabetes mellitus was diagnosed. Eight patients (44%, age range 15–28) showed evidence of epiphyseal closure on hand radiography at the time of diagnosis. Two patients (11%) had a bone age delay higher than two years.

GH basal and x ULNR-IGF1 medians at diagnosis were 70 (35–108) and 1.9 (1.7–2.9), respectively. Biochemical evidences for gonadal, thyroid, and adrenal deficiencies were found in sixteen (89%), fifteen (83%) and ten (55.5%) cases, respectively.

Tumor features

All pituitary tumors were macroadenomas, three (17%) with a maximum diameter ≥ 4 cm (giant adenomas). Suprasellar extension was noted in fifteen adenomas (83%). Of these, 5/15 adenomas (33%) presented intrasellar, 3/15 (20%) parasellar, and 7/15 (47%) both intrasellar and parasellar extension. Interestingly, pituitary hyperplasia alongside adenoma was observed in one (5%) patient. All tumors were immunoreactive for GH, with six adenomas (33%) also expressing PRL (Table 1). At diagnosis, three, six and four tumors were T2-hypo-, iso- and hyper-intense, respectively (Table 1).

Treatments and outcomes

Seventeen patients underwent pituitary surgery (sixteen transsphenoidal and one transcranial) as primary therapy, which was completely effective in only two cases. A second surgical approach was unsuccessful in seven patients. Postoperative radiotherapy (RT) was performed in two cases, resulting in disease control. Postoperative medical therapy was administered in the remaining thirteen patients and included SRL alone or in combination with the dopamine agonist (DA) cabergoline. Ten out of these 13 patients remained uncontrolled. Of these, five patients have received RT as a third treatment modality, which was successful in four cases.

One patient received SRL as primary therapy, but required additional transsphenoidal surgery for disease control. This patient had a hyper-intense signal on T2 (#5, Table 1). Other eight patients from iso- and hyper-intense group, received SRL as adjuvant therapy. Of these, three (#1, #11 and #14, Table 1) were responsive and five (#4, #7, #13, #15 and #17, Table 1) unresponsive to medical treatment. Only one patient with hypo-intense tumors signal (#18, Table 1) used SRL as secondary therapy and showed no response. The other two patients with hypo-intense tumors (#6 and #8, Table 1) were successfully treated with primary TSS.

Overall, the rate of success of treatment was 11% (2/18 patients) with monomodal therapy, 37.5% (6/16) with bimodal therapy (surgery + SRL/DA or surgery + RT), and 80% (4/5) with trimodal therapy (surgery + SRL/DA + RT). Patients were followed for a median of 12.5 years (7–21) and the median time to hormonal remission was 15.5 years (7–20).

Genetic abnormalities

Genetic analysis of AIP in DNA extracted from blood samples identified four heterozygous variants in PG patients (Supplementary Fig. 1). A novel c.788-2A > C variant (absent in gnomAD and 1000 Genome databases) was detected in one male patient. This variant is located in the canonical acceptor splice site of intron 5 and was classified as likely pathogenic according to ACMG criteria, weighted as very strong (PVS1). The three other AIP variants identified were previously described in patients with pituitary somatotropinomas as deleterious: p.Gln217Ter (c.649C > T, rs267606566, exon 5), p.Ala277Pro (c.829G > C, rs267606581, exon 6) and p.Arg304Ter (c.910C > T, rs104894195, exon 6). No AIP deletions were detected using ddPCR in our cohort of PG.

GPR101 single nucleotide variants (SNVs) were not observed in any case. Two sporadic female patients presented a germline microduplication in Xq26.3 as detected by ddPCR and confirmed by aCGH. In one of these patients, this microduplication was also observed in the tumor DNA. The CNVs identified in patients #16 and #17 also encompassed CD40LG, ARHGEF6, and RBMX, frequently duplicated with GPR101 in X-LAG (Fig. 1). Mutations in MEN1 and GNAS were excluded in all patients. The clinical and tumor characteristic of AIP mutated and X-LAG patients are described in Table 1. Figure 2 shows MRI pituitary images of one X-LAG patient.

Fig. 1
figure 1

Results of droplet digital polymerase chain reaction (ddPCR) and array comparative genomic hybridization (aCGH) analyses for the Xq26.3 region in the subset of AIP-mutation negative patients. a Copy number ddPCR assays for GPR101. Two copy number variation assays, Hs01818174_cn and Hs01730605_cn (Thermo Fisher Scientific), located respectively at the 5′ and 3′ end of the gene were employed. Reactions were performed both with and without the restriction enzyme HindIII, which was confirmed to not cut within the target and reference (RNASEP) amplicons. HindIII was used to reduce sample viscosity and off-target amplification, and by separating tandem gene copies, ensured proper random partitioning into droplets. Patients #16 and #17 showed a potential duplication of GPR101. b For patients #16 and #17, the ddPCR analysis was extended to include the centromerically adjacent CD40LG, ARHGEF6 and RBMX genes, commonly duplicated alongside GPR101 in X-LAG patients. The centromeric RBMX2 and the telomeric ZIC3, two genes never included in the X-LAG duplications reported so far, served as controls. All reactions were performed using HindIII. As expected, RBMX2 and ZIC3 showed no copy number gains. While some variability, likely due to the quality of the DNA, was observed for the other genes, the presence of a Xq26.3 duplication was confirmed. In both panels a and b, bars correspond to 95% confidence intervals; the dotted red lines crossing the y axis at 1.5 and 2.5 copy numbers represent the threshold for deletion and duplication in females, respectively. S Oral swab-derived DNA, T tumor-derived DNA. c Genome view of the Xq26.3 duplication detected by aCGH in blood-derived DNA from Patients #16 and #17, according to the UCSC Genome Browser (hg19; GRCh37/hg19). aCGH confirmed the ddPCR results

Full size image
Fig. 2
figure 2

T1-weighted a coronal and b sagittal MRI of patient #17 with X-LAG revealed a heterogeneous macro adenoma with suprasellar lesion extension

Full size image

Comparison of genetics, patients’ and tumor features and outcomes

Among all patients there were significant differences in age at diagnosis when comparing X-LAG to non-mutated (10.0: 5–15 vs. 18.5:16–24 years, respectively, P = 0.04, Table2) but not with AIP-mutated patients 16.5:13–18.5 years, P = 0.25, Table 2). Significantly younger age at onset of symptoms was observed for X-LAG when compared to non-mutated (7.0: 5–9 vs. 15:14–16 years, respectively; P = 0.03, Table 2) but not with AIP-mutated patients (12.0: 8–15.5 years; P = 0.24, Table 2). No statistical significance was observed when differences of ages at diagnosis and at symptoms onset were compared between AIP-mutated and non-mutated patients (P = 0.24 and P = 0.23, respectively). Although no statistically significant difference was identified in height at diagnosis between X-LAG, AIP mutated, and non-mutated patients (Table 2), a significant height Z-score difference was identified between X-LAG and non-mutated patients (6.2:5.7–6.8 vs. 3.1:2.9–3.9, respectively; P = 0.03 Table 2).

Table 2 The comparison of clinical, tumoral, and outcome data among AIP-mutated, X-LAG, and non-mutated patients with PG
Full size table

No other significant differences, such as GH and IGF-1 levels, tumor features, multi-modal therapy or disease control, were observed among these three categories of PG patients (Table 2). However, patients receiving trimodal treatment had larger tumors when compared to those receiving monomodal (3.8:3.1–4 vs. 1.65:1.2–2.1; respectively; P = 0.03). Also, tumor size had a weak positive correlation with height Z-Score (rho = 0.6; P = 0.03).

Discussion

Pituitary gigantism is a very rare disease, with few cases being described worldwide [2]. Nevertheless, significant advances in understanding the genetic causes of PG were achieved in the last years. The AIP gene was considered the most common cause of non-syndromic PG, accounting for approximately 29%, and GPR101 has been identified as an essential novel locus for this disease [2, 5, 10].

In the present study, a genetic investigation was undertaken in 16 out of 18 PG patients evaluated. AIP pathogenic variants were found in 4 cases (25%) and included a novel putative splicing variant (c.788-2A > C) in intron 5. This variant was observed in one male patient without familial history of accelerated growth or pituitary tumors, unlike the others AIP-mutated cases of our cohort who had family members with pituitary tumors. For familial cases, DNA samples from paternal uncle (patient #2, p.Arg304Ter) and father (patient #3, p.Gln217Ter), both with acromegaly, were available. The same AIP proband mutations were present in their respective relatives (data not shown). In fact, AIP mutations are frequently observed in familial cases of isolated somatotropinomas (FIPA) and were commonly found in PG patients [6, 8].

Surprisingly, in the most extensive multi-center international PG study involving 208 syndromic and non-syndromic cases, the majority of AIP-mutated patients were apparently sporadic [2]. However, the authors highlighted that these cases could be mistakenly interpreted as simplex, since AIP genetic family screening cannot be extensively performed and incomplete penetrance could lead to a family generation with no affected relatives [2, 5]. AIP mutated patients of our cohort showed invasive macroadenomas and did not achieve disease control with first-generation SRL monotherapy concurring with previous publications. AIP mutations were indeed described to confer resistance to first-generation SRL in patients with somatotropinomas [7, 22], and multi-modal therapy was required for hormonal control and/or residual size reduction of these tumors [23].

GH receptor antagonist pegvisomant (alone or combined with long-acting SRL) has been shown to be an effective modality for the treatment of pituitary gigantism [24,25,26]. However, this drug is not available in our public health system. Second-generation somatostatin multireceptor ligand pasireotide-LAR was administered with octreotide-LAR in three patients of cohort: one harboring the AIP c.788-2A > C intronic mutation (patient #1, Table 1), one with X-LAG (patient #17, Table 1) and one with unknow mutation status (patient #15, Table 1). Only the first one patient achieved disease remission. Recent reports have demonstrated a correlation between T2-tumor signal intensity and response to medical treatment with SRL, before or after surgery, in patients with somatotropinomas [27,28,29,30]. In particular, a better response to SRL therapy was observed in T2-hypotintense compared to hyper/isointense adenomas [27,28,29,30]. However, only one patient with T2-hypointense signal of our cohort received SRL therapy and this relationship could not be evaluated in the current study.

Xq26.3 genomic abnormalities, encompassing GPR101, occur in approximately 10% of PG and define the newly described X-linked acrogigantism (X-LAG). X-LAG is a subtype of PG characterized by early-onset gigantism, mostly affecting females [10]. Usually, these patients present with accelerated growth between 12 – 24 months and in all 36 cases described to date the disease onset occurred before four years of age [10,11,12, 21, 31]. Here, germline Xq26.3 microduplications were found in 2/16 (12,5%) of PG. Both were females with a sporadic disease and reported their symptoms at 5 and 9 years of age (patients #16 and 17, respectively; Table 1). We think this finding can be attributed to specific patients’ factors leading to difficulty recognizing their health situation, such as low intellectual and socioeconomic status and psychological aspects. Pediatric growth charts would be useful to evaluate these patients' growth velocity and, consequently, indicate the onset of illness. Unfortunately, these data were not available for both X-LAG patients.

Recently, somatic mosaicism was shown to occur in sporadic X-LAG male patients. In some of these patients the GPR101 duplication was identified in the pituitary DNA tissue, but not in leukocytes or saliva-derived DNA [32, 33]. Conceivably, postzygotic GPR101 mutations could be present in some of the male patients described in our cohort. However, only two male patients had pituitary tumor tissue available for exclusion of somatic Xq26.3 microduplications. Of note, the clinical characteristics of mosaic males are similar to those of the females/males with X-LAG due to germline Xq26.3 microduplications [21, 31,32,33].

Comparing X-LAG to non-mutated PG patients, there were significant differences in the average age at diagnosis (P = 0.04), age of onset of symptoms (P = 0.03), and height Z-score (P = 0.03). However, such differences were not seen when compared to AIP mutated patients. Classically, both AIP mutated and X-LAG patients showed a more aggressive clinical presentation, challenging its treatment strategy [21,22,23, 34]. According to this data, in the present study’s cohort, no patients harboring genetic abnormalities achieved disease control after transsphenoidal surgery and all of them required adjuvant therapy.

In conclusion, to our knowledge this report represents the largest cohort of PG patients followed at a single center. The prevalence of AIP mutations and GPR101 genomic abnormalities were in line to previous studies. Although the age of onset of symptoms in our X-LAG patients is higher than previously described, this result must be viewed in the context of some limitations, including insufficient availability of birth and early childhood data and a relatively small number of patients with genetic abnormalities evaluated.