Introduction

Pituitary neuroendocrine tumors (PitNETs) are common intracranial neoplastic lesions originating from the adenohypophysis [1, 2]. Although, according to epidemiological data, their prevalence was estimated in 25 cases per 100,000 population [3], the clinical prevalence of PitNETs by case-finding studies is higher and approaches 1 in 1000 inhabitants [4]. In a recent report from the United States Central Brain Tumor Registry, surgically resected PitNETs accounted for about 15% of all brain tumors [5]. PitNETs can also occur in pediatric age, albeit rarely: 3% of all intracranial neoplasms in children are PitNETs, involving females twice as often as males (2:1) [6, 7].

Pituitary tumors were previously classified by immunohistochemistry according to the hormones they secrete; i.e., growth hormone (GH), prolactin (PRL), adrenocorticotropic hormone (ACTH), thyroid stimulating hormone (TSH), luteinizing hormone (LH) and follicle-stimulating hormone (FSH), and hormone immunonegative tumors were classified as “null cell”. The World Health Organization (WHO) in 2017 introduced the diagnostic use of transcription factors detection for PitNET classification according to their lineage of origin: T-PIT-lineage for corticotrophs, SF1-lineage for gonadotrophs, and PIT1-lineage for somatotrophs, lactotrophs, mammosomatotrophs, and thyrotrophs [8,9,10]. The use of transcription factors detection for PitNET classification is recommended in tumors with an undetectable or unclear expression of pituitary hormones and has markedly reduced the number of so-called “null-cell” tumors [1, 8, 9]. Accordingly, the histological diagnosis and classification of PitNETs is currently based on the morphological features and immunohistochemical staining for pituitary hormones and, where appropriate, lineage-specific transcription factors [2, 9,10,11].

From a genetic viewpoint, recent molecular studies showed that sporadic pituitary tumors have low mutational burdens compared with other tumor types [12,13,14,15]. Inherited forms of PitNETs are estimated to account for around 5% of PitNETs, including familial and/or syndromic forms [15], the most frequent due to germline mutations in the MEN1 and AIP genes. Two recurrent somatic mutations have been identified in PitNETs, affecting the guanine nucleotide-binding protein alpha-stimulating (GNAS) gene in GH-secreting pituitary tumors and the ubiquitin-specific peptidase 8 (USP8) gene in ACTH-secreting pituitary tumors [12, 16,17,18]. Both were confirmed by recent pangenomic analysis to be present in > 5% of PitNETs [19, 20]. A small subset of PitNETs harbor somatic copy-number alterations affecting whole chromosome arms [13, 19, 21]. All these data suggest that alterations other than single recurring somatic mutations are involved in most sporadic PitNETs, and that tumor progression could be driven by several genetic/epigenetic factors still poorly understood [20].

In most human cancers, neoplastic cells acquire the ability to escape replicative senescence by activating telomere maintenance mechanisms crucial to their endless replication and immortalization [22,23,24]. Neoplastic cells depend on two mechanisms to maintain telomere length: (1) reactivation of telomerase and (2) a telomerase-independent mechanism mediated by homology-directed repair called alternative lengthening of telomeres (ALT) [25,26,27,28,29,30,31,32,33,34].

Telomerase reactivation is the most common mechanism (involved in around 90% of tumors). The main genetic alterations associated with telomerase activation are TERT promoter mutations and epigenetic changes, TERT gene amplifications, and structural rearrangements of TERT regulatory elements [25, 35,36,37,38]. In the present study, we only considered TERTp methylation, which can be seen quite frequently in PitNETs (16–28%), while we did not examine common TERTp hotspot mutations (C225T and C250T) or TERT amplifications/rearrangements because they have almost never been found in this type of tumor [39,40,41,42].

The other mechanism, ALT, is commonly associated with loss-of-function mutations in the chromatin remodeling genes, the α-thalassemia/mental retardation syndrome X-linked (ATRX) gene, and the death-associated domain protein (DAXX) [29, 33, 43]. ATRX inactivation can be driven by point mutations, insertions or deletions of bases, or large deletions, as well as by other genetic alterations not detected by direct DNA sequence analysis, such as promoter silencing mutations. These alterations are not localized to any specific domain of the protein, and ATRX loss of nuclear expression identified by immunohistochemistry (IHC) is considered a strong surrogate of ATRX loss-of-function mutations [37, 44, 45]. Functional inactivation of DAXX has also been reported, but less commonly and not in PitNETs [33, 46].

Heaphy et al. reported in 2011 [29] that the prevalence of ALT in human cancers is 3.73% and varies considerably among tumor subtypes. Their findings were subsequently confirmed by several other studies [25, 28,29,30, 32,33,34, 46,47,48]. To the best of our knowledge, only two studies analyzed telomere length in pituitary tumors [38, 49]. Boresowicz et al. reported variable telomere length in PitNETs, but no relationship with TERT alterations or tumor characteristics; they concluded that telomerase abnormalities had no role in pituitary tumor pathogenesis [38]. On the other hand, Heaphy et al. (2020) found ALT to be rare in pituitary tumors (2.8%), but more frequent in recurrent cases (36%) [49].

In the present study, we analyzed samples from 24 adult patients affected by recurrent PitNETs, including one case of metastatic PitNET. We examined cases of ALT activation and mean telomere fluorescence intensity, as well as ATRX loss and TERTp methylation. We correlated our findings with patients’ molecular and clinicopathological features. We also investigated both the primary tumors and subsequent relapses in 14/24 patients to analyze telomere length, TERTp methylation, and ALT during progression. We considered a second cohort of 12 pediatric PitNET as well, since neither ALT activation or TERTp methylation has been studied in this population, and telomerase-targeted and ALT-targeted therapies might hold promise for pediatric tumors in general.

Materials and Methods

Case Selection

Informed consent was obtained from all participants with adult recurrent and pediatric primary PitNETs. The histopathological diagnosis was based on the criteria established in the World Health Organization (WHO) Classification of Tumors of Endocrine Organs 2017, and tumor purity > 80% was confirmed in all samples by two neuropathologists (H.A. and F.G.) [5, 8, 9]. Tumors were classified at diagnosis according to immunohistochemical expression of pituitary hormones (PRL, ACTH, GH, FSH/LH, and TSH) as corticotroph, somatotroph, mammosomatotroph, lactotroph, thyrotroph, gonadotroph, and immunonegative. Immunostaining for the pituitary transcription factors SF1, T-PIT, and PIT1 was performed in hormones-immunonegative tumors or in the presence of equivocal results [8]. Lacking any universally accepted cut-off to define the prognostic value of p53 positivity for PitNETs, we elected to define p53 positivity when: more than 10 strongly positive nuclei/10HPF as suggested by Trouillas et al. [2], and/or when more than 5% of tumor cells as suggested by Saeger (Saeger PMID 17287410).

Only recurrent adult PitNETs that required more than one surgical procedure were considered for this study, for a total of 42 specimens from 24 patients. Twelve primary pediatric PitNETs (patients aged < 20 years old) were also studied and classified in the same way as their adult counterpart. All pediatric tumors were analyzed at the onset, and none was re-operated. All clinical and molecular data and results on telomere length are shown in the Supplementary Tables 1 and 2.

Immunohistochemistry for Assessing ATRX Nuclear Expression

Protein expression was evaluated by immunohstochemistry (IHC) with diaminobenzidine (DAB) staining. The streptavidin–biotin–immunoperoxidase technique was used on 3-μm sections of FFPE samples. IHC analyses were run on a Leica Bond III automated immunohistochemical stainer according to the manufacturer’s instructions. Primary anti-ATRX antibodies (NBP1-83077 Novus Biologicals, rabbit polyclonal, 1:1000 dilution) were incubated for 30 min at room temperature. ATRX immunohistochemical expression was quantified by counting the stained nuclei among a minimum of 300 cells in neoplastic regions, as described elsewhere [44, 47, 48]. Cases with ≤ 15% immunopositive tumor nuclei were judged likely to harbor ATRX-inactivating mutations, since loss of nuclear ATRX expression has been consistently associated with mutations scattered in multiple positions in the ATRX coding and non-coding sequences [33, 44, 47, 48]. The evaluation was performed by three independent observers (H.A., S.M, and F.R.B), and scoring was done in consensus. Endothelial cells, infiltrating inflammatory cells, and cortical neurons were generally positive and served as internal positive controls; external positive and negative control tissues were also included in the analysis. Only for samples with ATRX loss of expression, a second re-staining was performed, in order to exclude artifacts or misjudgements.

Methylation-Specific PCR for Assessing TERTp Methylation Status

A semiquantitative methylation-specific polymerase chain reaction (MS-PCR) was used to ascertain the methylation status of five CpG sites 600 bp upstream from the TERT transcription start site (UTSS region). Bisulfite DNA modification was performed using the EZ DNA methylation kit (ZYMO Research) according to the manufacturer’s instructions. The five CpG sites were targeted using specific primers to amplify the bisulfite-modified DNA. Two pairs of forward and reverse primers specific for methylated and unmethylated alleles, respectively, were used for MS-PCR. Details of the PCR cycling conditions and data analysis with ImageJ software (NIH) are available in previous works [47, 48] and in Supplementary Fig. 1.

Telomere-Specific Fluorescence In Situ Hybridization and Telomere Length Analysis

Telomere length was investigated on FFPE sections using telomere-specific fluorescence in situ hybridization (Telo-FISH), with FITC-PNA (peptide nucleic acid) probes (K532511, Dako) complementary to the telomeric repeat sequences, as previously described [33, 47,48,49]. The PNA probes do not recognize sub-telomeric sequences, thus enabling an exact measurement of telomere length.

FISH sections were examined with an AxioImager M1 microscope (Carl Zeiss) by two investigators (S.M. and F.R.B.). Signals were counted for a minimum of 200 tumor nuclei. Cases were identified as ALT-positive when ≥ 5% of tumor cells exhibited large, ultra-bright intranuclear foci of telomere signals, as previously described [29, 33].

Image analysis was performed using TFL-Telo software to quantify the mean telomere intensity [47, 48, 50]. Other cell types (e.g., infiltrating lymphocytes, necrotic areas) were excluded from the digital image analysis.

Telomere fluorescence thresholds were established using three different controls. External controls for ALT derived from previously studied cases of brain tumors [47, 48] were used as follows: ALT-positive tumors to set the threshold for ALT activation (23 samples including 16 pediatric high-grade gliomas with ATRX loss/H3.3-G34mut, and 7 medulloblastomas with ATRX loss; mean = 1546, standard deviation = 393); ALT-negative tumors to establish the threshold for inactive telomerase/ALT mechanisms (23 samples including 4 non-neoplastic brain tissues, 9 pediatric high grade gliomas with ATRX-pos/H3.3-wt/hTERTp-wt/hTERTp-unmethylated, and 10 medulloblastomas with ATRX-pos/H3.3-wt/hTERTp-wt/hTERTp-unmethylated; mean = 691, standard deviation = 172). Endothelial internal controls were used within each pituitary tumor sample, measuring telomeres intensity in endothelial cells of selected vascularized neoplastic areas (mean = 628, standard deviation = 74), for direct comparison with respective neoplastic cells (Fig. 1). This internal measurement was necessary to set the thresholds for the three different telomere intensity categories.

Fig. 1
figure 1

Telo-FISH images showing representative adult pituitary tumors from case n° 21 (L), n° 11 (M), and n° 9 (H) with a low (mean 200–500), medium (mean 500–800), or high (mean 800–1000) telomere intensity in neoplastic cells, as compared with endothelium. The picture on the right shows a representative ALT-positive sample (case n° 24) with ultra-bright signals in the cancer cells and a mean telomere intensity > 1000. The red ring surrounds the nuclei of endothelial cells, while the white ring encloses the nuclei of neoplastic cells. The histograms show the results of TFL-Telo software analysis on neoplastic areas, and the mean telomere intensity for each category

Full size image

TFL-Telo identified distinct telomeric profiles (Fig. 3b), with significant differences in telomere length (p < 0.001). Based on mean telomere intensity, each sample was assigned to one of three categories: low (L), medium (M), or high (H) intensity. PitNETs with a low telomere intensity had mean levels in the range of 200–500, those of medium intensity ranged between 500 and 800 (the same range of telomere intensity as in endothelial cells, see Fig. 3b), and those of high intensity between 800 and 1000. ALT-positive cases with ultra-bright signals had a mean telomere intensity > 1000 (Fig. 1).

Statistical Analysis

All data were collected using Microsoft Excel 2016 and analyzed using MedCalc statistical software (MedCalc Software Ltd, Belgium). The multifactor analysis of variance (ANOVA) was used to compare two or more groups. The results of the ANOVA were associated with p values, which were considered statistically significant when less than 0.05.

Results

Clinical and Pathological Characteristics of Adult Recurrent and Primary Pediatric PitNETs

Clinical and pathological characteristics of adult and pediatric cohorts of PitNETs are summarized in Tables 1 and 2, respectively.

Table 1 Clinicopathological and molecular features of 42 recurrent PitNET (from 24 patients)
Full size table
Table 2 Clinicopathological and molecular features of 12 pediatric primary PitNET
Full size table

Individual characteristics of the 42 PitNETs from 24 adult patients are summarized in Supplementary Table 1. Samples from both tumor primaries and one or two recurrences were available in 14/24 (58.3%) cases, while only the first recurrence was available in the remaining cases 10/24 (41.6%). Median age was 52.9 years (range 37–73); 11/24 (46%) were male and 13/24 (56%) were female.

According to pituitary hormone and transcription factors expression, adult recurrent PitNEts were classified as follows: half (12/24, 50%) were of gonadotroph origin, as indicated by clear FSH/LH positivity (n = 3) or SF1 expression (n = 9); 25% were of corticotroph origin (6/24) as indicated by ACTH secretion (n = 5) or T-PIT expression (n = 1); the remaining 25% derived from the Pit-1 lineage (6/24) including 1 GH-secreting/somatotroph (4.2%), 2 PRL-secreting/lactotroph (8.3%), 1 mixed GH/PRL-secreting/mammosomatotroph (4.2%), and 2 silent Pit-1 tumors (8,3%).

In the pediatric group, the median age was 13 years (range 5–19), and 9/12 (75%) were female. All pediatric PitNETs were functioning, including PRL-secreting (50%), followed by GH/PRL- (25%), ACTH- (16.7%), and FSH- (8.3%) secreting tumors.

Among recurrent adult PitNETs, five had a clinically aggressive course: one silent PIT1-positive tumor (n° 12) with a malignant evolution characterized at last surgery by multiple brain and spinal metastases; another invasive non-functioning PIT1-positive tumor (n° 13) recurring after 2 surgeries and radiotherapy, showing PRL immunostaining at last surgery; two invasive ACTH-secreting tumors (n° 17 and 24) with Cushing’s disease; and one giant regrowing fibrous gonadotroph PitNET (n° 18). All of them were male patients diagnosed with invasive macrotumors in their 5th/6th decade. Temozolomide (TMZ) was given in all cases with a partial response, in most cases after last surgery and together with re-irradiation and endocrinological treatment as appropriate (dopamine-agonists, somatostatin analogues, inhibitors of adrenal steroidogenesis). Nonetheless, the two patients affected by aggressive corticotropinomas eventually died from uncontrolled tumor growth and hypercortisolism. One post-operative death (n° 18) occurred due to massive stroke after transcranial surgery for tumor regrowth. Treatment and outcome data are reported in Supplementary Table 3.

Characterization of TERTp Methylation in Adult Recurrent and Pediatric Primary PitNETs

TERTp methylation of five CpG sites 600 bp upstream from the transcription start site (UTSS) was analyzed using MS-PCR (Supplementary Fig. 1). Among the adult recurrent PitNETs, we identified 6/24 patients (25%) with a tumor methylation value ≥ 30%, indicative of TERTp methylation (Table 1). The median age of patients with TERTp methylation (3 M and 3 F) was 51 years (range 41–59), with a predominance of ACTH-secreting tumors (4/6), a minority of PitNETs from gonadotroph lineage (1 FSH-secreting and 1 SF1-positive), and none derived from the Pit-1 lineage. These findings suggest that UTSS-TERTp methylation is relatively frequent in PitNETs and is not limited to a given group of hormone-secreting tumors. In addition, it was unrelated to further tumor progression or metastasis.

TERTp methylation could be studied at onset and on subsequent surgical procedures in 12 cases. Interestingly, a full concordance in TERTp methylation status was observed between primaries and recurrences in individual patients, indicating that TERTp methylation status remained stable throughout PitNETs progression. However, no significant relationship was observed between TERTp methylation status and telomere length: among the 6 TERTp methylated tumors, telomere intensity was either low (2 cases) or medium (2 cases), and the remaining 2 cases were ALT-positive (Supplementary Table 1).

Among the pediatric cases, TERTp methylation was identified in 4/12 specimens (33.3%) (2 PRL-secreting and 2 GH/PRL-secreting). The patients with TERTp methylation (3 F, 1 M) were 5, 9, 14, and 15 years old. Again, TERTp methylation showed no significant correlation with telomere length: telomere intensity could be either low (1 case), medium (2 cases), or high (1 case) (Supplementary Table 2).

Characterization of ALT and Telomere Length in Adult Recurrent and Pediatric Primary PitNETs

Using Telo-FISH, we analyzed a total of 40 samples from 24 adult patients with recurrent PitNETs (Fig. 3a). As shown in the figure, plotting the distribution of the mean telomere fluorescence intensity generated four different, clearly separate clusters (Fig. 3b).

ALT was identified in 3/24 patients (12.5%) and was present both at onset and at recurrences in individual patients, suggesting that ALT activation can be acquired as soon as the tumor develops, and remains stable during tumor progression (Table 1). Representative images of the ALT-positive tumors are shown in Fig. 2. One ALT-positive PitNET was found in a 72-year-old male patient whose disease recurred after 3 and 11 years from onset (case 2 Suppl. Table 1). Interestingly, this tumor evolved from a PRL-negative to a PRL (and Pit-1) positive phenotype, although pre-operative hyperprolactinemia was moderate. Ultra-bright telomeric signals were seen in the consecutive 3 samples, while ATRX nuclear expression was retained, suggesting that ALT was triggered by other alterations (Supplementary Fig. 2). The other two cases displaying ALT activation were invasive functioning corticotroph PitNETs and developed an aggressive behavior. The first one was observed in a 58-year-old male patient (case 17 Suppl. Table 1) with a moderate Cushing’s disease and early tumor regrowth 6 months and 1 year after first surgery. In this case, nuclear loss of ATRX was present at both onset and relapses, suggesting that ALT was triggered by an ATRX alteration (Supplementary Fig. 4). The second one was observed in a 54-year-old patient (case 24 Suppl. Tab. 1) presenting with a severe Cushing’s disease, who was re-operated 3 and 5 years after first surgery. In this case, ATRX nuclear expression was retained in all samples, suggesting alternative mechanisms for ALT activation. Both patients subsequently developed increasingly aggressive tumors and eventually died from uncontrolled tumor progression – 142 and 106 months after first surgery, respectively – despite multiple therapeutic interventions including repeated radiotherapy, temozolomide and endocrinological treatments, in the absence of documented metastasis. Overall, 3 out of 5 aggressive or malignant PitNETs were ALT-positive (Supplementary Table 3). Moreover, the mean progression-free survival (PFS) of ALT-positive cases was lower compared to ALT-negative PitNETs (30.6 months vs 81.7 months). Taken together, these observations suggest that ALT behave as an early marker of recurrent and even aggressive tumor aggressiveness.

Fig. 2
figure 2

ALT-activated samples analyzed with Telo-FISH. Three different patients (n° 13, 17, and 24) with elongation of telomeres via the ALT mechanism. H&E stains are representative of tumor samples (magnification 10 ×). Immunohistochemistry shows ATRX nuclear retention in two samples, while one (n°17) shows ATRX nuclear loss (magnification 20 ×). Telo-FISH images show neoplastic nuclei with large and very bright signals indicative of ALT (magnification 100 ×); in all patients, ALT was triggered at the onset and the mechanism persisted in the following two relapses

Full size image

ATRX nuclear expression was retained in all ALT-negative cases (21/24, 87.5%).

On the basis of mean telomere intensity (compared with that of endothelial cells), ALT-negative samples were assigned to one of the three telomere intensity categories – low (L), medium (M), or high (H). In the adult cohort with recurrent PitNETs, we found 2 H (8.3%), 12 M (50%), and 7 L (29.2%) cases, with no significant differences in the clinical, pathological, or molecular characteristics in the three groups (Fig. 3a). A full concordance in mean telomere intensity was observed between the tumors at onset and at relapse in the same patients, so that none changed telomere length category during progression (Fig. 3c); we found low telomere intensity in 1 patient, medium in 7, and high in 2, and patients were distributed in the same category.

Fig. 3
figure 3

Telomere intensity assessed in adult recurrent and pediatric primary pituitary neuroendocrine tumors. (a) Figure shows relevant clinicopathological (sex, age, hormones, transcription factors) and molecular features (Ki67, ATRX, TERTp) of adult recurrent PitNETs, grouped by mean telomere fluorescence intensity (L, M, H) and ALT activation. (b) Graph shows quantification of telomere fluorescence intensity by TFL-Telo within groups of adult recurrent samplescharacterized by low (mean: 200–500), medium (mean: 500–800), or high (mean: 800–1000) telomere intensity, and ALT (mean: > 1000); positive controls (mean = 1546, standard deviation = 393), negative controls (mean = 691, standard deviation = 172), and endothelial cells (mean = 628, standard deviation = 74) are included to establish thresholds (blue, black, and red lines). Four different, clearly separate clusters are evident. (c) Graph shows quantification of telomere fluorescence intensity by TFL-Telo in adult recurrent PitNETs analyzed both at onset and at relapses. The number of each sample is shown on the x-axis. The red circles indicate telomere intensity at the onset, the black squares, and blue triangles at first and second recurrences, respectively. The analysis shows a strong concordance in mean telomere intensity between onset and relapses for each sample analyzed. (d) Pie charts represent percentages of patients analyzed with ALT, or low, medium, or high telomere intensity for the whole cohort of adult recurrent PitNETs (left), adult PitNETs analyzed only at the onset (middle), and pediatric primary PitNETs (right)

Full size image

Regarding the pediatric PitNETs, no case of ALT was identified (0%), and nuclear expression of ATRX was retained in all samples (Supplementary Table 2 and Supplementary Fig. 3). Telomere length was classified as H in 2/12 cases (16.6%), M in 7/12 cases (58.4%), and L in 3/12 cases (25%), respectively (Fig. 3d). As in the adult cohort –no significant differences were found in patients’ molecular or clinicopathological characteristics according to telomere intensity.

Comparing primary PitNETs in the pediatric and adult cohort (Fig. 3d), a higher proportion of L telomeric profiles was observed among the pediatric tumors (25% vs. 7% of adult cases at the onset), while the proportion of pediatric cases with M or H telomeric profiles was similar to their adult counterpart (M: 58.4% vs. 57% of adult cases at the onset; H: 16.6% vs. 14% of adult cases at the onset). The main difference between primary pediatric and primary adult PitNETs was the absence of ALT in the first (ALT: 0% of pediatric vs. 22% of adult cases at the onset) (Fig. 3d). The higher proportion of tumors with low telomere intensity and the absence of ALT in pediatric PitNETs suggest that telomere maintenance is regulated differently in younger patients.

Discussion

PitNETs form a heterogeneous group of neoplasms derived from the adenohypophysis, with variable secreting, proliferative, and invasive patterns. In the WHO Classification of Tumors of Endocrine Organs (2017), they are classified according to their functional lineage, based on immunohistochemical expression of hormones and transcription factors [2, 8,9,10,11, 19]. About 40% of PitNETs are invasive, and a minority of them are clinically aggressive, defined by unusual rapid growth and multiple recurrences. Guidelines for the treatment of aggressive and malignant PitNETs have been proposed by the European Society of Endocrinology and include chemotherapy with temozolomide as a first-line treatment regardless of metastasis [63]. Where temozolomide fails to control disease progression, innovative molecular target therapies are being considered. Since aggressive and metastatic PitNETs are currently undistinguishable from a pathological point of view and are likely to be “the two sides of the same coin” [2], a key point is to identify early alterations associated with such an aggressive/metastatic potential, in order to adapt their follow-up and treatment. Therefore, ALT activation is an attractive candidate. Based on the role of telomere abnormalities in a variety of neoplastic conditions, we analyzed ALT and TERTp methylation in a series of recurrent PitNETs and in a series of sporadic pediatric cases.

To date, only one study has reported on the presence of ALT in pituitary tumors [49]. The authors found ALT in 3 (2.8%) of 106 PitNETs, 2 of them being recurrent. The series included a majority of non-functioning PitNETs, and ALT-positive tumors were either non-functioning (2 cases, undefined lineage of origin) or GH-secreting (1 case). These results led them to analyze a second cohort of 32 tumor samples from 22 patients with recurrent PitNETs, and they found that 8 (36%) of them were ALT-positive, leading them to conclude that ALT was enhanced in recurrent pituitary tumors [49] – the functional phenotype of this second group of ALT-positive PitNETs and patient’s gender and age were not specified. Our study also revealed ALT activation in samples from 3/24 patients with recurrent PitNETs (12.5%), including 2 male corticotroph and one male lactotroph tumors. This indicates that ALT may occur in PitNETs derived from different lineages. Noteworthy, we found ALT activation in both primary and recurrent tumor samples, whereas in the second cohort reported by Heaphy et al. 2020, ALT was also present since first surgery in 2 cases, but acquired during progression in 1 case [49]. Taken together, these findings support the hypothesis that ALT activation mostly occurs at an early stage of PitNETs tumorigenesis and then represents a stable mechanism maintained during disease progression, although secondary activation may occur later in tumor evolution. Since we observed ALT activation in a Pit1/PRL-positive aggressive PitNET, but not in the silent Pit1 metastatic PitNET, ALT may not be directly involved in metastatic progression. However, 2/3 ALT-positive PitNETs in our series died from aggressive corticotropinomas after 142 and 106 months, respectively, and the progression free survival (PFS) of ALT-positive cases was lower compared to ALT-negative PitNETs (mean 30.6 vs 81.7 months, respectively). Such findings suggest that ALT activation may be a risk factor for aggressive evolution in PitNETs, and not only for recurrence. In contrast, we studied for the first time ALT activation in an unselected series of pediatric PitNETs, and ALT was negative in all cases, suggesting that this is not a characteristic of early-onset PitNETs.

The mechanisms of ALT activation in PitNETs are still poorly known. In our series, only 1/3 case of ALT activation was associated with ATRX loss. In the study by Heaphy et al. 2020 [49], 2 out of the first 3 ALT-positive cases showed ATRX protein loss in the absence of ATRX mutation, while the third had normal ATRX expression and sequencing, but harbored a somatic mutation in EP300, a histone acetyl transferase that regulates transcription via chromatin remodeling [49]. ALT activation was also independent from DAXX mutations [49]. Therefore, the hypothesis is that ALT activation in PitNETs may be driven by additional unrecognized somatic mutations, as reported in other tumors (e.g., SMARCAL1, SLX4IP H3.3, or SLX4IP) [48, 49, 51, 52, 57]. Partial ATRX loss was also observed in a recurrent PitNET with acquired ALT, further supporting its causative role [49], but ATRX mutations by direct sequencing were not studied, which is a limit of our study. Because loss of ATRX nuclear expression on immunohistochemistry can be used as a surrogate marker of ATRX mutations with high sensitivity and specificity in gliomas [33, 44, 47], the observation of ATRX loss in the absence of mutations reported by Heaphy et al. is intriguing. This may be uncommon, since ATRX mutations were found in all cases of ATRX loss by immunohistochemistry in a recent series of aggressive and malignant PitNETs [56] and would deserve further investigation.

Overall, ATRX tumor expression was retained in 95.8% of the patients in our cohort – including the single case of metastatic PitNET – and in all ALT-negative cases, confirming that loss of ATRX is very rare in adult recurrent PitNETs. In this study, we observed ATRX associated with ALT in an aggressive male corticotropinoma, eventually leading to patient’s death. ATRX expression was also retained in a large series of 246 pituitary tumors, except one case displaying ATRX mutation and loss of nuclear expression in both primary and metastatic samples, suggesting a possible relationship between loss of ATRX and metastatic behavior [54]. The negative prognostic value of ATRX mutations is supported by other studies, although ALT was not considered [54,55,56]. Guo et al. found a somatic mutation in ATRX in an ACTH-secreting metastatic PitNET (in association with PTEN and p53 mutations) [55]. Casar-Borota et al. reported that ATRX mutations occur in a subset of aggressive PitNETs and carcinomas, especially corticotroph tumors (7/9) [54, 56]. Taken together, these findings suggest that ATRX loss may be especially involved in the aggressiveness of corticotropinomas. However, an aggressive prolactinoma with a fatal outcome and another metastatic somato-lactotroph PitNET were also found to harbor an ATRX mutation [56]. And, in contrast with our series, ATRX loss has also been reported in 3 prolactinomas issued from a series of 42 young-onset PitNETs (children and adolescents) [53]. Loss and/or mutation of ATRX is therefore not limited to corticotroph PitNETs and may be useful indicators of potential aggressiveness and/or metastatic potential in different functional phenotypes.

ALT-negative samples of adult recurrent PitNETs (87.5%) were classified according to their mean telomere intensity using the TFL-Telo software, and either low (29.2%), medium (50%), or high (8.3%) intensity was observed. No significant differences were found between the 3 groups of telomere intensity in terms of clinical or molecular features, such as patient’s age and tumor functional phenotype, Ki67, P53, or ATRX immunostaining. A full concordance in mean telomere intensity was observed between primary tumors and relapses in individual patients, suggesting that, similarly to ALT, telomere length generally remains stable as the disease progresses. Analogous findings were found in our pediatric cohort of PitNETs (all ALT-negative), which displayed either low (25%), medium (58.4%), or high (16.6%) telomere intensity, in the absence of significant clinical or biological differences among the 3 groups.

The biological significance and potential mechanisms of telomere length alterations on PitNETs are not fully understood. Both TERT expression and increased telomerase activity have been reported, especially in functioning tumors [64, 65]. TERTp mutations appear to be almost absent in PitNETs [38,39,40,41, 58]. In addition to TERTp mutations, TERTp methylation provides an additional regulatory mechanism for telomerase upregulation in many cancers [25, 31], although TERTp alterations do not always sufficiently activate telomerase expression to counteract telomere shortening [59]. We found that adult recurrent PitNETs harbored high levels of TERTp methylation in the UTSS region in 25% of cases, which is similar to previous data from the literature (16–27%) [41, 58], and no significant difference in telomere length or clinical/pathological characteristics was found between TERTp methylated and unmethylated cases. We suggest that TERTp methylation alone is not sufficient to activate elongation of telomeres, supporting that telomere length maintenance via telomerase in PitNETs may require upregulation promoted by multiple steps [59]. Previous larger studies included a majority of primary tumors; Miyake et al. reported that TERTp methylation (16% overall) was significantly more frequent in recurrent tumors, associated with a shorter progression-free survival and with upregulated TERT expression, suggesting that TERTp methylation could be a useful marker for predicting tumor recurrence in PitNETs [58]. In contrast, Köchling et al. observed no correlation between TERTp methylation (27% overall) and tumor recurrence, hormone secretion, or proliferation index, arguing against its prognostic value; nevertheless, intriguingly, reported that TERTp methylation was significantly more frequent in male primary tumors (40% vs 13% in female) [41]. They also noticed that TERTp was methylated in one metastatic PitNET and unmethylated in another one, which is similar to the metastatic case reported in our study [41]. In agreement with this study, we also found TERTp methylation status to be unchanged during tumor progression [41]. In addition, we observed TERTp methylation in 33% of pediatric PitNETs, which is slightly higher than in adult recurrent cases, but this result should be confirmed in larger series and compared with primary adult tumors. Overall, based on available data, the role of TERTp alterations in telomere elongation in PitNETs remains uncertain and would deserve further investigation.

In conclusion, our study shows that adult recurrent PitNETs may feature ALT activation, a telomere maintenance mechanism which appears to occur early in tumor development and maintained as the disease progresses. Moreover, ALT activation was associated with tumor aggressiveness and reduced progression free survival. It may therefore be useful to analyze ALT in the diagnostic workup of adult invasive PitNETs as a potential predictor of aggressive behavior. At present, ALT has not been involved in pediatric PitNETs, but further studies may be useful to confirm this point. Indeed, several ALT-targeted drugs have recently been tested on several types of cancer, including recombination inhibitors, histone deacetylase (HDAC) inhibitors, or G-quadruplex stabilizer [60,61,62], which may represent a promising strategy for the treatment of aggressive ALT-positive PitNETs.