Somatic Activating ESR1 Mutation in an Aggressive Prolactinoma

Abstract

Context and Objective

The genetic profile of prolactinomas remains poorly understood. Our objective is to identify somatic genetic alterations associated with prolactinomas and to report the identification of an activating ESR1 mutation (ESR1_Y537S) in an aggressive prolactinoma.

Setting

Brigham and Women's Hospital.

Design

Massively parallel-sequencing panel (OncoPanel) was performed in a cohort of patients with prolactinomas to identify mutations and copy number variation.

Results

Twenty subjects (mean age, 38.6 years; 12 women and 8 men) were included in this study. A somatic ESR1_Y537S mutation was identified in an aggressive prolactinoma in a postmenopausal woman. No SF3B1 or other somatic mutations were identified. The median number of copy number variation events identified in our samples was 46; the prolactinoma with ESR1_Y537S had the highest number with 233 events. In breast cancer, ESR1_Y537S has been shown to activate estrogen receptor alpha independent of ligand binding. In patients with resistant breast cancer and ESR1_Y537S, elacestrant, a second-line estrogen receptor degrader, improves progression-free survival. Therefore, given the lack of response to multimodality therapies, elacestrant was initiated in this patient after the third cycle of radiotherapy. Elacestrant, along with radiotherapy, controlled tumor growth and significantly reduced prolactin levels.

Conclusion

Molecular profiling allowed the identification of ESR1_Y537S, in an aggressive prolactinoma. ESR1_Y537S was not detected early in the course of the disease and is likely conferring tumor aggressiveness. This finding emphasizes the significance of estrogen receptor signaling in prolactinomas. It also allowed the use of targeted therapy with successful control of disease progression.

Prolactin-secreting adenomas (prolactinomas) are the most common type of pituitary adenoma, accounting for 50% to 60% of all pituitary tumors (1, 2). Eighty percent of prolactinomas are smaller than <10 mm (microadenomas), and they are more frequent in women (3). Prolactinomas can cause hypogonadism and galactorrhea (4). They arise from lactotrophs, which express estrogen receptor α (ERα), encoded by ESR1 (5). Prolactinomas are most frequently diagnosed in women of reproductive age, with up to a 10-fold higher incidence than in men in the same age group; this sex bias is no longer present after menopause (4). Treatment with dopamine agonists leads to tumor regression and hormonal control in about 90% of patients (6, 7). The few patients (10%) with tumors resistant to medical therapy may experience an aggressive disease course, with tumor invasion of surrounding structures and multiple recurrences (1, 6, 8, 9). Sex-related differences have been reported in relation to tumor behavior, with men having a more aggressive course and more treatment-resistant prolactinomas, compared to premenopausal women (1, 4, 10).

The molecular mechanisms underlying prolactinoma tumorigenesis and predictors of disease aggressiveness are poorly understood (4). The age-dependent sex bias in prolactinoma incidence suggest that estrogen may play a role in tumor behavior. Physiologically, estrogen stimulates lactotroph cell proliferation (11, 12). Additionally, estrogen induces prolactin transcription, synthesis, and secretion, which explains why, in the case of hyperprolactinemia, prolactin levels can normalize postmenopausally (13-15). Furthermore, estrogen may play a role in tumor growth during pregnancy, which occurs more often with macroadenomas (16).

Prolactinomas are thought to be monoclonal neoplasms; a somatic genetic event may result in a growth advantage over surrounding pituitary cells (5, 17). Few recurrent alterations in sporadic prolactinomas have been described and are limited to recent identification of a recurrent somatic mutation in splicing factor 3 subunit B1 (SF3B1), predominantly in aggressive prolactinomas (18, 19). Prolactinomas with SF3B1 mutation were associated with higher prolactin levels and poor progression-free survival (18). Molecular studies suggest that SF3B1 participates in prolactinoma tumorigenesis by activating estrogen signaling pathways (18).

The aim of this study was to identify somatic genetic alterations associated with prolactinomas using a gene panel. The striking finding of the present study was the identification of a somatic ESR1 mutation in an aggressive prolactinoma in a postmenopausal woman.

Materials and Methods

Cohort

Patients who underwent surgery to remove a pituitary tumor at the Brigham and Women's Hospital (BWH) between 2013 and 2020 were offered the option to participate in the tumor profiling genomic initiative (OncoPanel) (20). We selected all patients (n = 20) in the entire OncoPanel cohort diagnosed with prolactinoma to include in our study, some of whom were included in a previous publication (21).

Reviews of patients' charts were performed and clinical parameters including sex, age of diagnosis, adenoma size, disease remission after surgery, and Ki-67 proliferation index were collected for all patients. Tumor size was determined before the surgery from which the molecular evaluation was performed and classified as microadenoma (<1 cm) or macroadenoma (≥1 cm). Persistent disease after surgery was determined based on elevated prolactin levels after surgery, with or without radiological evidence of the residual tumor. Insufficient response to dopamine agonists was determined by the primary physician.

The diagnosis of prolactinoma was made based on the presence of both elevated prolactin levels and pathologic diagnosis of pituitary adenoma with immunostaining for prolactin in the tumor specimen.

This study was approved by the BWH and Dana Farber Cancer institute institutional review boards and conducted with written informed consent in accordance with the Declaration of Helsinki.

Prolactin Assays

Prolactin measurements were performed using different assays over time. In the patient with the ESR1_Y537S prolactinoma, prolactin was measured in a laboratory outside our institution between 2002 and 2006. The reference range of this immunoassay provided by the laboratory was 3 to 30 ng/mL. From 2006 to July 2011, prolactin levels were measured using Access chemiluminescent immunoassay (Beckman Coulter, Fullerton, CA) with a normal range of 2.7 to 26.6 ng/mL. Since August 2011, prolactin levels were measured using an electrochemiluminescence immunoassay “ECLIA” (Elecsys Prolactin II assay, Cobas, reference 03203093-190). The reference range for this method is 4.79 to 23.3 ng/mL; this range encompasses 95% of subjects, excluding the lowest 2.5% and those higher than 97.5%.

Immunohistochemistry

Immunohistochemistry was performed on 4-µm formalin-fixed paraffin-embedded tissue sections following pressure cooker antigen retrieval (pH 6.1 citrate buffer; Target Retrieval Solution, Dako) using a monoclonal mouse anti-Ki-67 antibody (1:400 dilution; clone MIB-1; Agilent, RRID: AB_2890068), a polyclonal rabbit anti-prolactin antibody (1:500 dilution; Agilent, RRID: AB_1542808), and a monoclonal rabbit anti-ER antibody (1:30 dilution; clone SP1; Biocare Medical, RRID: AB_843887). The anti-ER antibody epitope is located at the C-terminus of the ER protein, which is distal to the ligand binding domain where the mutation is located, allowing ER staining to remain positive despite the ESR1 mutation (22). Appropriate positive and negative controls were used throughout.

Genomic Analysis

OncoPanel, a massively parallel-sequencing panel, was performed to identify genomic variants and copy number variation (CNV). The OncoPanel was part of the Dana-Farber/BWH Cancer Center's PROFILE clinical research study, which included genetic profiling of canonical cancer-associated genes (20, 21, 23, 24). OncoPanel included 275 genes in version 1, 300 genes in version 2, and 447 genes in the third, current version (Supplementary Table S1) (25). In our series of 20 prolactinomas, the OncoPanel version 1 was used for tumor profiling in 2 specimens, version 2 in 13, and, finally, version 3 in 5 specimens. Importantly, SF3B1 and ESR1 genes were included in all 3 versions of OncoPanel.

Genomic DNA was isolated from formalin-fixed, paraffin-embedded tumor tissue using a Qiagen kit (56704) as described (20). The samples were analyzed in the Center for Advanced Molecular Diagnostics at BWH, a CLIA-certified laboratory (20). Genomic analysis of tumor DNA was performed following a previously described protocol (20, 21).

Genomic Analysis Focusing on ESR1 Mutation

Molecular analysis of blood DNA was performed to characterize the identified ESR1_Y537S mutation as somatic or germline. Genomic DNA was isolated from peripheral blood leukocytes using Qiagen kits (55114) as described (20). The ESR1_Y537S flanking region was amplified by PCR using the following primers: forward 5′-CAAAGGCATGGAGCATCT-3′, reverse 5′-TTATCTGAACCGTGTGGGAG-3′; with a melting temperature of 55 °C. Subsequently, Sanger sequencing was performed at the DF/HCC DNA Resource Core.

We then aimed to investigate whether ESR1_Y537S may have driven initial tumor formation or whether it was acquired as the tumor progressed, as usually seen in breast cancer (see the following section). To this end, we extracted DNA from tumor samples from the patient's first and second surgeries, amplified the region flanking ESR1_Y537S by PCR using the same primers described previously, and performed Sanger sequencing. Sanger sequencing results were confirmed by droplet digital PCR (ddPCR), using ddPCR Supermix (Bio-Rad) and custom TaqMan probes and primers, as previously described (26, 27). The following primers were used to perform ddPCR: forward 5′-CAGCATGAAGTGCAAGAACGT-3′, reverse: 5′-TGGGCGTCCAGCATCTC-3′. Probes for ddPCR were: 5′-VIC-CCCCTCTATGACCTGC-MGB-NFQ-3′ (Y537wt) and 5′-FAM-CCCTCTCTGACCTGC-MGB-NFQ-3′ (Y537S). Gene transcript ID: ENST00000206249.8. The reference sequence GRCh38/hg38 was used.

Liquid Biopsy

To detect the circulating somatic variant ESR1_Y537S in the patient's plasma, cell-free DNA (cfDNA) analysis was performed. A peripheral blood sample (6-10 mL) was collected, and plasma was isolated within 2 hours and then stored at −80 °C until processing. Cell-free DNA was extracted from plasma samples using the QIAmp Circulating Nucleic Acid Kit (Qiagen) according to the manufacturer's protocol. A sensitive assay using ddPCR was performed, as previously described and used clinically in patients with breast cancer (26, 27). High sensitivity and specificity of this method have been demonstrated in a cohort of patients with metastatic breast cancer (26). The detection threshold of ddPCR is 10 genome copies/mL plasma (26). Briefly, TaqMan PCR reaction samples were made using ddPCR Supermix (Bio-Rad) and custom TaqMan probes and primers prepared specifically for this assay. The PCR plate was set using ddPCR reaction mixture with cfDNA samples. The heat-sealed PCR plate was placed on the Automated Droplet Generator (Bio-Rad) and then on a conventional thermal cycler for amplification. The PCR plate was read on the QX200 droplet reader (Bio-Rad) and the analysis was performed using QuantaSoft software (Bio-Rad). Samples were studied in triplicate.

Statistical Analysis

Categorical variables were described using proportions, whereas continuous variables used mean ± standard deviation or median and interquartile range. To evaluate differences between sexes, a 2-sided t test was used and nonparametric test was used for continuous variables. In the statistical analysis of CNV, 1 outlier was identified and was excluded from the comparison analysis between males and females. Statistical significance was defined as P < .05. IBM SPSS Statistics, version 28.0.1.0, was used for statistical analysis.

Results

Clinical and Pathological Characteristics

In our cohort of 20 patients with surgically resected prolactinomas, the mean age at the time of diagnosis was 38.6 ± 14.5 and 60% of our patients were female. Men were diagnosed with prolactinomas at an older age compared to women (mean: 48.4 ± 13.8 vs 32.1 ± 11.2 years; P = .016 (Fig. 1A). Reasons for surgery included insufficient response to dopamine agonist in 8 patients (40%), drug intolerance in 9 patients (45%), tumor apoplexy and acute visual deterioration in 2 patients (10%), and personal preference in 1 patient (5%). The clinical and pathological characteristics of our cohort are summarized in Table 1. Radiological invasion into the sphenoid and/or cavernous sinus was detected in 5 (25%) patients; 4 of them were male. Fifteen patients (75%) underwent 1 surgical procedure and 5 patients (25%) had 2 or more surgeries performed. Three patients (15%) needed radiotherapy. Four prolactinomas had Ki-67 ≥ 10%. Two patients in our cohort were considered to have aggressive tumors, according to the European Society of Endocrinology guidelines (28). The first was the patient with the tumor harboring the ESR1 mutation. The second was a male patient who received multiple modalities of treatment, including 5 surgeries, radiotherapy, temozolomide, and nivolumab, without successful control of the prolactinoma. The tumor had extensive local invasiveness (temporal fossa, petrous apex, and the right middle ear) and a Ki-67 of 7.5%. The patient died from tumor progression 4 months after the fifth transsphenoidal surgery at the age of 59 years.

Figure 1.

Age at diagnosis and copy number variation (CNV) in males and females. (A) Age at diagnosis of prolactinoma in males and females. yrs, years. Values represent mean ± SD. Statistical analysis was performed by unpaired 2-tailed t-test; *P = .016. (B) The number of CNV events in resected tumor samples in males and females. The outlier represents the prolactinoma harboring the ESR1_Y537S mutant. Statistical analysis was performed using the Mann-Whitney test. The outlier was excluded from the statistical analysis; *P = .016.

Table 1.

Clinical and pathological characteristics of the cohort

	Total	Male	Female	P value
Patients, n (%)	20	8 (40%)	12 (60%)
Age at diagnosis, y, mean ± SD	38.6 ± 14.5	48.4 ± 13.8	32.1 ± 11.2	.016
Insufficient response to dopamine agonist, n (%)	8 (40%)	3 (37.5%)	5 (41.7%)	.852
Macroadenoma, n (%)	15 (75%)	7 (87.5%)	8 (66.7%)	.603
Radiologically invasive tumor, n (%)	5 (25%)	4 (50%)	1 (8.3%)	.038
Ki-67 >3%, n (%)	11 (55%)	3 (37.5%)	8 (66.7%)	.362
Ki-67 >10%, n (%)	4 (20%)	1 (12.5%)	3 (25%)	.619
CNV, median (IQR)	40 (1-80)	68.5 (48-103.5)	19 (1-52)	.016
Persistent disease after surgery, n (%)	10 (50%)	6 (75%)	4 (33%)	.17

Radiologic invasion of the sphenoid and/or cavernous sinus in MRI was not assessed in 1 male patient. Ki-67 proliferation index was determined in the tumor sample in which the genomic profiling was performed; 5 of them were recurrent tumors. The 4 tumors with Ki-67 >10% were also included in the group of tumors with Ki-67 >3%. Persistent disease after surgery was determined based on persistent elevation in serum prolactin levels after surgery, with or without radiological evidence of the residual tumor.

Abbreviations: CNV, copy number variation; IQR, interquartile range; SD, standard deviation.

OncoPanel Mutation Analyses

Of the 20 tumors subjected to OncoPanel analysis, a missense mutation in the gene encoding ERα, ESR1_Y537S, was identified in the tumor specimen from a female patient obtained following her fourth surgery, at an allele frequency of 29% in 405 reads (Figs. 2 and 3). This tumor had a mutation burden of 5.323 mutations/Mb. Other variants identified in this tumor are reported in Supplementary Table S2 (25).

Figure 2.

Twenty one-year clinical summary of a patient with an aggressive prolactinoma harboring an ESR1_Y537S mutation. Top: Timeline to depict important events and treatments during the patient's course over time. Bottom: T1 coronal sections of brain MRIs. White arrows point to the adenoma. (A) November 2004, 21 months after initial transsphenoidal surgery (TSS): hypo-enhancing 7-mm lesion within the left inferior lateral aspect of the pituitary gland representing residual pituitary adenoma. (B). June 2009, MRI before the second TSS: extensive, irregular sellar lesion measuring 3.7 × 1.6 × 1.8 cm with superoposterior extension ascending into the third ventricle, left lateral expansion into the sphenoid bone and filling the left sphenoid sinus. (C) August 2009, 2 months after the second TSS: superior intracranial tumor remnant that could not be removed because it had invaded arterial branches from the basilar artery. (D) July 2010, preoperative MRI before the third surgery (craniotomy): giant 5-cm macroadenoma, contacting and slightly displacing the optic chiasm and tracts, encasing left cavernous internal carotid artery. (E) July 2016, 5 years after stereotactic radiation therapy (SRT): 0.8 × 0.6-cm lesion of residual adenoma. (F) December 2019, preoperative MRI before the fourth TSS: 4.3-cm lesion with increasing outward protrusion into the sphenoid sinus and persistent mass effect on the cavernous segments of the internal carotid arteries bilaterally. (G) January 2021, 1 year after the fourth TSS: adenoma measuring up to 6.1 cm, involving the bilateral cavernous sinuses, the clivus, and sphenoid sinuses with elevation of the pituitary stalk and optic chiasm. (H) February 2022, 1 year after the second SRT, showing a significant decrease in adenoma size to 3.7 cm. (I) January 2023, 2 years after the second SRT, showing a recurrent increase in the size of prolactinoma with involvement of bilateral cavernous sinuses and clivus and extension into the suprasellar and prepontine cisterns. (J) September 2023, 5 months after the third cycle of SRT, showing prolactinoma invading the cavernous sinus with extension along the optic apparatus, involvement of the bilateral internal carotid arteries, and extension into the sphenoid/nasal cavity. CAB, cabergoline; IMRT, intensity-modulated radiation therapy; PRL, prolactin; SRT, stereotactic radiation therapy; TMZ, temozolomide.

Figure 3.

Serum prolactin measurements throughout the patient's course and in relation to treatment. The patient was treated with cabergoline at doses of up to 6 mg/week until May 2024, except from 2006 to 2008, when she was lost to follow-up. Dashed line corresponds to the upper limit of reference for serum prolactin (normal range, 2.8-29.2 ng/mL). IMRT, intensity-modulated radiation therapy; TMZ, temozolomide; PRL, prolactin; CAB, cabergoline; TSS, transsphenoidal surgery; SRT, stereotactic radiation therapy.

We first aimed to establish whether ESR1_Y537S was a germline or somatic mutation. Sanger sequencing of germline DNA extracted from blood did not identify the mutation, confirming that ESR1_Y537S was a somatic mutation (Supplementary Fig. S1A) (25).

Subsequently, through analysis of tumor samples from the patient's previous surgeries, we assessed whether ESR1_Y537S may have driven initial tumor formation or whether it was acquired as the tumor progressed. Sanger sequencing and ddPCR analyses of DNA from resected tumor tissue from the first 2 surgeries were performed; ESR1_Y537S was not identified (Supplementary Figs. S1B and 1C) (25). Tissue from the third surgery was not available for genetic analyses.

No mutations in SF3B1 or other pathogenic somatic variants were identified in our cohort. It is important to highlight that no somatic variants considered pathogenic were identified in the male patient (described earlier) with an aggressive prolactinoma. Two female patients harbored germline variants in MEN1 (p.E45V and p.Q447*), together with loss of heterozygosity in MEN1 in the prolactinoma. One female patient harbored a germline BRCA1 mutation (p.E23Vfs*17). These germline mutations were reported in a previous study (21).

Copy Number Variation Analysis

High numbers of copy number variation (CNV) events were identified in our samples, with a median of 46 events, with 64% gene amplifications and 36% gene losses in the cohort. Men had significantly more CNV events compared to women (median, 68.5 vs 19.0; P = .016; Fig. 1B). There was no association between the number of CNV events and persistent disease postsurgery (P = .48), radiological tumor invasion (P = .13), or Ki-67 > 3% (P = .82). However, the patient with the aggressive prolactinoma harboring ESR1_Y537S had the highest number of CNV events in the entire group of pituitary adenomas subjected to OncoPanel analysis at our institution (data not shown).

Loss of chromosome 11p was detected in 6 (30%) of the adenoma samples. Of these 6 patients, persistent disease after surgery and Ki-67 > 3% were both observed in 4 (66%) patients. A greater proportion of men had loss of chromosome 11p compared to women, although this difference was not statistically significant (37.5% vs 25%, respectively; P = .64). Another recurrent chromosome alteration was a gain of chromosome 9 identified in 7 (35%) patients ( 4 men and 3 women). Of these 7 patients with a gain of chromosome 9, 3 (43%) had persistent disease after surgery, which is comparable with the overall rate of persistent disease in our cohort (50%). (Table 1)

The prolactinoma harboring the ESR1 mutation had the highest number of CNVs of the entire cohort, with 233 events (Fig. 1B). This tumor genome exhibited large chromosomal alterations, including gain of whole chromosomes 5, 7, and 8; gain of chromosome arms 1q, 2q, 14q, and 20q; loss of whole chromosome 10; and loss of chromosome arms 3q, 4q, 9p, and 11p (Fig. 4). At the gene level, there was an extensive list of genes with amplifications or deletions. Genes targeted by copy number alteration are further detailed in Supplementary Table S3 (25).

Figure 4.

Copy number variation (CNV) in the patient's tumor specimen obtained from her fourth surgery. Genomic distribution of somatic CNV, color-coded for chromosomes. The vertical axis is the log base 2 ratio of sample copy number relative to a pooled normal control. The value of 0 on the vertical axis indicates no difference from normal (diploid). Each dot represents 1 event; 0 represents absence of CNV; dots above 0 represent copy number gain, and dots below 0 represent copy number loss.

The male patient with an aggressive prolactinoma (described previously) had 54 CNV events. The sample from his fourth surgery was subjected to array comparative genomic hybridization, which demonstrated a whole-arm gain of 1q, 104.3 Mb single-copy gain of 3q, 63.0 Mb single-copy loss of 10q leading to PTEN single copy loss, and 26.4 Mb single-copy 13q deletion leading to RB1 single copy loss.

Identification of ESR1_Y537S in Cell-free DNA

The aggressive and rapidly progressive growth of the prolactinoma harboring the ESR1 mutation led us to investigate the presence of cell-free tumor DNA in the circulation. For this purpose, plasma was collected 11 months after the patient's fourth surgery. ddPCR detected a mutant allele frequency of the ESR1_Y537S of 0.65% in the plasma of this patient, corresponding to 76 genome copies/mL plasma. The identification of a mutant allele frequency between 0.02% to 1% and greater than 10 genome copies/mL plasma are considered positive.

Immunohistochemistry of the ESR1_Y537S prolactinoma

The histopathology of the initial tumor and subsequent samples revealed disruption of normal tissue architecture (Fig. 5, hematoxylin and eosin). Prolactin immunohistochemistry was diffusely positive in all samples (Fig. 5, prolactin). Interestingly, the Ki-67 proliferation index was much lower in the initial tumor and became markedly increased in subsequent recurrences (Fig. 5, Ki67). ERα staining was diffusely positive in all samples analyzed (Fig. 5, ERα). Our method does not allow ERα quantification.

Figure 5.

Representative histopathology of initial tumor (2003) and subsequent recurrences (2009, 2010, and 2020). Photomicrographs of prolactinoma samples over time. Hematoxylin and eosin stains showing disruption of the normal pituitary architecture; monomorphic sheets of large cells with hyperchromatic nuclei and somewhat basophilic cytoplasm. Prolactin immunohistochemistry was diffusely positive. Ki67 staining, reflecting tumor cell proliferation, was low in 2003 (1.1%) as typically seen in pituitary adenomas. In 2009, 2010, and 2020, Ki67 was atypically high: in 2009 and 2010, Ki67 was 10%; in 2020, in the sample in which the ESR1 mutation was identified, Ki67 was 15%. ERα (estrogen receptor alpha) staining was diffusely positive in all samples, including those before the identification of the mutation, suggesting that this ESR1 activating mutation does not alter antibody binding to ERα. The ER antibody epitope is located at the C-terminus of the protein, which is distal to the ligand binding domain where the mutation is located, allowing ER staining to remain positive despite the ESR1 mutation. Our method of immunostaining does not allow ERα quantification. ERα staining was not performed in the sample from 2010.

Clinical Course of the Patient With a Prolactinoma With ESR1_Y537S

The patient initially presented in 2003 at age 48 years, when she was premenopausal, for evaluation of galactorrhea, amenorrhea, and elevated serum prolactin of 118 ng/mL (reference range, 3-30 ng/mL) (Figs. 2 and 3). She reported a history of regular menstrual cycles and 3 pregnancies, not requiring fertility induction. The initial brain magnetic resonance imaging (MRI) showed a pituitary adenoma measuring 1.5 × 1.5 cm (the image was from an outside hospital and was not available). She underwent transsphenoidal pituitary surgery (TSS) in February 2003 and pathology confirmed diagnosis of prolactinoma (Fig. 5). There was no report of tumor invasiveness on the initial MRI and pathology reports. After surgery, prolactin levels remained elevated, 85 ng/mL (reference, 3-30 ng/mL). Treatment with cabergoline was initiated, with reduction of prolactin levels but not to the normal range The patient's last menstrual period reported was in October 2004, at age 49 years, which may have been related to menopause or secondary to elevated prolactin levels. An MRI scan in November 2004 showed a right sellar lesion (Fig. 2A, bottom). She was lost to follow-up until May 2008, when she returned with persistent hyperprolactinemia (prolactin 800 ng/mL), and treatment with cabergoline was restarted at a dose of 1 mg/week, which was up-titrated to 2 mg/week. In June 2009, prolactin levels were still elevated despite cabergoline treatment, and MRI revealed a large prolactinoma (Fig. 2B, bottom). Subsequently, the patient received multiple modalities of treatment available for aggressive pituitary adenomas, without adequate control of tumor progression. Over the course of her care, she underwent 4 surgeries for tumor removal (3 TSS and 1 craniotomy), medical therapy with maximum tolerated doses of cabergoline (up to 6 mg weekly), medical therapy with temozolomide, lanreotide, and everolimus, 3 cycles of radiation therapy (2 courses of stereotactic radiation [2011/2023], and 1 course of intensity-modulated radiation therapy (2021) (Figs. 2 and 3) (28, 29). The pathology report from the first TSS revealed a Ki-67 proliferative index of 1%; after the second TSS, Ki-67 was at least 10%, which is high for pituitary adenomas (Fig. 5, Ki67). During her disease course, the patient developed several complications secondary to tumor growth and/or radiotherapy, including panhypopituitarism, bilateral optic neuropathy with severe vision loss, left third and sixth cranial nerve palsies, cognitive decline, and cerebrospinal fluid leak. She had temporary tumor stabilization after each cycle of radiotherapy, with prolonged control of tumor growth and prolactin levels for 7 years after the first stereotactic radiation, but thereafter the prolactinoma continued to grow. No evidence of metastasis was detected on a thoracic and abdominal computed tomography scan. The activating ESR1 mutation, ESR1_Y537S, was identified in tumor tissue resected during her fourth surgery. In January 2023, elacestrant, a second-line ERα degrader, was approved by the Food and Drug Administration for treatment of advanced or metastatic breast cancer harboring an ESR1 mutation. Given her lack of response to currently available therapies and the identification of ESR1_Y537S together with ERα staining in the tumor (Fig. 5, ERα), off-label elacestrant was initiated at a daily oral dose of 345 mg (standard dose used to treat breast cancer) in June 2023 after the third cycle of radiotherapy was completed. Her prolactin levels decreased significantly, to less than 200 ng/dL, 1 year after the third cycle of radiotherapy and 4 months after the initiation of elacestrant (Fig. 3). As of August 2024, she has remained neurologically stable, with progressively lower serum prolactin levels and a stable tumor volume noted on the most recent brain MRI.

Discussion

Prolactinomas are diagnosed predominantly in premenopausal women (4). Most prolactinomas are easily managed with medical treatment with dopamine agonists; however, a subset of prolactinomas, especially in men, are resistant to this conventional treatment (6). Recent studies identified SF3B1 mutations associated with aggressive prolactinomas (18, 19).

In this study, we used OncoPanel to screen for somatic mutations in our cohort of patients with prolactinomas with tissue available from surgery. Our cohort does not reflect the classic history of prolactinomas with a predominance of microadenomas, a higher proportion of females compared to males, and straightforward management with satisfactory results (4). On the contrary, here we describe surgical cases with a disproportionate number of macroadenomas, a sex ratio with a higher proportion of men, and invasive tumors, consistent with other surgical series (30-32). Although the mainstay of prolactinoma management is dopamine agonist therapy, surgery can be considered in some circumstances (4). In this study, the majority of the patients underwent surgery because of either insufficient response to dopamine agonist therapy or drug intolerance, similar to other studies of surgically treated prolactinomas (31, 32).

Significant sex-related differences in prolactinoma have been observed (10). In agreement with the literature, in our cohort prolactinomas in men were diagnosed at older ages than in women (1, 33). Also consistent with the literature, there was a higher proportion of men with radiological invasive tumors than women (34, 35). Furthermore, genomic disruption was statistically higher in males compared to females. This result is consistent with a previous study, which suggests that the more aggressive behavior of tumors in men may be related to an increase in chromosome alterations (10). In line with these findings, the aggressive tumor in a male patient in our study had losses of 2 tumor suppressor genes, PTEN (on chromosome 10q) and RB1 (on chromosome 13q), which could have contributed to the aggressive behavior of this adenoma.

Chromosomal instability may correlate with advanced tumor grade (36-38). However, no association between genome disruption and parameters of aggressiveness, such as Ki-67 >3%, persistent disease postsurgery, and radiological tumor invasion, was observed in our study. Genomic instability has been associated with pituitary adenoma subtype, and prolactinomas are reported to be one of the most disrupted subtypes (37, 39, 40). In our cohort, most of the prolactinomas had some level of genomic instability. The recurrent genomic disruption found in our series included gain of chromosome 9 in 35% and loss of chromosome 11p in 30% of the patients, aligning with findings from previous studies (10, 37, 41). The loss of chromosome 11p, also detected in our patient with the ESR1_Y537S mutation, has been associated previously with aggressive prolactinomas and other malignancies, such as breast, ovary, and thyroid, and may be related to prolactinoma tumor progression (41, 42). Five tumors (13%) in our cohort had quiet genomes; however, the OncoPanel does not cover the whole genome, and we may have missed CNV events in our analyses.

A remarkable finding of our study was the identification of a somatic activating ESR1 mutation in a prolactinoma that progressed to become an aggressive tumor in a postmenopausal woman, when estrogen levels are low and prolactin levels usually decline (15, 16, 43). ERα activation is critical for physiologic lactotroph function, contributing to lactotroph proliferation and prolactin synthesis (5). Estrogen-driven malignancies (eg, cancers of the breast and endometrium) can acquire somatic alterations that constitutively activate estrogen signaling pathways to stimulate proliferation and tumor progression (44). Accordingly, SF3B1 somatic mutation, recently reported in aggressive prolactinomas, contributes to the pathogenesis of prolactinomas through the gain-of-function of estrogen-related receptor gamma involved in estrogen signaling pathways (18, 19). Another potential mechanism in the estrogen-modulated process of prolactinoma tumorigenesis is the increased expression of pituitary of tumor transforming gene (PTTG) (12). It has been demonstrated that estrogen regulates the expression of PTTG, bFGF, and VEGF, which may participate in prolactinoma formation (12). Because estrogen signaling pathways participate in prolactinoma pathogenesis, it seems reasonable to use antiestrogen therapy to treat such adenomas. Antiestrogen treatments, such as tamoxifen and the aromatase inhibitor anastrozole, have been used as add-on treatments to a dopamine agonist in prolactinomas resistant to this treatment, with a good response in reducing prolactin levels (45, 46). The response of prolactinomas to pharmacotherapy targeting estrogen suggests that the tumor cells are sensitive to and partially dependent on estrogen action. Therefore, an increase in estrogen signaling through an activating mutation of the ERα (ESR1_Y537S) may promote greater tumor cell proliferation.

Pathogenic ESR1 mutations in codons 536-538 within the ERα ligand binding domain result in an agonist-like conformational change similar to wild-type ERα bound to estrogen, enabling constitutive receptor activation (47). In breast cancer models, it has been demonstrated that activating ESR1 mutations in the ligand binding domain induce a change in the ERα transcriptional network, which promotes more aggressive behavior (48, 49). Moreover, somatic ESR1 mutations in breast cancer are rarely present in treatment-naïve disease, and only emerge after cycles of antiestrogen hormonal therapy (50, 51). The identification of the ESR1_Y537S gain-of-function mutation in our patient's tumor only after her fourth surgery, when the prolactinoma was more aggressive despite the patient being postmenopausal, suggests that this mutation emerged or was selected for in the context of a low estrogen environment, similar to the selective pressure conferred by antiestrogen therapy in breast cancer (50, 51). We cannot rule out that ESR1_Y537S was present since the beginning of the disease course at low frequency and we were not able to detect it with the methods used.

It has been shown in prolactinomas that low ESR1 expression can correlate with treatment resistance and higher tumor grade, suggesting that the expression of the wild-type ESR1 is an important prognostic marker (10, 52, 53). An ESR1 activating mutation can exacerbate growth-related pathways, leading to lactotroph proliferation, thereby increasing susceptibility to prolactinoma progression. In the postmenopausal state, the ESR1_Y537S activating mutation potentially promotes an advantage for tumor cells to proliferate in a low estrogen milieu, compared to cells expressing wild-type ESR1, resulting in evolution into a new adaptive phenotype with more aggressive behavior. This tumor transformation (or adaptation) into a more aggressive phenotype has been well demonstrated in resistant breast cancer, when classic hormonal therapy induces a low estrogen milieu, providing growth advantage to the cells with ESR1_Y537S or other ligand binding mutations (50).

A striking hallmark of this aggressive prolactinoma harboring the ESR1_Y537S mutation was genomic instability, evidenced by multiple whole-chromosome and arm-level alterations, including loss of chromosome 11p, and a number of CNV events that were more frequent compared to other prolactinomas in our study and in most pituitary tumors with CNV reported in the literature (37, 41). Although data are not extensive on mutational burden in pituitary tumors, the average tumor mutational burden in this study of 200 pituitary tumors was less than 1 mutations/Mb, which is much lower than in this prolactinoma harboring the ESR1 mutation (TMB: 5.3 mutations/Mb) (54). Such genomic instability and increased mutation burden may be a consequence of a series of cumulative events, including increased cell proliferation induced by ESR1_Y537S and DNA damage resulting from treatments (17, 55). Radiotherapy and temozolomide, both used before the identification of ESR1_Y537S, can induce hypermutation, and radiotherapy can also exacerbate chromosomal instability (56, 57). Therefore, we cannot exclude the possibility that radiotherapy or temozolomide treatment may have induced ESR1 mutation.

Measurement of cfDNA is a technique used to monitor disease burden effectively in cancers. However, cfDNA has rarely been detected in benign and slow-growing pituitary adenomas, likely because of low cell proliferation rates in these adenomas (58, 59). The presence of ESR1_Y537S in the recurrent/progressive tumor in our patient permitted us to look for the presence of tumor-specific cfDNA in peripheral blood. The ability to successfully detect ESR1_Y537S was likely because of the prolactinoma's high cell proliferation rate, calling into question this adenoma's benign status. Indeed, the case presented here, with an acquired somatic mutation that conferred aggressive behavior to the tumor, suggests that pituitary adenomas can exist across a continuum that approaches many of the criteria that characterize malignancy: invasion of surrounding structures, incurability and recurrence, rapid growth, high tumor proliferation index with a Ki-67 >10%, and genomic instability (60). Because of the difficulty in detecting tumor-specific circulating cfDNA derived from pituitary adenomas, the role of this technique is not well established in this tumor type. Our findings here demonstrate that this is achievable in cases of aggressive tumors with high cell proliferation and consequent shedding of DNA from the tumor and may be a valuable tool to be added to the management of these challenging tumors.

With the progression of the patient's prolactinoma despite multiple treatment modalities and the identification of ESR1_Y537S as a potential cause of aggressiveness, we pursued a personalized treatment targeting ESR1 activity. The first-generation selective estrogen receptor degrader, fulvestrant, has been shown to have potent activity against lactotroph proliferation in rat cell lines (GH3 and MMQ) expressing wild-type ERα (61). However, fulvestrant has poor oral bioavailability and limited efficacy against ESR1 mutant breast cancer (62). The second-generation selective estrogen receptor degrader, elacestrant, improves progression-free survival in patients with resistant breast cancer and ESR1_Y537S, with a safe profile (62, 63). Three months after the patient received her third cycle of radiotherapy and serum prolactin levels were still >1000 ng/mL, indicating tumor activity, elacestrant was initiated as salvage therapy. Elacestrant, in combination with radiotherapy, significantly reduced serum prolactin levels to <150 ng/dL for the first time in 5 years and remained stable 1 year after initiation of elacestrant. Interestingly, since the tumor had become aggressive, this level of prolactin had not been achieved using a variety of different treatment modalities, including isolated radiotherapy. The control of tumoral mass and decline in prolactin level may be due to a synergistic effect of radiotherapy and elacestrant that should be investigated in future studies. Although the identification of ESR1_Y537S in an aggressive prolactinoma is an isolated finding and further studies screening for ESR1 mutations in aggressive prolactinomas are needed, human and in vitro studies in breast cancer support the hypothesis that this mutation is conferring aggressive tumor behavior in this postmenopausal patient. It is important to emphasize that ESR1 mutations may occur later in the evolution of a prolactinoma in a subset of cells, as described in breast cancer, and some mutation screening techniques (such as Sanger sequencing) may not have sufficient sensitivity to identify clones harboring ESR1 activating mutations (64). Moreover, the sample extracted for molecular analyses may not contain the clones of cells with the mutation. In breast cancer, a temporal selection of the ESR1 mutation has been demonstrated, which was identified in metastatic tissues but not in the matching untreated initial tumors (50).

In summary, we describe the utility of molecular profiling to identify a key somatic mutation in an aggressive prolactinoma. The identification of the ESR1_Y537S somatic activating mutation revealed a genetic basis for this atypical behavior, underscored the importance of estrogen receptor signaling pathways in the tumorigenesis of prolactinomas, and allowed us to use a potential effective targeted therapy.

Abbreviations

BWH

Brigham and Women's Hospital

cfDNA

cell-free DNA

CNV

copy number variation

ddPCR

droplet digital PCR

ER

estrogen receptor

MRI

magnetic resonance imaging

TSS

transsphenoidal surgery

Funding

Innovation Evergreen Award, Brigham and Women’s Hospital to A.P.A.

Disclosures

The authors have nothing to disclose.

Data Availability

The datasets analyzed in the present study are available upon reasonable request to the corresponding author.