Approach to the Treatment of a Patient with an Aggressive Pituitary Tumor

Abstract

A small subset of pituitary adenomas grows despite maximal treatment with standard therapies; namely, surgery and radiotherapy. These aggressive tumors demonstrate 2 patterns of growth: they may be locally aggressive or metastasize distantly, either hematogenously or through the spinal fluid. Further surgery and radiotherapy may be helpful for palliation of symptoms, but they are rarely definitive in the management of these malignant tumors. The only chemotherapy with established activity in the treatment of pituitary tumors is the alkylating agent temozolomide. At most, 50% of patients exhibit an objective response to temozolomide and the median time to progression is short; thus, there remains a significant unmet need for effective treatments within this patient population. Several targeted agents have reported activity in this tumor type—including small molecule inhibitors, checkpoint inhibitors, and other biologics—but remain investigational at this time.

Case: A 46-year-old woman presented with weight gain, amenorrhea, and myopathy. She was found to have an elevated urine free cortisol value (UFC) of 689 µg/24 h (normal < 45 µg) in the setting of an elevated plasma ACTH of 99 pg/mL. Magnetic resonance imaging (MRI) scans revealed a pituitary macroadenoma that was invasive to the clivus, body of the sphenoid bone, tuberculum sellae, and sphenoid sinus. By transsphenoidal approach, she underwent subtotal resection of the sellar tumor, and pathology confirmed a corticotroph pituitary adenoma with a Ki-67 of 5% and p53 immunoreactivity. Postoperative MRI scan showed residual tumor and UFC was 90 µg/24 h. Radiotherapy was recommended but she declined and was enrolled on a clinical trial investigating the somatostatin pan-receptor ligand, pasireotide; however, after 4 weeks on treatment, because of rapid progression causing respiratory compromise resulting from bilateral dysfunction of cranial nerves IX and XII, she was intubated before placement of a tracheostomy. She subsequently received emergent radiotherapy (RT) with 1000 cGy in 5 fractions through parallel opposed conventional RT. This was followed by a 3-dimensional conformal plan for an additional 3200 cGy in 16 fractions with a final 3-dimensional conformal cone down plan of 800 cGy in 4 fractions. She had a biochemical and radiographic response accompanied by a dramatic clinical improvement resulting from normalization of cortisol levels and near resolution of the cranial neuropathies, which permitted decannulation, as previously reported (patient 3 in the case series) (1). Unfortunately, 2.5 years later her disease progressed, confirmed by an elevated UFC of 754 µg/24 h, plasma ACTH of 88 pg/mL, and recurrent tumor growth (Fig. 1A).

Figure 1.

Imaging of recurrent disease and the corresponding radiotherapy plans (photon and proton). (A) Timeline of the patient’s treatment; the images in panels B-F are from the timepoint in red. *Sample on which whole exome sequencing was performed. (B) An axial T1 postcontrast sequence; the recurrent disease is in the left jugular foramen. The recurrent disease is better visualized with (C) ¹⁸F-FDG and (D) ⁶⁸Ga-DOTATATE PET. Notably, with ⁶⁸Ga-DOTATATE there is less background uptake in the visualized brain. (E) Photon RT plan. (F) Comparative proton RT plan. In the proton RT plan, the brainstem is largely spared reirradiation, which is important given that the brainstem was already treated to the lifetime radiation limit after the first course of RT. ¹⁸F-FDG, 18-fluorodeoxyglucose; ⁶⁸Ga, Gallium-68; PET, positron emission tomography; RT, radiotherapy.

Clinically significant pituitary adenomas, tumors that are either hormonally active or of sufficient size to cause visual impairment, are often addressed with first-line treatments, such as surgery or, in the subset that are prolactinomas, dopamine receptor agonists. Even the small subset that exhibit invasive behavior and demonstrate growth following initial therapy, are rarely life-threatening tumors because they are typically controlled with additional local therapies (specifically, additional surgery and/or radiation). This article focuses on the treatment of the rare subset of patients with pituitary tumors that are not only invasive, but also exhibit aggressive behavior, as characterized by resistance to these standard therapies. The natural history of aggressive pituitary tumors that cannot be controlled with surgery and RT is unrelenting, locally destructive growth, which results in significant morbidity and mortality and, for some, the development of metastases.

In this article, we review the clinical behavior of aggressive pituitary tumors with a focus on management considerations. We discuss the primary treatments that are available to these patients, namely reresection, reirradiation, and temozolomide—an oral cytotoxic chemotherapy for which there are ample retrospective data and treatment guidelines supporting its use. We then discuss off-label investigational approaches of unestablished benefit with an emphasis on opportunities for targeted therapy. Given the rarity and complexity of treatment, patients with this disease process should be referred to a center with multidisciplinary expertise in the management of these tumors when possible.

Clinical Behavior

Aggressive pituitary tumors, which demonstrate resistance to standard therapies (surgery and radiation), fall into 2 general categories. They can remain confined to the skull base and cause significant morbidity because of locally destructive behavior, or they can metastasize, at which time they are called pituitary carcinomas. By virtue of their locally destructive behavior, aggressive pituitary adenomas can cause significant disability. From a mass effect standpoint, these tumors cause symptoms by provoking headache—often migraine or trigeminal autonomic cephalgias (2)—and cranial neuropathies. Through suprasellar extension, these tumors may cause a classic bitemporal hemianopsia or asymmetric vision loss in the bilateral eyes resulting from eccentric compression of the optic chiasm/optic apparatus. Additionally, locally aggressive pituitary tumors can cause visual compromise of an individual eye from unilateral optic nerve compression. Diplopia is another common complication of aggressive growth as cranial nerves III, IV, and VI pass through the cavernous sinus. Besides causing an eye that is depressed and abducted, dysfunction of cranial nerve III also causes pupillary dilation and ptosis, which can progress to complete closure of the affected eyelid and functional loss of vision. V1 and V2 also reside within the cavernous sinus, and thus numbness and paresthesia in the upper two-thirds of the face may develop, though this is typically a later symptom.

Patients with locally aggressive pituitary tumors may succumb to their tumor because of hypothalamic and frontal/temporal lobe destruction, which can cause somnolence, hyperphagia, obstructive hydrocephalus, seizures, and even diabetes insipidus (a complication that is only known to occur as a consequence of mass effect very late in the tumor’s course (3)). As these malignancies progress, when the tumor is functional, hormonal control becomes increasingly difficult, sometimes with life-limiting consequences, as seen in the subset of patients with ACTH- or GH-secreting tumors (4-9). In patients with ACTH-secreting tumors, bilateral adrenalectomy is often required to treat the hypercortisolemia, which can lead to Nelson syndrome (NS), in which patients develop elevated plasma ACTH levels from loss of feedback inhibition and possibly more aggressive tumor growth (10). In a recent series, the 10-year progression-free survival for patients following the diagnosis of NS was 62%. Among the subset of patients who failed initial management of NS (with surgery, RT, pasireotide, or observation), 28% (5/18) demonstrated aggressive tumor behavior and 11% of patients (2/18) developed metastases (11).

Pituitary carcinomas, pituitary tumors that have metastasized, follow 2 patterns of metastatic spread. Pituitary tumor cells can travel through the spinal fluid, resulting in leptomeningeal metastases, which appear as distant deposits that reside on the surface of the brain, brainstem, or spinal cord. Alternatively, the tumor cells can spread through the bloodstream, resulting in distant metastases anywhere in the body. The epidemiology of aggressive pituitary adenomas and pituitary carcinomas has not been well defined. Statistics on the incidence and prevalence of this problem is uncertain as this is a malignancy that is not adequately captured by many registries including the Central Brain Tumor Registry of the United States and the National Cancer Institute’s Surveillance, Epidemiology and End Results program (12). Although it is clear that pituitary carcinomas are rare, they may be more common than appreciated as the problem is likely underdiagnosed because the identification of distant metastases requires a high index of suspicion. Institutional cohorts have reported that only 0.2% to 0.4% of pituitary tumors develop metastases, and a European effort estimates 4 new cases of pituitary carcinoma per 10 000 000 person-years (13-15).

Small case series have suggested that the median survival is less than a year (14), with the caveat that there is a definite subset of patients with pituitary carcinomas that follow a relatively indolent course. Although leptomeningeal dissemination carries a dismal prognosis nearly universally across cancer types, it is possible that the situation is different among pituitary carcinomas. For these tumors, spread through the spinal fluid may have a more favorable prognosis than hematogenous spread (14). The literature includes a report of a patient with leptomeningeal metastases from an ACTH-secreting pituitary carcinoma who had a complete radiographic response to surgery and whole-brain RT and remained without evidence of recurrence at the time of publication 21 years later (16).

Surveillance for metastatic disease may be indicated in any patient with a tumor that has demonstrated growth following treatment with RT. Among functional tumors, an increase in circulating hormone concentrations may also prompt investigation for metastatic disease. The best imaging modality for identifying distant sites of disease has not been defined. 18-fluorodeoxyglucose (¹⁸F-FDG) positron emission tomography (PET) is the most common radiopharmaceutical used in clinical practice for staging patients with malignancies. Uptake of ¹⁸F-FDG is seen in all metabolically active tissue types, such as the brain and liver, but also accumulates at tumor sites and areas of inflammation. Although the sensitivity of ¹⁸F-FDG-PET for malignancies is high, neuroendocrine tumors, including pituitary adenomas, may fail to exhibit hypermetabolism, particularly the more differentiated forms (17). Additionally, leptomeningeal metastases may be difficult to discern because of the high background uptake of neural tissue, and the spatial resolution is low (about 5-8 mm), which limits the detection of micrometastases.

Because somatostatin receptors are overexpressed on pituitary cells and many neoplasms of the pituitary gland (18, 19), it has been suggested that radiolabeled octreotide derivatives such as Gallium-68 (⁶⁸Ga) DOTATATE may be of utility in the management of this tumor, including the identification of distant metastases (20, 21). ⁶⁸Ga-DOTATATE primarily targets SSTR2 (22), which is expressed in the majority of pituitary tumors expressing TSH and GH. For nonfunctioning tumors and tumors secreting ACTH and prolactin, which have lower densities of this receptor (23), tracers with broader affinity such as ⁶⁸Ga-DOTATOC, binding also SSTR5 and SSTR3, may be more sensitive (22). Because of low background avidity in neural tissue, ⁶⁸Ga-DOTATATE can potentially discriminate leptomeningeal deposits but could miss micrometastases because of higher background uptake in the liver and several other organs, low avidity in certain primary tumors, and variable uptake at metastatic sites, presumably from clonal heterogeneity (24).

To rule out leptomeningeal dissemination, MRI scans of the total spine with and without gadolinium and lumbar puncture are often required. Clear indications for a workup for leptomeningeal dissemination include back pain, radicular pain, signs and symptoms of cord compression/cauda equina syndrome, or the identification of leptomeningeal deposits in the field of view of the MRI scan performed for surveillance of the pituitary tumor.

Standard-of-Care Treatments

Historically, the standard treatments for local control are additional surgery and RT. Aggressive pituitary adenomas are typically macroadenomas at the time of diagnosis (14, 25), for which a surgical cure was not achieved for reasons including invasion into unresectable structures such as the cavernous sinus. With further growth, the tumor becomes even less amenable to surgery by transsphenoidal, and even transcranial approaches, and frequently does not result in long-term stabilization. Moreover, each additional surgery places the patient at higher risk of complications: specifically, stroke, diabetes insipidus, and spinal fluid leak. Therefore, the primary indication for repeat surgery is decompression of the optic chiasm as a subtotal resection in other locations is often of limited benefit.

Because locally aggressive pituitary adenomas are typically macroadenomas at the time of initial presentation, these tumors often cannot be treated with stereotactic radiosurgery (SRS) because of size. With SRS, an ablative dose of radiation is administered in 1 to 3 fractions using one of several RT platforms. When a tumor is too large or too close (within 3-5 mm) to a sensitive structure such as the optic chiasm, SRS is not feasible and the total dose is typically fractionated over 25 to 30 fractions over 5 or 6 weeks (26, 27). Fractionation is less destructive to normal tissue and results in high rates of tumor control in the majority of patients with pituitary macroadenomas (in excess of 90% local control at 5 years) (27). The optic chiasm, optic nerve, and other nearby structures have established radiation tolerances. When these tolerances are surpassed, patients are at risk of developing complications including blindness and radiation necrosis of the brain. A contemporary collection of patients who received reirradiation developed radiation induced optic neuropathy in 13.3% and temporal lobe necrosis in 13.3% (28, 29). The cumulative dose of radiation administered to every structure is known before treatment initiation; for that reason, patients can receive individualized counseling on their expected risk for developing these complications based on their anatomy and the details of prior RT. Unfortunately, there are no treatments for radiation-induced optic neuropathy, and when it occurs, permanent visual loss usually occurs over a period of days to weeks (30).

The risks of reirradiation can be partially mitigated by advances in RT, which permit the design of increasingly conformal treatment plans. Conventional external beam radiation treats tumors with photons. Although proton radiation has been in clinical use for decades, it has only recently become widely available. When used therapeutically, protons deliver the majority of their energy at the tissue depth where the particle comes to rest (31); in contrast, with photon RT, a significant proportion of energy is deposited distant from the target, resulting in less conformal plans.

Although proton RT can reduce the dose delivered to nontumor tissue such as the temporal lobes of the brain, the orbit, and brainstem, no sparing of directly adjacent structures can be achieved.

Temozolomide

The only chemotherapy with documented effectiveness in the treatment of pituitary tumors is the oral alkylating chemotherapy agent, temozolomide. Temozolomide is primarily used for the treatment of primary brain tumors but has shown activity in neuroendocrine tumors of the pancreas and lung (small-cell lung cancer) (32, 33). Treatment with temozolomide exerts a cytotoxic effect by causing DNA damage, primarily through the addition of methyl groups to the O6 position of guanine residues, though adducts on other residues and other positions on guanine residues may also contribute (34). When a tumor cell divides, the methyl groups that were added to the 06 position of these guanine residues result in the insertion of a thymine rather than a cytosine (35). In a tumor cell with a functional mismatch repair system, this DNA damage ultimately results in cell apoptosis. A mechanism of resistance to temozolomide is expression of the gene, MGMT, which encodes a protein that catalyzes the removal of these methyl groups. In glioblastoma, epigenetic silencing of MGMT (MGMT promoter methylation) effectively turns off expression of the gene and is an established biomarker of treatment response to temozolomide (36). In pituitary tumors, it has been reported that MGMT protein expression as assessed by immunohistochemistry is a superior biomarker for predicting treatment response, though this has not been prospectively validated and is plagued by the qualitative nature of the assay, which precludes standardization (37).

Though a prospective trial has not been performed investigating the use of temozolomide in pituitary tumors, there are now several large case series supporting a relatively high response rate. As it stands, temozolomide remains an off-label treatment for pituitary tumors, though it is now included in the Endocrine Society and European Society of Endocrinology guidelines (25, 38). Unfortunately, without US Food and Drug Administration (FDA) approval or its addition to National Cancer Center Network Guidelines for this indication (39), insurance denials are not uncommon in the United States at the time of publication.

The largest cohort in the literature is an electronic survey of the European Society of Endocrinology, which captured 166 patients who were treated with this therapy (40). An objective response rate was seen in 37% (specifically, a complete response in 6% and a partial response in 31%). In this publication, a partial response was defined as a 30% regression without specifying measurement parameters. The reproducibility of this result is limited as this survey did not use validated response criteria, such as Response Evaluation Criteria in Solid Tumors (41), which bases response on longest diameter, or Response Assessment in Neuro-oncology Criteria (42), which uses the product of the longest diameter (43). Among patients with an objective response to temozolomide, tumor shrinkage is typically identifiable after 3 or 4 months, and when the tumor is functional, radiographic response typically correlates with a biochemical response (25). Higher response rates closer to 50% have been reported by other case series, suggesting that a 37% objective response rate may be a conservative approximation (37). In this cohort of 166 patients, stable disease was reported in 33%. The clinical significance of stable disease is unclear as the study cohort included in this publication was quite heterogeneous, including a subset of patients who never received radiation and may have had relatively indolent disease.

When given as a single agent, temozolomide is typically dosed based on body surface area at 150 to 200 mg/m²/d for 5 consecutive days every 28 days. A cycle is 28 days because the neutrophil and platelet nadir for temozolomide is between 21 and 28 days. Temozolomide is often well tolerated by patients (44). The common side effects are fatigue, rash, transaminitis, constipation, nausea and vomiting that is typically well controlled by antiemetic premedication, and thrombocytopenia that rarely requires platelet transfusion. Because the mechanism of action of temozolomide is the induction of DNA damage, temozolomide is teratogenic and, for that reason, contraception during treatment and for a period after cessation of the drug is essential. Because there is a small chance that temozolomide could cause primary ovarian insufficiency in a woman or azoospermia in a man, egg and sperm banking is offered. With that said, it is often unnecessary because many individuals go on to have children without using banked gametes after waiting an appropriate duration of time (45). Last, there is a small risk that temozolomide could cause a myelodysplastic syndrome or leukemia that is estimated to occur at a frequency of 10 cases per 1000 patient-years (46). To put this in perspective, the risk of secondary malignancy (especially a malignant glioma or meningioma) with RT is also a concern (47), and RT can cause other complications such as hypopituitarism, optic neuropathy, and, more rarely, vascular complications.

Many controversies exist in terms of the optimal use of temozolomide in this patient population. Two of the major controversies are: (1) the optimal duration of treatment and (2) whether temozolomide should be used in combination with other therapies. In a survey of the European Society of Endocrinology, the median time to first progression after cessation of treatment was only 12 months, though durable responses have been reported (37, 40). When growth occurs during observation following an initial response, the response rate to a second course of temozolomide is thought to be low. In the European Society of Endocrinology survey, of the 18 patients with outcome data that progressed on observation after completing a first course of temozolomide and were rechallenged with this treatment, 2 of 18 had another partial response and 5 of 18 had stabilization for an unreported duration (40). It has been suggested that longer courses of temozolomide could lengthen the time to first progression, though the evidence supporting this possibility is limited (48, 49). In the trial that resulted in the FDA approval of temozolomide for the treatment of glioblastoma, patients received only 6 monthly cycles of temozolomide (44). Although it is common practice for temozolomide to be continued for 12 cycles and beyond for glioblastoma, retrospective studies have not identified a survival benefit with protracted courses of therapy in primary brain tumors (50). Though it is possible that longer courses of temozolomide (continuation of treatment until progression of disease) suppress tumor growth and are beneficial, it is also possible that longer durations of treatment provide no additional benefit, increase toxicity including the risk of bone marrow failure and leukemia (46, 51), and create resistance to future alkylator chemotherapy (52).

The 2 main combination treatments that are described in the literature are (1) concurrent capecitabine and temozolomide and (2) concurrent temozolomide and RT. Like temozolomide, capecitabine is an oral cytotoxic chemotherapy. As a prodrug that is metabolized to the antimetabolite 5-fluorouracil (53), capecitabine interferes with DNA synthesis, replication, and repair and exerts a cytotoxic effect via inclusion into replicating RNA (54). The most common side effects are bone marrow suppression, diarrhea, hand-foot syndrome, nausea, vomiting, and fatigue. Capecitabine has been used in combination with temozolomide in the treatment of nonpituitary, neuroendocrine tumor types. In a small case series of 4 patients with aggressive pituitary tumors and low MGMT expression (suggesting responsiveness to single-agent temozolomide), a high response rate was reported to this capecitabine and temozolomide (CAPTEM) regimen (1). Among these 4 patients, 2 had a complete response, 1 had a partial response, and 1 had stable disease. A prospective clinical trial is assessing the response rate of this combination (NCT03930771) and will hopefully elucidate whether the addition of capecitabine results in higher response rate or merely adds toxicity.

Combining temozolomide with RT is of major interest after the European Society of Endocrinology survey reported an objective response rate of 71% to this combination over an unclear period of observation (40). This finding is consistent with preclinical evidence that indicates that temozolomide is a radiosensitizer (55). Because the average patient with a pituitary adenoma has a high rate of control with radiation alone, it remains unclear which patients might benefit from the addition of temozolomide to RT, even if it were known that these 2 treatments act synergistically. Possible subsets of patients who might derive additional benefit from multimodal treatment are patients with tumors with histological features associated with aggressive disease, tumors that demonstrate rapid growth before RT, and tumors that demonstrated inadequate response to a first course of RT.

Case: At age 50, the patient progressed both clinically (with worsening hormonal hypersecretion and recurrent cranial neuropathies) and also radiographically (Fig. 1A). She underwent a second subtotal transsphenoidal resection to decompress her optic chiasm, which revealed a corticotroph adenoma with 5 mitoses per 10 high-powered fields, a Ki-67 labeling index of 15% to 20%, and MGMT expression in less than 15% of tumors cells by immunohistochemistry. She was referred to radiation oncology; at the time, only photon RT was available. Given the need to overlap RT fields in the brainstem, the risk of a second course of RT was felt to be prohibitive. She was treated with ketoconazole and cabergoline without biochemical control (UFC remained elevated at 1028 µg/24 h with a plasma ACTH of 88 pg/mL). She was then enrolled on a clinical trial investigating CAPTEM, which administered capecitabine 1500 mg/m²/d (maximum daily dose of 2500 mg) divided into 2 doses on days 1 through 14 and temozolomide 150 to 200 mg/m²/d divided into 2 doses on days 10 to 14 of a 28-day cycle. A bilateral adrenalectomy was planned; however, after 3 cycles, her plasma ACTH decreased from 68 to 15 pg/mL and serum cortisol decreased from 20.5 to 1.3 µg/dL, obviating the need for the surgical procedure because treatment had rendered her adrenally insufficient, requiring hydrocortisone replacement. After 5 cycles of CAPTEM, her ACTH and cortisol decreased to 5 pg/mL and 0.5 µg/dL, respectively, the tumor had decreased in size by 50%, and the recurrent cranial neuropathies had largely resolved. After 10 cycles of CAPTEM, her MRI showed complete radiographic response and plasma ACTH was undetectable, as previously reported (1). She was continued on CAPTEM for 8 years on the presumption that cytotoxic chemotherapy would have a suppressive effect on the minimal residual disease. Initially, each cycle of CAPTEM was 28 days. Over time, the cycles were extended to every 42 days, then to every 3 months. During this time, many of her Cushing comorbidities resolved, including diabetes and hypertension, but she continued to require treatment for depression, anxiety, and hyperlipidemia. After 8 years, while being maintained on CAPTEM, the patient progressed biochemically when her plasma ACTH became detectable at 27 pg/mL with a serum cortisol of 2.3 μg/dL. A proopiomelanocortin (POMC) level was measured at this time and returned elevated at 179 fmol/mL (normal < 50 fmol/mL), suggesting the development of an undifferentiated tumor that does not efficiently process the POMC precursor to ACTH. It was at this time that she was referred to Memorial Sloan Kettering Cancer Center for further evaluation. The site of progression was initially occult because the recurrence was distant from the sella (Fig. 1B). ¹⁸F-FDG PET identified the site of progression in the left jugular foramen (Fig. 1C) and CAPTEM was discontinued. Further evaluation was performed with ⁶⁸Ga-DOTATATE PET, which identified no other sites of disease (Fig. 1D). With the availability of proton RT, permitting brainstem sparing, she did well following proton reirradiation except for the development of vocal cord paralysis, left-sided hearing loss, and vestibular dysfunction (left cranial nerve VIII and X dysfunction), both of which were expected complications of treatment. Although the vestibulopathy resolved, the hearing loss persisted. The vocal cord paralysis required no intervention and did not impact her breathing or ability to swallow. In Fig. 1E, we show the photon plan in which one-half of the brainstem would receive > 1750 cGy; the cumulative dose administered with photon RT was expected to place the patient at high risk for radiation necrosis of the brainstem, a catastrophic complication. In contrast, Fig. 1F is the proton plan, where the majority of the brainstem received < 1750 cGy. One year following reirradiation, at age 59, she was found to have liver metastases, which was confirmed by biopsy. Because of insufficient tumor from the liver biopsy, whole exome sequencing of her temozolomide-naïve recurrent sellar tumor and a matched normal was performed. This tumor was found to be hypermutated with an extraordinarily high burden of subclonal mutations in the absence of microsatellite instability/mismatch repair deficiency (Fig. 2A).

Figure 2.

Mutational burden and clonality. (A) Subject of this article has a hypermutated tumor on whole exome sequencing. The majority of the mutations in the sequenced resection have a low cancer cell fraction, which is the fraction of cancer cells predicted to harbor the mutation. (B) Mutational burden of a different, previously reported patient with a temozolomide (TMZ)-naïve, locally recurrent tumor with a more typical mutation burden on whole-exome sequencing, and the mutational burden of the TMZ-exposed metastasis with alkylator-induced somatic hypermutation that subsequently developed in this same patient (92).

Second-line medical treatments

All second-line medical therapies following temozolomide should be considered investigational and of unproven benefit because the data are largely limited to case reports and small case series. Aside from temozolomide, the limited experience using cytotoxic chemotherapy in the management of aggressive pituitary tumors has been disappointing, with only isolated objective responses reported to combinations such as cisplatin/etoposide (56), lomustine/5FU (57), and 5FU/cyclophosphamide/doxorubicin (Table 1) (58).

Table 1.

Additional Cytotoxic Chemotherapy Regimens Reported in the Literature

	No. of Cases	Partial Response	Stable	Progression	Reference
Carboplatin and etoposide	5			5	(40, 92, 102)
Cisplatin and etoposide	7	2	2 (1 was stable for 1 y; the other was stable for > 1 y)	3	(40, 56, 103-106)
Cisplatin, etoposide, and tamoxifen	1		No objective response in 1/1		(107)
Carboplatin	3		No objective response in 3/3		(57)
Etoposide	2			2	(40)
Etoposide and cyclophosphamide	1			1	(40)
Cisplatin and doxorubicin	1		1 (stable over an unknown duration)		(40)
Capecitabine	1			1	(40)
Cisplatin and 5-FU	1			1	(40)
Doxorubicin and 5-FU	1			1	(40)
Lomustine, procarbazine, etoposide	1		1 (stable > 1 y)		(108)
Lomustine, procarbazine, vincristine, and tamoxifen	1	1			(103)
Lomustine and 5-FU	9	1	No objective response in 8/9		(57)
Oxaliplatin and 5-FU	1			1	(40)
5-FU, cyclophosphamide, and doxorubicin	2	1	1 (stable over a 2- to 3-y period)		(58, 109)
Cyclophosphamide, vincristine, and dacarbazine	2		2 (stable over an unknown duration)		(110, 111)
Cyclophosphamide, doxorubicin, and dacarbazine	1		1 (stable over an unknown duration, at least 2 mo)		(112)
Methotrexate and 5-FU	1		1 (stable over a 2-y period)		(113)
Cisplatin, vinblastine, and bleomycin	1			1	(114)

Abbreviation: 5-FU, fluorouracil.

Because molecular profiling has become available, biomarker-driven targeted therapies are increasingly being considered. We will review 5 targeted therapies with reported single-agent activity: lapatinib, everolimus, vascular endothelial growth factor (VEGF) inhibitors, peptide receptor radionuclide therapy (PRRT), and checkpoint inhibitors. An additional targeted therapy, the CDK 4/6 inhibitor, palbociclib was reported to induce regression of a pituitary macroadenoma that had not demonstrated aggressive growth and will not be discussed further but requires further study (59).

Lapatinib

Lapatinib, a tyrosine kinase inhibitor which inhibits epidermal growth factor receptor (EGFR) and ErbB2 (Her2) signal transduction, is a targeted therapy of interest in corticotroph and lactotroph tumors for different molecular reasons. Recurrent ubiquitin specific peptidase 8 (USP8) mutations have been identified in 17% to 62% of corticotroph tumors, with most studies reporting a 31% to 36% prevalence (60, 61). These gain-of-function mutations result in decreased degradation and increased recycling of EGFR to the cell surface because of enhanced USP8 deubiquitinase activity, leading to an increase in POMC transcription and ACTH secretion. It follows that targeting EGFR in USP8 mutated corticotroph tumors may be an effective tumor-directed therapy, though this remains speculative.

Despite the finding that EGFR expression in pituitary tumors may be associated with aggressive behavior (62), some studies indicate that USP8 mutations are found in corticotroph tumors that are smaller, less invasive, and follow a more benign course (61, 63, 64). For this reason, EGFR-targeted therapy may be less relevant for aggressive tumors, though this remains uncertain. There is only a single known example of an aggressive corticotroph tumor that was treated with lapatinib; in this negative case report, USP8 mutational status was not reported (65). A trial is currently investigating the use of gefitinib, another EGFR inhibitor, in a small cohort of ACTH-secreting tumors (NCT02484755).

Recurrent USP8 mutations do not occur in prolactinomas. However, it has been demonstrated that EGFR and ErbB2 (HER2), the targets of lapatinib, are highly expressed in these tumors (66). Although the preclinical evidence suggested promise (67), available clinical data indicate more limited activity. A prospective clinical trial enrolled 6 patients with aggressive prolactinomas to treatment with lapatinib and reported a minor response (22% tumor volume reduction) in 1 patient, stable disease in 3, and progressive disease in 2 patients at 6 months. Notably, a reduction in prolactin level was observed in 3 of 6 subjects (68). Lapatinib is generally well tolerated, with the most common grade 3 toxicities being diarrhea, nausea and vomiting, transaminitis, neutropenia, and rash (69, 70).

Case: Interrogation of the USP8 gene on whole-exome sequencing of the tumor and a matched normal revealed 2 subclonal mutations: a silent mutation in exon 14 and a S755T missense mutation in exon 15, which is outside the 14-3-3 binding motif. Both USP8 mutations in this hypermutated tumor are believed to be clinically insignificant because of their presence in a minority of tumor cells and significant doubt over their functional significance.

Everolimus

The mammalian target of rapamycin (mTOR) pathway, which is involved in metabolism, cell growth, and proliferation, is another potential treatment target for aggressive pituitary tumors. This pathway may be upregulated in pituitary tumors compared with normal pituitary tissue (71), and may be associated with cavernous sinus invasion (72). In vitro studies have suggested that everolimus, an mTOR inhibitor that is approved to treat neuroendocrine tumors, could have antiproliferative and proapoptotic activity in pituitary tumors (73), and that this pathway may be upregulated in prolactinomas in particular (74).

The clinical data supporting its use are sparse. To date, everolimus has been used in at least 7 aggressive pituitary tumor cases: 1 prolactinoma, 3 ACTH adenomas/carcinomas, and 3 pituitary tumors of unknown histology (13, 40, 74-76). Response has been mixed, with a partial response seen in 1 patient with a prolactinoma that was receiving concomitant therapy with cabergoline, and stable disease of less than a year in a patient with an ACTH-secreting carcinoma that received concomitant treatment with palliative RT (75). The remaining 5 tumors did not respond radiographically (13, 40, 76). Adverse effects can include myelosuppression, hyperglycemia, hyperlipidemia, fatigue, rash, stomatitis, and diarrhea (77).

Case: The vast majority of the genetic alterations in the sequenced resection are subclonal and unlikely to be genetic drivers. Although whole exome sequencing does not provide compelling evidence in support of mTOR pathway activation, pathway activation may have developed as the tumor underwent malignant transformation. Additionally, mTOR pathway activation can occur via other mechanisms, including epigenetic changes, which would not be identified by this analysis and for that reason, everolimus remains a treatment consideration.

VEGF inhibition

VEGF plays a role in tumor formation by stimulating angiogenesis and regulating the tumor microenvironment. Some data suggest that high VEGF expression may be associated with pituitary tumor invasiveness or subtype (78), and that targeting angiogenesis may be of therapeutic value in pituitary tumors specifically (79, 80). Intriguingly, the prolactinomas that develop in a mouse model of multiple endocrine neoplasia, type 1, in which a copy of the MEN1 gene has been knocked out, respond to treatment with an anti-VEGF-A monoclonal antibody, as shown by a decreased mean tumor doubling-free survival and lowering of serum prolactin compared to mice treated with control (81). Concordant results were seen in a Hmga2/T mouse model that develops hemorrhagic prolactin-secreting adenomas when treated with a VEGF inhibitor (79).

Bevacizumab is a monoclonal antibody that binds VEGF and prevents it from interacting with its receptor on endothelial cells and is the primary anti-VEGF therapy that has been tried in pituitary tumors. The common side effects of bevacizumab are fatigue (which is typically mild), hoarseness, and hypertension. Serious, but uncommon side effects of this drug include clotting (either venous or arterial), hemorrhage, wound healing problems, gastrointestinal perforation, reversible posterior leukoencephalopathy syndrome, and proteinuria, which is a cumulative dose-dependent complication (82). The VEGF receptor can also be targeted with tyrosine kinase inhibitors, of which some are more selective than others; they include axitinib, apatinib, and sunitinib. Because bevacizumab can be safely combined with RT and cytotoxic chemotherapy, it is often used in combination with other agents, which complicates interpretation of these data. Notably, there are case reports suggesting single-agent activity of bevacizumab in this tumor type. In the European Society of Endocrinology survey, of the 3 patients who received bevacizumab monotherapy, 1 had a partial response (40). Although there are certainly negative case reports (83), prolonged stabilization of a pituitary carcinoma with bevacizumab alone has been described (84). Taken together, clinical data on the efficacy of anti-VEGF therapy for aggressive pituitary tumors are limited and need further exploration.

PRRT

Somatostatin receptor expression has been demonstrated on pituitary tumors in a subtype-dependent manner (85). Somatostatin receptors can be targeted with several different somatostatin receptor ligands, of which some can be radiolabeled. Based on this, treatment with a somatostatin receptor ligand that delivers a cytotoxic dose of radiation (PRRT) is the next logical step after the visualization with radiolabeled diagnostic tracers, such as ⁶⁸Ga-DOTATATE, but only when the tracer establishes that the tumor adequately overexpresses somatostatin receptors (eg, uptake equal or higher than normal liver) (86).

PRRT is performed with octreotide derivatives radiolabeled with therapeutic isotopes, such as Lutetium-177 (¹⁷⁷Lu) or Yttrium-90 (87). ¹⁷⁷Lu-DOTATATE has been approved for use by the FDA for advanced gastro-entero-pancreatic neuroendocrine tumors, based on the Neuroendocrine Tumors Therapy-1 trial and prior European experiences (88). The treatment was well tolerated, with the most common side effects being nausea, vomiting, abdominal pain, diarrhea, fatigue, and bone marrow suppression. Data supporting the use of PRRT in the treatment of pituitary adenomas are still limited. A recent review captured 20 patients with aggressive pituitary tumors who were treated with PRRT using several different radiolabeled octreotide derivatives (89, 90). Of the 20 described cases, 3 showed partial response and 3 had stable disease, of which all 6 cases were temozolomide-naive. Notably, all but 1 of the patients without prior temozolomide exposure showed either stable disease or a treatment response. Systematic clinical trial data are needed to determine response rate, which would be expected to track with density of SSTR expression.

Case: Because the patient’s liver metastases were undetectable on ⁶⁸Ga-DOTATATE PET due to low receptor density, ¹⁷⁷Lu-DOTATATE was not felt to be a treatment consideration.

Checkpoint inhibitors

Mismatch repair deficiency (MMRd) is a known cause of hypermutation and can be somatic or inherited. Notably, there is a report that suggests an association between Lynch syndrome (a germline defect in 1 copy of the MSH2, MSH6, MLH1, or PMS2 gene) and the development of aggressive pituitary tumors (91). Hypermutation occurred in this patient’s temozolomide-naïve tumor in the absence of MMRd. This maintenance of mismatch repair was predicted because temozolomide is only cytotoxic in tumor cells capable of mismatch repair and this tumor demonstrated exquisite sensitivity to this drug. Because temozolomide is an alkylator, it can be the cause of hypermutation. When a tumor cell acquires MMRd under the selective pressure of temozolomide, temozolomide induces mutations without causing the tumor cells to undergo apoptosis. The end result is the development of a tumor with a heavy mutational burden, including potentially oncogenic mutations that could theoretically cause malignant transformation (52). Recently, this phenomenon, alkylator-induced somatic hypermutation, was reported in a pituitary tumor during transition from a locally aggressive tumor to a pituitary carcinoma (Fig. 2B) (92, 93).

It is known from the melanoma and non-small-cell lung cancer literature that mutational burden is associated with treatment response to checkpoint inhibitors, specifically responses to antibodies against CTLA-4 (such as ipilimumab) and PD1 (such as nivolumab and pembrolizumab) (94, 95). Antibodies that block CTLA-4 on T cells prevent this molecule from disengaging the immune system, resulting in T-cell activation, and blocking PD1 releases a brake that prevents the immune system from attacking an individual’s normal cells. The correlation seen between clinical benefit from checkpoint inhibitors and tumor mutational burden is thought to be due to the creation of neoantigens, which are presented on major histocompatibility complex class I proteins on tumor cells, which can trigger a T-cell-dependent immune response (94). It has been suggested that mutations with higher clonality are better at eliciting an effective treatment response (96). One checkpoint inhibitor, pembrolizumab, has achieved FDA approval for treatment of any tumor, including pituitary tumors, with microsatellite instability because microsatellite instability is a marker of a high somatic mutational burden. This unprecedented tumor agnostic FDA indication is based on data from 149 patients with microsatellite-high or mismatch repair deficient cancers collected over 5 uncontrolled, multicenter clinical trials.

In the report describing alkylator-induced somatic hypermutation in a pituitary tumor, a dramatic response to the checkpoint inhibitors, ipilimumab and nivolumab, was reported (92). After an initial biochemical response with plasma ACTH decreasing from 45 550 to 59 pg/mL, the dominant liver metastasis regressed by 92% and the intracranial component regressed by 59% on volumetric assessment using manual segmentation over 6 months. It is unclear whether this response was mediated by hypermutation or whether pituitary tissue is uniquely sensitive to immunotherapy. Evidence in support of the latter assertion include preclinical evidence supporting activity of anti-PD-L1 antibodies against pituitary tumors in a mouse model of Cushing disease (97) and human data, in which high rates of hypophysitis occur in patients treated with checkpoint inhibitors (97, 98). Hypophysitis is reported to occur in patients treated with ipilimumab (anti-CTLA-4) alone at a rate of 4% to 15% (98-100); it occurs less commonly in patients treated with nivolumab (anti-PD1) and more commonly when ipilimumab is combined with nivolumab (101). A clinical trial (NCT04042753) is further investigating the role of this combination, nivolumab plus ipilimumab, in the treatment of patients with aggressive pituitary tumors. Although combination treatment with dual checkpoint blockade may be advantageous, it is known that it has greater toxicity than monotherapy. In 1 study, severe/life-threatening (grade 3 or 4) immunologic adverse events occurred at a rate of 59% with combination nivolumab and ipilimumab, in contrast to a rate of 21% in patients who received nivolumab alone (101). The most common adverse effects of any severity are rash, pruritis, fatigue, and diarrhea.

Case: Although the low cancer cell fraction of the vast majority of the mutations in the tumor may be a barrier for triggering an effective immune response, the hypermutation of this patient’s tumor, which may have been augmented by 8 years of exposure to temozolomide, provides a compelling argument for considering a checkpoint inhibitor.

Conclusion

Aggressive pituitary tumors are malignant tumors that are understudied for a variety of reasons. These tumors are not appropriately tracked and so the scope of the problem is largely unknown. Because these tumors are relatively uncommon, even major pituitary tumor centers are unable to support a clinical trial because of the rarity of these patients. Because of this limitation, the only agent with documented effectiveness, temozolomide, has only retrospective studies supporting its use. In the current case, tumor control was achieved after an initial response while the patient remained on maintenance temozolomide plus capecitabine over an 8-year period. The optimal duration of temozolomide-based therapy remains to be determined. To facilitate prospective trials into this uncommon tumor, a clinical trial consortium is needed to facilitate the development of multicenter trials. Patients who are unable to receive temozolomide or do not respond or progress on this agent should be preferentially enrolled onto clinical trials as additional treatments are needed.

Glossary

Abbreviations

¹⁸F-FDG

18-fluorodeoxyglucose

⁶⁸Ga

Gallium-68

¹⁷⁷Lu

Lutetium-177

CAPTEM

capecitabine and temozolomide

EGFR

epidermal growth factor receptor

FDA

Food and Drug Administration

MMRd

mismatch repair deficiency

MRI

magnetic resonance imaging

mTOR

mammalian target of rapamycin

NS

Nelson syndrome

PET

positron emission tomography

PRRT

peptide receptor radionuclide therapy

RT

radiotherapy

SRS

stereotactic radiosurgery

UFC

urine free cortisol

VEGF

vascular endothelial growth factor

Additional Information

Disclosure Summary: A.L.L. has received research funding from Bristol-Myers Squibb for a phase II trial investigating nivolumab plus ipilimumab in the treatment of aggressive pituitary tumors and from NantOmics to investigate MGMT quantification via mass spectrometry and its relationship with treatment response to temozolomide. T.J.Y. receives research funding from Kazia and AstraZeneca and consults for Debiopharm and Galera. L.B. receives research funding from Advanced Accelerator Applications/Novartis; has unpaid consultancies with Advanced Accelerator Applications/Novartis, Ipsen, Curium, and Clovis Oncology; and is on the speaker’s bureau for Advanced Accelerator Applications/Novartis, ITM, and Ipsen. V.T. is a cofounder and scientific advisor for BlueRock Therapeutics Inc. E.B.G. is an investigator for research grants to MSKCC from Ionis, Novartis, Corcept, Strongbridge Biopharma, and Bristol-Myers Squibb. The remaining authors have no disclosures to declare.

Data Availability

The datasets generated during and/or analyzed during the current study are not publicly available but are available from the corresponding author on reasonable request.