Pituitary carcinoma (PC) is defined as a pituitary adenoma that has metastasized outside the sellar region.1 Unlike pituitary adenomas that are common and usually benign,2 PCs are aggressive, rare tumors that account for an estimated 200–300 cases annually in the USA.3

Similar to pituitary adenomas, PCs originate from the various cell types within the anterior pituitary gland and can be functioning (hormone secreting) or non-functioning. PCs more often are lactotroph or corticotroph tumors, producing prolactin (PRL) or adrenocorticotropic hormone (ACTH), respectively. These tumors can be clinically apparent, causing signs and symptoms of hormone excess, or they can be clinically silent, demonstrating expression of PRL or ACTH via immunohistochemistry but not causing Cushing syndrome or clinical features of hyperprolactinemia.

Given the rarity of PC, no well-defined standard therapy exists. Most patients undergo resection if feasible, focal radiotherapy, and medical therapy (for functioning tumors). Temozolomide has emerged as a first-line chemotherapy for PC.4 The combination of temozolomide with capecitabine (CAPTEM) has also been shown to result in high response rates and prolonged survival in PC.5 However, prolonged use of temozolomide can be associated with bone marrow suppression, and recurrences are common in clinical practice. With the advent of immunotherapy as a fourth pillar for the treatment of various solid malignancies, there has been interest in exploring the role of checkpoint inhibitors (CPIs) for treatment of pituitary tumors.

CPIs, including antibodies against programmed death-1 (PD-1), programmed death-ligand 1 (PD-L1) and cytotoxic T-lymphocyte-associated protein 4 (CTLA-4), significantly prolong survival in 30%–40% of patients with various solid cancers.6 Mutation status, PD-L1 expression and the presence of tumor-infiltrating lymphocytes (TILs) are potential biomarkers of response to CPIs.6 Different thresholds have been used to define a ‘hypermutator’ phenotype.7 8 Tumors with a large number of somatic mutations (commonly >10 mutations per Mb) and/or mutations in mismatch repair (MMR) or DNA polymerase genes are referred to as hypermutated tumors.7 8 A hypermutator phenotype is seen across different tumor types with varying frequencies and can either happen de novo or as a result of exposure to alkylating agents.9 Marked response to CPIs have been reported in two patients with corticotroph PC, including one tumor with a hypermutator phenotype.10 11

Here, we report four cases of refractory PC treated with pembrolizumab. Two out of four patients had a radiographic and hormonal response to CPI therapy.

In this open-label, phase II trial, patients with advanced rare cancers whose tumors had progressed within the previous 6 months received pembrolizumab 200 mg intravenously every 21 days.12 Immune-Related Response Evaluation Criteria In Solid Tumors (irRECIST)13 was applied to report responses. The adverse events were graded according to the National Cancer Institute Common Terminology Criteria for Adverse Events V.4.03. PD-L1 staining was performed by Qualtek using Merck 22C3 antibody for PD-L1. Based on the percentage and intensity of membrane staining, an H-score from 0 to 300 was assigned to tumor samples.14 To measure TILs, we performed a morphological assessment of H&E-stained sections to determine the overall average abundance value of TILs within tumor nests for the whole sample, using a scale of 0 (absent) to 3. High TIL score was defined as a TIL density score ≥2. PD-L1 and TIL scores were completed by a board-certified pathologist.

The data cut-off for report of these data was May 4, 2020. It was planned to enroll 15 patients in this cohort; however, due to rarity of PC and resultant slow accrual, we decided to report on the current four patients.

We treated four patients with PC with pembrolizumab (table 1). All patients tolerated the treatment well with mild adverse events (table 2).

CPIs have changed the outcomes of many patients with cancer, and their approvals by the US Food and Drug Administration (FDA) are on the rise, including the first tumor-agnostic approval for pembrolizumab for microsatellite instability-high or mismatch repair deficient solid tumors.15 Despite small numbers of patients, our current report of four cases, which supplements four previously published cases,10 11 provides the basis for further defining the role of CPIs in the treatment of PC.

In this study, we describe the positive radiographic and hormonal responses to pembrolizumab in two patients, both of whom had ACTH-secreting PC refractory to prior lines of systemic therapy. Patient 1’s response has been durable for 42 months. Genomic analysis of tumor tissue obtained from this patient after 16 cycles of temozolomide demonstrated mutations in 76 genes of a 400-gene panel, including MSH2 G862E and MSH6 G1157D, indicative of a hypermutator phenotype. MSH2 and MSH6 are essential mismatch repair (MMR) system components that detect base-base mismatches or insertion/deletion loops and perform error removal and DNA resynthesis.16 MMR-deficient tumors have a large proportion of mutant neoantigens that result in in vivo expansion of neoantigen-specific T-cell clones that are reactive to the mutant neopeptides within the tumor and therefore increase responsiveness to CPI.17 Patient 2’s tumor after exposure to temozolomide was not sampled so its mutation status postchemotherapy is unknown. Interestingly, the two tumors that did not demonstrate a response (patients 3 and 4) did not demonstrate a hypermutator phenotype after exposure to temozolomide.

In patient 1, hypermutation might have been due to exposure to temozolomide; however, the patient’s temozolomide-naïve tissue is not available for NGS testing for comparison. Lin et al reported a case of marked response to nivolumab (anti-PD-1) and ipilimumab (anti-CTLA-4) in a corticotroph PC.11 The authors performed genomic analysis of the tumor before and after exposure to temozolomide and demonstrated the emergence of a hypermutator phenotype after temozolomide exposure. Conversely, Caccese et al reported rapid disease progression of a corticotroph PC with MMR deficiency treated with pembrolizumab.18 This case, however, in contradistinction to the first published case and our two responders, was complicated by ongoing hypercortisolism, which may have attenuated the beneficial effects of CPI therapy. A correlation between MMR deficiency and high mutational burden has been reported in gliomas,19 but this correlation in PC remains to be identified.

Elevated PD-L1 expression is a predictor of response to anti-PD-1/PD-L1 therapies in some tumors, and companion PD-L1 diagnostic expression assays have been approved by the FDA. One study evaluated PD-L1 RNA and protein expression in 48 pituitary tumors and demonstrated increased expression of PD-L1 RNA and protein in functioning (somatotroph and lactotroph) pituitary adenomas compared with clinically non-functioning (null cell and gonadotroph) adenomas.20 Elevated PD-L1 expression was positively correlated with increased TILs and was enriched in primary tumors compared with recurrent tumors. In our case series, all patients had negative PD-L1 staining, including our patient with a durable response, indicating that PD-L1 staining may not be an indicator of response to CPIs in PC. However, this needs to be confirmed in larger studies. Two of four patients with available pretrial TIL data had TIL scores of 2 (patient 1 (responder) and patient 4 (non-responder)). It is not yet clear if PD-L1 expression and/or the presence of TILs are predictors of response to CPIs in pituitary tumors.

The composition of various types of immune cells within the tumor microenvironment and their interaction with tumor cells play an important role in the ability of CPIs to mount immune responses against pituitary tumor cells. One study that analyzed CD11b-expressing cells (expressed on monocytes, neutrophils, natural killer cells, granulocytes and macrophages) in 16 non-functioning pituitary tumors demonstrated that a higher proportion of CD11⁺ cells (>10%) was associated with a more expansile and mitotically active tumor.21 This study also demonstrated that tumors invading the cavernous sinus had a greater presence of immunosuppressive macrophages (M2) than did non-invasive tumors (M2/M1 ratio >1 vs M2/M1 ratio <1). The correlation between increased number of immunosuppressive M2 macrophages and invasive behavior has been described in many cancer types.22 Another study evaluated different immune cell populations in 35 human pituitary tumors and demonstrated increased number of CD68⁺ macrophages in sparsely granulated GH and null cell tumors compared with densely granulated GH and corticotroph tumors.23 In addition, the number of CD68⁺ macrophages has been shown to positively correlate with tumor size and local invasion.24 Evaluating the immune cell composition of tumors is needed for successful clinical development of CPI therapy in pituitary tumors. Our study did not include an analysis of the tumor immune cell composition.

Drugs that target CTLA-4 are more likely than other CPIs to cause hypophysitis and pituitary dysfunction, and their combination with medications that target PD-1 further heightens that risk.25 Isolated ACTH deficiency can also occur after use of single-agent PD-1 inhibitors.26 Importantly, although recovery of anterior pituitary hormone secretion can be seen, central adrenal insufficiency is usually permanent.27 This raises the question as to whether there may be an increased intrinsic susceptibility of corticotroph tumors to immune checkpoint inhibition, as observed in our study (patients 1 and 2) as well as the two other previously published cases.10 11

It has also been hypothesized that pituitary tumors may particularly be sensitive to anti-CTLA-4 antibodies due to the ectopic expression of CTLA-4 on pituitary tumors, which may be in part responsible for higher rates of hypophysitis after anti-CTLA-4 versus anti-PD-1 therapy. Our case of a responder treated with pembrolizumab monotherapy demonstrates that PD-1 blockade alone may be sufficient to induce a remarkable response in a hypermutated PC. It would be interesting to know if patients without a hypermutator phenotype who did not respond to PD-1 blockade alone would have a response to dual anti-CTLA-4 and anti-PD-1 therapy. The ongoing phase II trials of nivolumab and ipilimumab in patients with aggressive pituitary tumors (NCT04042753 and NCT02834013) have the potential to address this question.

Determining the efficacy of CPIs is difficult in rare tumors such as PC due to the challenges involved with conducting large clinical trials for an orphan disease. Here, we report four patients with PC treated with pembrolizumab—in whom two partial responses were seen in patients with ACTH-secreting PC. Further studies are needed to determine if the response to CPIs in patients with PC is unique to corticotroph tumors and to further clarify how the response correlates to the mutational status of the tumor and the tumor microenvironment.