The Role of Programmed Cell Death Ligand 1 Expression in Pituitary Tumours: Lessons from the Current Literature

Pituitary tumours represent 15% of all intracranial tumours and usually have an indolent behaviour [1, 2]. However, high disease burden may deeply jeopardize patients’ health and quality of life, due to excessive or decreased secretion of pituitary hormones, as well as due to local tumour mass effects [1‒3]. Aggressive pituitary tumours correspond to about 10% of all cases, and have been defined as radiologically invasive tumours with unusually rapid tumour growth rate, or clinically relevant tumour growth despite optimal standard treatment [4‒6]. However, the occurrence of central nervous system or distant metastatic deposits, defining a pituitary carcinoma, is rare comprising 0.1 up to 0.25% of all pituitary tumours [4, 7].

The expression and modulation of immune checkpoint pathways in tumour cells, such as programmed cell death protein 1 (PD-1), programmed cell death ligand 1 (PD-L1) or cytotoxic T-lymphocyte-associated protein 4 (CTLA-4), promotes immune tolerance allowing tumour cells to escape anti-tumour immune responses [8‒11]. PD-L1 expression in tumour cells has been shown to predict the clinical outcomes and the response to immune checkpoint inhibitors (ICIs) in different cancers, such as non-small cell lung carcinoma, oesophageal squamous cell carcinoma, gastric adenocarcinoma, melanoma, and breast cancer [12‒17]. ICI efficacy relies on the immune recognition and T-cell mediated destruction of malignant cells [12, 13, 18, 19]. Despite their high anti-tumour efficacy, ICI carries the risk of adverse effects: fever, rash, myalgia, pneumonitis, diarrhoea as well as endocrinopathies, including thyroiditis, adrenal insufficiency, diabetes, and hypophysitis [8]. ICI-induced hypophysitis has been described in up to 17% of patients taking ICI [20, 21], suggesting that the pituitary gland is a potential immunogenic target, which paved the way for further investigation of ICI in patients with aggressive or metastatic pituitary tumours [8, 22].

Patients with aggressive or metastatic pituitary tumours, progressive or refractory to temozolomide, have currently few alternative therapeutic options [4, 23]. Novel treatments for advanced pituitary tumours are being investigated and/or experimentally used, including bevacizumab, mTOR inhibitors, tyrosine kinase inhibitors, peptide receptor radionuclide therapy, and also ICI [24‒27], particularly since the first successful case of a corticotroph carcinoma treated with ipilimumab and nivolumab [28]. However, knowledge and robust data on effectiveness, safety and predictive biomarkers of response to ICI are still lacking [25, 29].

Recent studies have been investigating the role and usefulness of PD-L1 expression as a prognostic biomarker, as well as a potential predictor of responsiveness to ICI [24, 25, 27, 28, 30‒37]. We review the current literature on PD-L1 expression in pituitary tumours, focusing on its biological significance and prognostic usefulness.

A PubMed search was conducted using the following terms (isolated and in combination): “pituitary adenoma”, “pituitary tumour”, “pituitary carcinoma”, “pituitary neuroendocrine tumour”, “PitNET”, “tumour microenvironment”, “PD-L1”, “PDL1”, and “immune checkpoints”. All indexed manuscripts published in English until January 2024 were reviewed and evaluated, as well as any relevant articles from the references list of each publication. Initially, we identified a total of 16 studies [10, 11, 18, 22, 24, 38‒48]. As shown in online supplementary Table 1 (for all online suppl. material, see https://doi.org/10.1159/000539345), from these 16 studies comprising 1,371 patients, PD-L1 expression is higher in functioning pituitary tumours in 9 studies [11, 18, 39‒41, 45, 46, 48, 49], especially in somatotroph, lactotroph, and pituitary transcription factor 1 (PIT-1) tumours. After a thorough screening, 10 studies were excluded from our analysis on the basis of: (i) did not report the percentage of PD-L1 positivity status as assessed by immunohistochemistry (IHC); (ii) did not use IHC positivity cutoffs equal or above 1–5% of pituitary tumour cells immunoreactive for PD-L1; and/or (iii) did not specify the PD-L1 positivity status according to the pituitary tumour functional status. Six studies comprising a total of 704 pituitary tumours with available immunohistochemical data on PD-L1 expression status were then considered for our pooled analysis. These studies, as well as the criteria for PD-L1 positivity by IHC and the used anti-PD-L1 antibodies, are summarised in the Table 1.

Functioning pituitary tumours were regarded as tumours that produced hormones to a clinically significant syndrome in line with the data reported in each study. Pituitary tumours were considered invasive, proliferative, and/or recurrent as also established in each study. Proliferation criteria relied on the proliferation index Ki-67 ≥3% [10, 38], and in one study [22], proliferation was considered based on the presence of at least 2 of the following 3 criteria: >2 mitoses per 10 high-power field, Ki-67 ≥3%, p53 positive. The defining criteria for an invasive pituitary tumour also varied across the studies, consisting of cavernous sinus invasion only, or cavernous and/or sphenoid sinus invasion (Table 2). Recurrence, evaluated in 3 studies [22, 38, 39], and intended as high serum hormone levels in functioning pituitary tumours after treatment, or reappearance and regrowth of tumour mass in patients with nonfunctioning pituitary tumours, was considered as established by each study.

Although PD-L1 has been widely studied in cancer, including in pituitary tumours, little consensus exists regarding the best way to determine a significant PD-L1 expression and which methods or cutoffs should be adopted to define an expression of PD-L1 relevant from a biological standpoint [10, 11, 18, 22, 24, 38‒40, 42‒44]. Different assessment criteria and methods regarding PD-L1 expression in pituitary tumours have been reported. Previous studies investigated PD-L1 expression by IHC in tumour cells [10, 11, 18, 22, 24, 38‒40, 45], while others were based on mRNA [11] or RNA sequencing-related techniques [41‒43].

Different IHC techniques including different antibodies, the use of image software interpretation, multiplexed immunochemistry, and different or not specified subcellular topography of the staining (cytoplasmatic and/or membranous staining), as well as different cutoff values to consider PD-L1-positive expression, render the comparison between studies rather challenging. In order to provide a comparative analysis concerning the PD-L1 expression in pituitary tumours across the studies, we defined PD-L1 positivity when ≥1–5% of pituitary tumour cells were immunoreactive for PD-L1, as most studies adopted such cutoff (≥1–5%) for considering the PD-L1 expression in pituitary tumours as positive [10, 18, 22, 25, 38, 39]. Similar approach and cutoff values have been used in other cancers, including in anaplastic thyroid cancer [50, 51], breast cancer [52], prostate cancer [53], among others [54, 55]. Studies assessing the expression of PD-L1 in pituitary tumours following a qualitative manner or a different approach than the estimation of the percentage of tumour cells with PD-L1-positive staining were excluded from our pooled analysis.

The statistical analysis was performed using the SPSS software (version 26.0, IBM, USA). Data are presented as absolute number and percentages for categorical variables. The comparative analysis concerning the PD-L1 status and pituitary tumour functioning status, as well as the clinico-pathological features (invasion, proliferation, and recurrence) were carried out using χ² test, and Fisher’s exact probability test, as appropriate. p values <0.05 were considered statistically significant.

In our analysis, we included 6 original studies where PD-L1 expression was assessed by IHC and defined PD-L1 positivity when ≥1–5% of pituitary tumour cells were immunoreactive for PD-L1 (Table 1), encompassing a total of 704 pituitary tumour cases: 384 (54.5%) non-functioning tumours and 320 (43.5%) functioning tumours (Table 2). The tumour subtype was specified in 264 of these 320 functioning pituitary tumours, corresponding to 60 somatotroph, 69 lactotroph, 68 corticotroph, 6 thyrotroph, and 61 plurihormonal tumours (Table 2).

Overall, 248 out of 704 pituitary tumours (35.2%) had a positive PD-L1 expression. PD-L1 positivity varied remarkably across the 6 included studies, from 18% up to 47.2% (Table 1). The percentage of PD-L1 immunoreactive cells versus tumour cells was variable across the 3 studies reporting this data [18, 22, 25], in most cases lower than 50%. In the series of Turchini et al. [18], 40 out of 125 (32%) PD-L1-positive tumours had a percentage of PD-L1 immunoreactive cells/total tumour cells >50%, while in the Suteau and coworkers [22] study, this percentage was only 12%, and none of the four PD-L1-positive tumours reported by Ilie et al. [25] had a percentage of PD-L1 immunoreactive cells/total tumour cells >50% (Table 1).

The proportion of PD-L1-positive tumours was significantly lower in nonfunctioning pituitary tumours (26%) than in the functioning counterparts (46.3%). Among all PD-L1 positive tumours, 59.7% (n = 148) corresponded to functioning tumours. Such association between PD-L1 positivity and functioning pituitary tumours was significant in 2 individual studies [10, 39], with one study [10] describing a marked difference in the PD-L1 positivity rates between non-functioning versus functioning tumours (18.9% vs. 58.8%, p < 0.001) (Table 2).

Data concerning PD-L1 positivity across different functioning tumour subtypes were provided in 4 studies [10, 18, 22, 25], comprising a total of 610 patients (Table 2). Somatotroph and lactotroph tumours showed higher rates of PD-L1-positive tumours (respectively, 56.7% and 53.6%) than thyrotroph (33.3%) or corticotroph tumours (20.6%). Of the 61 patients with plurihormonal tumours where PD-L1 expression data were available [10, 18, 22], 46 cases (75.4%) had a positive expression of PD-L1 (Table 2). Some studies included in our analysis reported data on invasion (n = 2) or proliferation (n = 3). Overall, 39.8% (66/166) and 27.2% (97/357) of pituitary tumours were, respectively, classified as invasive or proliferative, as defined by the criteria followed in each study. Three studies, comprising 233 patients, provided information about recurrence [22, 38, 39]: 23 patients (9.9%) were identified with recurrent pituitary tumours (Table 1). Overall, the proportion of PD-L1 positivity did not differ between invasive (24.2%) and noninvasive (17.2%) pituitary tumours (Table 3). In contrast, PD-L1 positivity rates were significantly higher in proliferative tumours than in non-proliferative counterparts (43.3% vs. 23.9%; p < 0.001) (Table 3). The overall rates of PD-L1 positivity among recurrent or primary pituitary tumours were 13% and 21.9%, respectively (Table 3). Suteau et al. [22] also investigated the association between PD-L1 expression and clinical outcomes, reporting no correlation between expression of PD-L1 and postoperative evolution (complete remission, remnant stability, disease controlled by medical therapy, revision neurosurgery or radiotherapy), neither with progression-free survival.

The cohort of Ilie et al. [25] comprised 15 aggressive pituitary tumour cases treated with ICI, of which 6 were carcinomas (4 corticotroph and 2 lactotroph) and 9 were aggressive pituitary tumours (5 corticotroph and 4 lactotroph). There were 4 PD-L1 positive cases: 3 lactotroph and 1 corticotroph tumour. In the subgroup of lactotroph tumours, PD-L1 expression data were available in 5 out of 6 cases, being positive in 3 of them, including the case that had a partial response to ICI. This responsive case had a lower percentage of PD-L1 immunoreactive cells in the primary pituitary site (30%) than in the lung metastasis (40%). One lactotroph tumour displaying 40% PD-L1 positive tumour cells had stable disease on ICI treatment, while the lactotroph tumour that progressed on ICI had 10% PD-L1 positive tumour cells. In corticotroph tumours, PD-L1 expression data were available for 8 out of 9 cases and were negative in seven of them, including the 3 tumours that showed partial response and the tumour that showed stable disease to ICI therapy, while it was positive (5% positive tumour cells) in the corticotroph tumour that progressed on ICI [25].

PD-L1 is a transmembrane receptor that interacts with PD-1 and B7-1-causing immune system suppression. PD-1 and PD-L1 interaction blocks the activation of T lymphocytes and inhibits immune responses. Within the tumour, PD-L1 expression may be observed in infiltrating immune cells and neoplastic cells. PD-L1 expression on tumour cells was shown to downregulate the immune system and lead to immune evasion in many cancers [55‒58]. Recent studies assessed the expression of PD-L1 in pituitary tumours, from the perspective of its potential as a biomarker for predicting the biological behaviour, outcomes, and prognosis [10, 11, 18, 22, 38‒44, 46, 47]. PD-L1 was also suggested to play a key role in the pathogenesis of pituitary tumours through the cyclooxygenase 2 (Cox-2)/PD-L1/arginase 1 signalling pathway [44]. The main findings from the studies investigating PD-L1 in pituitary tumours are summarised in online supplementary Table 1.

Putting together the data from the 6 original IHC-based studies included in our pooled analysis, few considerations can be made: (i) a significant amount of pituitary tumours express PD-L1, although the PD-L1 expression pattern may be heterogeneous and variable, which may well be due to differences in methodologies, antibodies, or criteria to define PD-L1 immunostaining positivity across the studies, as well as other factors such as to the heterogeneity of the biology and the microenvironment across different pituitary tumours; (ii) nonfunctioning tumours seem to express lower levels of PD-L1 than functioning pituitary tumours; (iii) PD-L1 expression appears to be higher in proliferative pituitary tumours; (iv) PD-L1 expression seems not to correlate with pituitary tumour invasion; (v) PD-L1 appears not to be associated with recurrence, with recurrent tumours tending to have lower rates of PD-L1 positivity than primary pituitary tumours.

PD-L1 expression levels as assessed by IHC are highly variable across the studies. The 6 included studies reported PD-L1 positivity rates in less than half of cases, ranging from 18% up to 47.2%. However, other IHC-based studies that followed different methods, and thus not included in our pooled analysis, reported higher rates of PD-L1 positivity. Mei et al. [11] reported a positive PD-L1 expression in all 48 studied pituitary tumours, while Zhao and coworkers [44] described a PD-L1 positivity rate of 78% in a cohort of 55 pituitary tumours patients. In fact, the data from such studies are difficult to compare with each other, bearing in mind the heterogeneous and variable IHC methodologies used to assess PD-L1 expression. Different anti-PD-L1 antibodies have been used, and the immunodetection concordance between assays may be very poor [59]. Different antibody clones were used, including research-only-marked clones, where optimization and validation may be more challenging than clones previously validated in clinical trials of cancer immunotherapy [22]. Different IHC-related techniques may lead to distinct results, including the employment of automated versus manual staining techniques (which may vary according to the commercial brands), as well as the preparation, type, and size of samples used for IHC studies. PD-L1 immunostaining is generally valid only when there are more than 100 tumour cells in a sample, thus the results may vary significantly in PD-L1 staining across a full slide versus on tissue microarrays [18]. There is also a lack of consensus of PD-L1 assessment criteria, including for the best method (quantitative vs. qualitative), the optimal cutoff for considering PD-L1-positive expression, and also regarding the immunostaining localization to be considered (plasma membrane vs. cytoplasmic/nuclear staining). Currently, as for other membrane receptors, such as somatostatin receptors, only the membranous staining is specific [18, 22]. Currently, it is not clear if the PD-L1 positivity is more relevant than the intensity of PD-L1 expression, an important aspect requiring further investigation in pituitary tumours, given that in other cancers, it was shown that the PD-L1 staining intensity may be more determinant than the positivity itself for clinical outcomes and predict response to anti-cancer treatments [60, 61]. PD-L1 immunohistochemistry can be challenging to interpret, and a poor interobserver variability has been reported [62]; this may be even more impactful in pituitary tumours, given the prominent cytoplasmic staining for PD-L1 in the tumour cells, which may obscure the classical PD-L1 membranous pattern, regarded as the truly positive staining [18]. The heterogeneity across different pituitary tumours, the variable predominance of aggressive pituitary tumours across the series, and the different criteria to define invasive or proliferative tumours, may further contribute for inconsistent results across the studies [22, 25].

Inconsistent or discrepant observations emerged from studies based on mRNA detection when compared with the data generated from IHC studies. Uraki and coworkers [43] reported lower PD-L1 mRNA expression levels in invasive pituitary tumours in comparison with the noninvasive counterparts, and a trend for an association between PD-L1 mRNA expression and higher Ki-67. Beyond the above-mentioned limitations related with the IHC studies, such discrepancies may result from the lack of a strict correlation between mRNA and protein expression of PD-L1, as previously shown by Mei and coworkers [11] in a cohort of pituitary tumours, as well as in other cancers, such as in breast cancer [63]. Discrepancies between mRNA and protein expression of PD-L1 may also derive from inter-assay discordances attributable to technical factors such as antibody epitope, assay/platform performance, as well as due to intra-tumour heterogeneity both at genetic and epigenetic levels [63, 64]. However, Shi et al. [47] described a good correlation between PD-L1 expression at mRNA and protein levels.

Our pooled data suggest that nonfunctioning pituitary tumours may express lower levels of PD-L1, which is in accordance with the data from 2 large series [10, 18]. Another series also reported lower PD-L1 expression in silent corticotroph and null cell tumours in comparison to gonadotroph tumours [43]. However, other studies reported no differences between nonfunctioning and functioning tumours [22, 42, 44, 47].

Higher expression of PD-L1 has been described in proliferative tumours in several studies [45, 49, 65]. Consistently, an increased expression of CD80 and CD86 (CTLA-4 ligands) was found in more proliferative tumours [66]. However, Mei et al. [11] did not find an association between PD-L1 and proliferation, whereas Suteau et al. [22] suggested a possible link between PD-L1 expression and less pituitary tumour proliferation.

Higher expression of PD-L1 has been described in invasive pituitary tumours, including in the paediatric and adolescent setting [38, 47]; however, this is not supported by other studies [22, 44, 65], neither by our pooled data from the IHC-based studies. In contrast, Uraki et al. [43] described lower levels of PD-L1 mRNA expression in invasive nonfunctioning tumours in comparison to the noninvasive counterparts.

A positive association between PD-L1 expression and primary tumours has been suggested in two studies [11, 39], which aligns with our pooled data from IHC-based studies as the proportion of PD-L1 positivity tended to be lower in recurrent (13%) than in primary (21.9%) pituitary tumours. In contrast, Shi et al. [47] reported an association between high PD-L1 expression and recurrence in paediatric and adolescent pituitary tumour patients.

The expression of PD-L1 in pituitary tumours may theoretically influence other biological behaviour and mechanisms beyond invasion, proliferation, and recurrence, such as the modulation of the immune microenvironment or angiogenesis [3, 67]. The expression of PD-L1 by pituitary tumour cells has been correlated with the number of infiltrating lymphocytes, particularly CD8+ T lymphocytes [10, 11, 40]. Lower expression levels of PD-L1 were observed in recurrent tumours, which also tended to have higher immune infiltrates as a possible consequence of the differential expression of PD-L1 [40]. Endothelial markers have been associated with more prominent immune infiltrations, as well as with increased expression of immune checkpoint markers, hence PD-L1 and other immune checkpoint molecules may be involved in the angiogenic mechanisms in pituitary tumours [3, 40].

The expression of PD-L1 in different cancers, and thus the PD-1/PD-L1-mediated tumour immune escape, may be influenced and modulated by different components of the microenvironment, more prominently T lymphocytes, but also other immune cells such as macrophages and stromal cells, as well as cytokines, chemokines, and growth factors [68]. Rather than being a pre-determined interaction, tumour cells may upregulate PD-L1 expression in response to immune assaults, revealing a dynamic and challenging regulatory mechanism [11].

Pituitary tumour-infiltrating T lymphocytes were positively associated with higher PD-L1 expression in the tumours [10, 11]. A possible explanation would be that lymphocytic infiltration may stimulate PD-L1 expression in tumour cells as an immune escape mechanism [11, 22]. Particularly in functioning pituitary tumours, a significant increase of CD3+ and CD4+ populations, but not CD8+ T lymphocytes, accompanied PD-L1 positivity [10, 11]. A lower level of CD8+ lymphocytes in pituitary tumours expressing higher PD-L1 levels was described in paediatric and adolescent cohort [47]. This negative correlation may be due to the CD8+ T lymphocyte role in inducing immunity and PD-L1 effect in promoting immune tolerance.

Macrophages are the most abundant immune cell in the microenvironment of pituitary tumours [67, 69, 70] and were described to be significantly more abundant in PD-L1 negative tumours [24, 65]. However, no association has been found between M2-like macrophages (which have a pro-tumour activity) and the expression of PD-L1 in such tumours [65]. In PIT-1 positive tumours, a notable M2 macrophage infiltration accompanied the overexpression of PD-L1, and both were associated with invasion and/or tumour growth [46].

Turchini et al. [18] reported a high percentage of positivity for PD-L1 in 23 out of 28 PIT-1 plurihormonal tumours, with 10 of these showing strong staining (>50% of neoplastic cells positive for PD-L1), which together with the data from Uraki et al. [43] showing relatively high mRNA expression of PD-L1 in PIT-1 positive nonfunctioning tumours, support the potential involvement of the transcription factor PIT-1 in the modulation of PD-L1 expression [18, 43, 46]. Other transcription factors, such as GATA2, regulate the constitutive expression of PD-L1 and PD-L2 in brain tumours [71]. The Cox-2/PD-L1/arginase 1 pathway has been involved with the occurrence and development of pituitary tumours, and Cox-2 was also hypothesized to positively modulate the expression of PD-L1 in these tumours [44].

Another possible regulatory factor of PD-L1 expression in pituitary tumours may include the pituitary hormones, particularly in the case of hormone excess for functioning tumours, as suggested by Wang et al. [10] who reported positive correlations between the expression levels of PD-L1 and serum levels of prolactin, growth hormone (GH), adrenocorticotropic hormone (ACTH), and cortisol. Additionally, GH can directly modulate the microenvironment and activate some key tumourigenic mechanisms, such as the epithelial-to-mesenchymal transition pathway [72, 73], which may in turn further influence the expression of PD-L1 [74‒76]. The expression of PD-L1 may also been driven by several molecular and genetic mechanisms, such as loss of PTEN, activating mutations of EGFR, or increased STAT3 and AP-1 activation [77‒79], although these pathways and related mechanisms remain to be investigated in pituitary tumours.

The different compositions and distributions of the intestinal flora of pituitary tumour patients may exert systemic effects on the immune system. A recent study showed that somatotroph tumour-faecal microbiota transplantation promotes tumour growth in mouse models by increasing the number of PD-L1-positive cells in tumours, herein enhancing the escape of tumour cells from the immune response; additionally, there may be a compensatory increase in the infiltration rate of CD8+ T cells when their cytotoxic function is inhibited [48].

Nivolumab and pembrolizumab are anti-PD-1 monoclonal antibodies that specifically inhibit PD-1 (expressed on the surface of T-lymphocytes) from binding to PD-L1 expressed on tumour cells, thus promoting immune recognition and tumour cell destruction [8, 9, 80, 81]. The findings from current studies, including cell line and murine models of pituitary tumours [39], seem to indicate that PD-L1 may be a potential target in pituitary tumours. This is further supported by several case reports and small series of patients with aggressive or metastatic pituitary tumours who received anti-PD-L1 drugs, some with successful biochemical and/or radiological responses (reviewed in detail by Ilie and Raverot group [82‒84]). Additionally, the occurrence of hypophysitis in ICI-treated cancer patients suggests that the pituitary gland may constitute a potential immunogenic target.

Pathological markers are often used to predict responsiveness and/or select patients for anticancer treatments, but currently there are no biomarkers to predict ICI efficacy in patients with aggressive or metastatic pituitary tumours. A multicentric French study conducted to assess the efficacy and predictors of immunotherapy response in patients with aggressive or metastatic pituitary tumours, reported partial response rates of 33.3% in patients with corticotroph tumours, and of 16.7% in prolactinoma patients. In this series, pituitary carcinomas seemed to respond better to ICI therapy than aggressive tumours (4 out of 6 carcinomas had partial response, while none of the 9 aggressive pituitary tumours responded), and PD-L1 expression was negative in the 4 responsive corticotroph tumours, suggesting that a negative PD-L1 staining should not preclude the use of ICI [25]. These data suggest that routine immunohistochemical studies for PD-L1 may not be necessary for patients with aggressive or metastatic pituitary tumours being considered for anti-PD-1/PD-L1 inhibitors, given their poor value in predicting response to ICI; however, more studies are needed to fully elucidate this. In contrast, in other cancers, PD-L1 IHC is routinely used to select cases amenable for immunotherapy given its usefulness as predictive biomarker for ICI responsiveness [51, 55, 85].

Although most cancer patients treated with ICI will not develop hypophysitis as a side effect, it is known that a significant proportion of ICI-treated patients will. Data from a large meta-analysis comprising 101 studies and including 19,922 ICI-treated patients reported a pooled estimated incidence of ICI-induced hypophysitis ranging from 0.5 to 10.5%, depending on the type of ICI and as to whether is used isolated or in combination [86]. Anti-PD-1 drugs induce hypophysitis less commonly than anti-CTLA-4 drugs (0.5–1.1% vs. 1.8–5.6%, respectively), but when anti-PD-1 and anti-CTLA-4 drugs are used in combination, the incidence rates increase up to 8.8% (with ipilimumab plus nivolumab) or 10.5% (with ipilimumab plus pembrolizumab) [86]. Nevertheless, most ICI-treated patients will not develop hypophysitis, suggesting a complex interplay between immune system, genetic and environment factors in the predisposition for ICI-related hypophysitis. The pathophysiology of ICI-induced hypophysitis is not completely understood, but few mechanisms have been proposed: pituitary toxicity caused by binding to PD-1 and CTLA-4 which are expressed in normal pituitary cells; circulating anti-pituitary antibodies and lymphocyte infiltration of the pituitary tissue; molecular mimicry with T-cell activity against same antigens present in tumour cells and also in pituitary cells; activation of immune responses due to increasing levels of inflammatory cytokines; epitope spreading as a result of a general nonspecific inflammation secondary to the tumour lysis [20, 87‒91]. The model of ICI-induced hypophysitis may be useful to understand the pattern and mechanisms of the ICI efficacy in some (but not all) patients with advanced pituitary tumours, where a similar interplay between immune system, environment, and genetic factors, together with tumour-specific factors, may explain why some patients respond to ICI while others do not respond [28, 33, 36].

PD-1 and PD-L1 expression may play a role in the biology of cancer and predict aggressiveness and responsiveness to ICI in several cancers. The research focused on the expression of PD-L1 in pituitary tumours has been growing in recent years and shows that PD-L1 is expressed in a significant number of cases. Functioning pituitary tumours typically show higher expression levels of PD-L1. Lactotroph and somatotroph tumours, in particular, appear to show higher PD-L1 levels than other functioning subtypes. While the rates of positive PD-L1 expression may be higher in proliferative tumours, there seem to be no meaningful associations with invasion or recurrence, even though individual studies reported associations between PD-L1 expression and tumour invasion or recurrence.

However, many factors related to the research methodologies and techniques used to study PD-L1 expression, including the lack of consensus about assessment criteria, antibodies for IHC studies, interobserver variability, and heterogeneity across different pituitary tumours, typically lead to variable or discrepant results and may lead to several limitations in comparing the studies and in reaching definitive conclusions. Therefore, to date, the role of PD-L1 expression in the biology and in determining a more aggressive behaviour or poorer clinical outcomes in pituitary tumour patients remains controversial.

PD-L1 expression in pituitary tumours may be modulated by autocrine hormone secretion, PIT-1 positivity, Cox-2 expression and/or by tumour-infiltrating T lymphocytes, macrophages, and potentially other microenvironment components, as well as molecular pathways and genetically-driven factors, but also may be affected by gut microbiota, suggesting complex interactions between tumour cells, tumour microenvironment, and immune system. The complex PD-L1 regulatory mechanisms remain an area of interest to be further investigated.

Favourable response to ICI, particularly anti-PD-1 drugs (nivolumab and pembrolizumab), may occur in a subset of patients with aggressive or metastatic pituitary tumours where other therapeutic lines have previously failed. These positive responses have been documented regardless of PD-L1 positivity, supporting that ICI treatment should be considered in aggressive or metastatic cases, irrespective of the PD-L1 expression status. However, much remains uncertain about the response and usefulness of PD-1/PD-L1 blockade in pituitary tumour patients. Currently, there is no single surrogate marker to predict the benefit or the response to ICI, including the assessment of PD-L1 expression by IHC. Robust data concerning efficacy, safety, and predictive biomarkers or prognostic factors are still needed and justify further clinical and translational research.

The authors have no conflicts of interest to declare.

P.M. is supported by the Neuroendocrine Tumor Research Foundation (NETRF).

M.L.P. and P.M. contributed to conception and design of the work; data acquisition; analysis and interpretation; manuscript writing; reviewing; and final approval. E.L.N. and A.L.S. contributed to conception and design of the work, critical reviewing, and final approval of the manuscript. F.T. contributed to critical reviewing and final approval of the manuscript.