Background
Prolactinomas are among the most prevalent secretory pituitary adenomas and are usually controlled by dopamine agonists, with or without adjuvant surgical and radiation therapy [
1]. However, there are limited treatment options for patients who harbor drug-resistant prolactinomas, where surgical resection either failed or early recurrence occurred after surgery or radiotherapy. Temozolomide (TMZ) is an oral alkylating agent and has shown promise against glioblastoma [
2]. More recently, successful use of TMZ in treating aggressive pituitary adenomas and pituitary carcinomas has been widely reported [
3]. However, the mechanism by which TMZ acts on pituitary tumors remains unclear. Previous studies have documented that O
6 methylguanine-DNA-methyl transferase (MGMT) correlates with the response of glioblastomas to TMZ [
4,
5]. Similarly, low-level MGMT is also associated with a better prognosis in TMZ-treated adenoma patients [
6]. Interestingly, prolactinomas are more sensitive to TMZ, with 75 % of published cases showing a sustained response [
5]. However, accumulating evidence indicates that low-level MGMT is more frequently observed in prolactinomas [
5,
7,
8], suggesting a potential association between responsiveness and reduced MGMT expression [
5]. Despite the fact that MGMT expression might help predict prolactinoma response to TMZ, to our knowledge, the expression profile of MGMT in prolactinomas has not been systematically described.
The mechanisms underlying the genetic, epigenetic, and transcriptional regulation of MGMT expression are not fully understood [
9‐
11]. Promoter methylation represents one of the major factors silencing MGMT gene expression and predicts a favorable outcome in patients with glioblastomas exposed to TMZ [
2]. However, systematic studies have identified a role for MGMT promoter methylation in silencing among pituitary adenomas [
5,
6,
12]. Notably, recent data implies that promoter methylation is not the leading mechanism contributing to low pituitary adenoma MGMT expression [
5,
7]. At the same time, methylated MGMT promoters did not tend to correlate with a response to TMZ exposure [
5]. Moreover, although the majority of sporadic pituitary adenomas are monoclonal, MGMT staining was vastly different between tumors, which again suggested that transcriptional modifications are involved in regulating MGMT expression [
12‐
14].
The role of p53 in modulating MGMT expression is controversial. Several studies have suggested an inverse correlation of p53 with MGMT expression [
15,
16]. However, it has also been shown that accumulation of p53 protein suppresses MGMT expression and promotes cell sensitivity to alkylating agents [
17‐
20]. Furthermore, wild-type p53 protein abrogates MGMT expression by binding directly to the MGMT promoter or by sequestering specificity protein 1 (sp1) [
9]. Interestingly, p53 appears to stimulate MGMT expression, and p53 inhibition sensitizes human glioma cells to TMZ [
10,
21]. Additionally, wild-type p53, rather than mutant p53, promotes MGMT expression by modulating MGMT methylation by reducing DNMT1 expression in lung cancer cell lines [
22], while transduction of IMR90 fibroblasts (human fetal lung cells) with a wild-type p53-adenoviral vector reduced MGMT expression [
19]. In brief, the effect of p53 on MGMT expression may depend on its level and status in specific tissues. The reasons for these inconsistent findings remain elusive, and may be cell type dependent [
20]. According to previous studies, high-level p53 is more frequent in aggressive and atypical pituitary adenomas [
23], consistent with a higher incidence of low-level MGMT expression amongst more aggressive pituitary adenomas [
24]. Moreover, although p53 is one of the most commonly inactivated genes in human cancer, it is rarely mutated in pituitary adenomas [
25]. We therefore asked whether p53 was involved in mediating prolactinoma MGMT expression.
To address these questions, we carried out a retrospective study to evaluate MGMT expression with immunochemistry in a large prolactinoma cohort. Importantly, the potential mechanisms underpinning the suppression of MGMT expression, including MGMT promoter methylation and p53 regulation, were explored.
Discussion
Recently, increasing evidence suggests that TMZ might be a rational therapeutic option for patients with refractory pituitary adenomas and carcinomas [
3,
5]. Surprisingly, prolactinomas appear to be more sensitive to TMZ than the other types of pituitary adenomas [
5]. Meanwhile, successful TMZ therapy is closely associated with low-level MGMT expression [
3,
6]. Herein, profiling MGMT expression and understanding its potential modulation in prolactinomas may pave the way for TMZ treatment of prolactinomas. Our study evaluated MGMT expression in a large cohort of atypical prolactinomas, the majority of which exhibited low-level expression. Systematic analysis showed that MGMT staining was negatively associated with its promoter methylation and p53 protein expression, indicating both epigenetic and transcriptional mechanisms are potentially involved in modulating MGMT expression in prolactinomas.
The MGMT status in pituitary adenomas has been evaluated in several previous studies by immunostaining, showing variable expression levels in different subtypes of pituitary adenomas [
7,
8,
24,
31‐
33]. McCormack et al. first compared MGMT levels in different types of pituitary adenomas [
7]. Their results showed that only 13 % of tumors (
n = 88) demonstrated low-level MGMT expression (<10 %); however, 50 % of the prolactinomas exhibited low-level MGMT expression. Lau et al. examined MGMT expression in 30 pituitary carcinomas and 30 pituitary adenomas [
29]. Low MGMT expression (<25 %) was found in 57 % of carcinomas and 60 % of invasive adenomas. Similarly, 80 % of the prolactinomas (
n = 10) showed low-level MGMT expression. In nonfunctional pituitary adenomas, low MGMT expression (≤50 %) was confirmed in only 24 % of samples as reported by Widhalm et al. (
n = 45) [
31]. In addition, Whitelaw et al. summarized all published TMZ-treated prolactinomas, with 86.7 % staining positive at less than 10 % [
5]. Thus, prolactinomas seem more likely to have reduced MGMT expression compared with other subtypes. Consistent with previous studies, 78 % of samples stained positive for low-level MGMT (25 %), with 44.9 % at less than 10 %. However, in two recent reports, low-level MGMT expression (<25 %) was observed in 91.6 % of GH adenomas (
n = 36) [
32], and 60 % of pituitary corticotroph adenomas (
n = 40) [
24]. More cases are needed to create a more accurate profile of MGMT expression in each subtype of pituitary adenoma.
MGMT expression tended to be more common in aggressive subtypes. Takeshita et al. reported 71 % of Crooke’s cell adenomas (
n = 7) had very low MGMT expression (<5 %), whereas only 1/17 ordinary corticotroph adenomas showed low MGMT expression [
34]. However, in another study, only 50 % of Crooke’s cell adenomas (
n = 12) showed low-level MGMT staining, although all subtype I pituitary adenomas (
n = 7) were immunopositive at <10 % [
24]. Fealey et al. showed that all silent subtype 3 (SS3) pituitary adenomas (
n = 23) had reduced MGMT expression (≤50 %), and 78 % were negative for MGMT immunoreactivity [
35]. In contrast, Salehi et al. reported that 50 % of carcinomas (
n = 10) and 92 % (
n = 11) of SS3 pituitary adenomas showed low MGMT staining (<10 %) [
12]. Similarly, we observed low-level MGMT expression more frequently in atypical prolactinomas. However, no significant relationship was found between MGMT staining and tumor size, recurrence, invasiveness, or Ki-67 index, which is similar to previous studies [
3,
24,
32]. Nevertheless, these clues suggest a rational approach to administer TMZ to refractory adenomas.
In the present study, prolactinoma patients with low-level MGMT expression were younger than those with elevated expression. This is consistent with the observation that MGMT immunoreactivity was inversely correlated with patient age, again suggesting a role for methylation in regulating MGMT expression [
36], although this is also likely a reflection of the younger age of the prolactinoma patients. However, no apparent relationship between MGMT expression and age was observed in patients with Cushing disease or GH adenomas [
24,
32], which might be because of the various mechanisms of MGMT expression in different adenoma subtypes.
The precise mechanisms responsible for MGMT expression remain poorly understood. MGMT promoter methylation, well known as one of the proposed mechanisms underlying suppression of MGMT expression, is associated with the reduced MGMT protein level commonly seen in primary human neoplasms [
37]. Moreover, MGMT promoter methylation is correlated with improved response to TMZ in glioblastoma patients [
2]. However, few systematic studies are available that examine the association between MGMT promoter methylation and MGMT expression in pituitary adenomas [
5,
6,
12]. Therefore, the role of promoter methylation in regulating MGMT expression in pituitary adenomas remains controversial. In the study by McCormack et al., only 9 % of the cases (
n = 46) displayed methylated promoters, although a significant inverse correlation was found between MGMT expression and promoter methylation [
7]. This inverse relationship between MGMT expression and promoter methylation was also observed here in all ten tumors (21.7 %), with methylated promoters showing low-level MGMT staining. In addition, MGMT promoter methylation tended to be more common in the aggressive pituitary adenomas, with 33 % in pituitary carcinomas, 42 % in SS3 adenomas, and 43 % in aggressive pituitary adenoma [
12,
38]. This is consistent with the fact that the aggressive subtypes are associated with low-level MGMT expression. In contrast, a significant relationship between methylation status and MGMT immunoexpression was not observed in the carcinoma and SS3 pituitary adenomas [
12], and a methylated MGMT promoter did not tend to correlate with a response to TMZ exposure [
5]. Together, MGMT promoter methylation is likely to explain low-level MGMT expression in some, but not all, pituitary tumors, as further regulation may occur at the transcriptional, post transcriptional, or translational levels.
The tumor suppressor p53 is expressed in response to stress, and plays a central role in pituitary adenoma pathogenesis [
25]. Although p53 is widely mutated in many human cancers, it is rarely mutated in pituitary adenomas [
25]. In the present study, almost 40 % of the samples had 5 % or more positive staining of p53, which was negatively associated with MGMT immunoexpression, as has been confirmed in other human cancers, including breast, lung, and pancreatic [
15,
16]. Moreover, MGMT suppression was associated with p53 activation, and accumulation of p53 increased the cell response to alkylation agents [
17‐
20]. Additional evidence revealed that wild-type p53 suppresses MGMT by binding directly to the MGMT promoter or by sequestering Sp1 [
9]. On the contrary, some reports suggest that p53 may positively regulate MGMT expression, and that p53 inhibition sensitizes human glioma cells to TMZ [
10,
21]. Expression of wild-type p53 was necessary for inducing MGMT mRNA and protein by ionizing radiation [
18]. Additionally, wild-type p53 rather than mutant p53 promoted MGMT expression by modulating MGMT methylation, presumably by reducing DNMT1 (DNA-methyltransferase 1) expression in lung cancer cell lines [
22].
Our study has several limitations. First, the number of specimens for evaluating promoter methylation was relatively small, which may fail to describe the complete picture of prolactinoma MGMT promoter methylation and its correlation with tumor and patient characteristics. Second, because of insufficient tumor tissue, we failed to detect MGMT expression by western blot, but relied solely on immunohistochemistry staining. The lack of standard scoring, and operator variability, made it more difficult to determine accurately the level of MGMT expression. Therefore, three independent observers assessed immunoreactivity, with the data analyzed using the most widely used methods. Third, patients in China have no excess to cabergoline. Some patients included in the study may have been treated with cabergoline and not needed surgery or temozolomide treatment. Moreover, these data are not applicable to all prolactinoma patients, but only to the type of specimens analyzed here. Thus, it is difficult to say whether MGMT reduced expression was a general characteristic of prolactinomas. In addition, because all patients were not treated with TMZ before or after surgery, a possible relationship between MGMT level and TMZ intolerance could not be established. Finally, because of the nature of retrospective studies, the causality of MGMT silencing and promoter methylation or p53 protein activation could not be established. Although an inverse correlation between p53 and MGMT was found, it was weak. Nevertheless, our results strongly suggest that promoter methylation is not the sole mechanism underlining MGMT silencing in prolactinomas.
Competing interests
The authors declare that they have no competing of interest.
Authors’ contributions
XJ, HW, and YZ participated in the study design. XJ, BH, XW, and BS participated in the immunohistochemistry, MSP, and data analysis. HW, DH, and ZM were responsible for the surgery and sample collection. XJ, HW, and YZ participated in editing and proof reading. All authors read and approved the final manuscript.