Aggressive pituitary tumours: the role of temozolomide and the assessment of MGMT status

Authors

  • Ann I. McCormack,

    1. Cancer Genetics Unit, Hormones and Cancer Group, Kolling Institute of Medical Research, Royal North Shore Hospital, Sydney, NSW, Australia
    2. Department of Endocrinology, Oxford Centre for Diabetes, Endocrinology and Metabolism, Churchill Hospital, Oxford, UK
    Search for more papers by this author
  • John A. H. Wass,

    1. Department of Endocrinology, Oxford Centre for Diabetes, Endocrinology and Metabolism, Churchill Hospital, Oxford, UK
    Search for more papers by this author
  • Ashley B. Grossman

    1. Department of Endocrinology, William Harvey Research Institute, Barts and the London School of Medicine, London, UK
    Search for more papers by this author

Dr Ann McCormack, Cancer Genetics Unit, Kolling Institute of Medical Research, Royal North Shore Hospital, St Leonards, NSW 2065, Australia. Tel.: +61 2 9926 7176; fax: +61 2 9926 8484; e-mail: ann.mccormack@sydney.edu.au

Abstract

Eur J Clin Invest 2011; 41 (10): 1133–1148

Abstract

Background  Aggressive pituitary tumours are associated with substantial morbidity and mortality. Treatment options are often limited, and chemotherapy has been reserved as salvage therapy although historically results have often been disappointing. However, temozolomide, an oral alkylating agent, has recently demonstrated significant activity against these tumours. A DNA repair protein, 06-methylguanine-DNA methyltransferase (MGMT) has been suggested as a biomarker to predict response to temozolomide in pituitary tumours.

Materials and methods  This paper will review the current literature on temozolomide and pituitary tumours and discuss the recent controversy surrounding the value of determining the MGMT status in this tumour group. A PubMed search was performed to retrieve articles, using the terms ‘pituitary tumour’ and ‘temozolomide’.

Results  Overall, 24/40 (60%) of the published cases demonstrated a response to temozolomide therapy. The highest response rates were seen amongst prolactinomas (73%) and ACTH-secreting tumours (60%), whilst nonfunctioning pituitary tumours exhibit lower response rates (40%). Responsivity is typically evident in the first 3 months of therapy and may be dramatic and sustained. Low MGMT expression, as determined by immunohistochemistry, is associated with a high response rate (76%), whilst high MGMT expression has not been associated with responses. MGMT promoter methylation does not correlate with temozolomide response.

Conclusions  Temozolomide is the first chemotherapeutic agent to show substantial response rates in aggressive pituitary tumours. MGMT immunohistochemistry, but not MGMT methylation analysis, shows promise as a predictive tool. Prospective clinical trials are now necessary to more accurately determine the efficacy of this agent in this patient group.

Introduction

Aggressive pituitary tumours are a distinct biological and clinical entity [1,2]. These tumours exhibit continued growth despite multimodal therapy, including surgery and radiotherapy. Whilst these tumours have malignant potential, the term pituitary carcinoma is strictly reserved for those tumours with demonstrated craniospinal or systemic metastases [3]. However, even when not strictu sensu carcinomas, they may be very difficult to control and ultimately prove to be lethal. It is now widely accepted that pituitary carcinomas arise through the progressive accumulation of genetic alterations [4], although there are very occasional cases of apparent de novo presentation [5]. Aberrations of genes involved in regulation of cell cycle progression, growth factor signalling pathways, angiogenesis and cellular migration appear to be particularly important in pituitary tumour progression.

Recent epidemiological studies have demonstrated a higher prevalence of clinically significant pituitary tumours than has been previously recognised, with 1 case occurring per 1064–1289 people [6,7]. However, the prevalence of aggressive pituitary tumours is not clear. Strict pituitary carcinoma is very rare, accounting for 0·2% of pituitary tumours [8]; however, the frequency of invasive pituitary tumours, defined as infiltration of surrounding tissue, is reported at 40–50% [4,9]. The WHO category of ‘atypical pituitary adenoma’ is a pathological classification that identifies a tumour with features that may indicate the potential for aggressive behaviour [10]. The frequency of atypical pituitary adenomas in the German Registry of Pituitary Tumours was recently reported at 2·7% [11].

The majority of debilitating symptoms associated with aggressive pituitary tumours relate to mass effects from tumour bulk, particularly loss of vision and headaches associated with sellar disease. Endocrine symptoms arising from hormone hypersecretion can be problematic, particularly those associated with active Cushing’s disease. The majority of patients with pituitary carcinoma die within 1 year of diagnosis [8].

Historically, treatment options for these tumours have not been satisfactory. Debulking surgery is rarely curative, but may offer substantial symptom control for variable periods of time [12]. Radiotherapy, particularly when applied early in the course of disease, may prove useful in controlling tumour growth but only occasionally induces true tumour regression [13,14]. The experience with stereotactic radiosurgery is limited and may only be useful for small tumours [15]. One of the hallmarks of an aggressive pituitary tumour is resistance to dopamine agonists or somatostatin analogues [12]. Various chemotherapeutic regimens have been used as salvage therapy, although any response is usually partial and transient [4].

However, temozolomide, an oral alkylating agent, has recently demonstrated promising activity against these tumours, and experience with its use is growing at a rapid rate. Furthermore, 06-methylguanine-DNA methyltransferase (MGMT), a DNA repair protein, which specifically removes the alkylating lesion induced by temozolomide, has been suggested as a biomarker to predict response to temozolomide in pituitary tumours [16,17]. In this paper, we will review the use of temozolomide in the management of aggressive pituitary tumours and discuss current controversies surrounding the value of determining the MGMT status and the methods employed in its testing.

Temozolomide

Mechanism of action

Temozolomide is a second-generation alkylating agent of the imidotetrazine class, first synthesised in 1984 [18]. It is structurally and functionally related to dacarbazine. Unlike dacarbazine, which requires hepatic conversion to the active derivative, methyl-triazeno-imidazole-carboxamide (MTIC), temozolomide is spontaneously activated at physiological pH. Degradation of MTIC to the highly reactive methyldiazonium ion is responsible for forming toxic methyl adducts with DNA bases. Temozolomide has other advantages to dacarbazine: it can be administered orally with 100% bioavailability, it readily crosses the blood–brain barrier, and it displays predictable linear pharmacokinetics with little interpatient variation [19,20].

The critical methyl adduct produced by temozolomide and responsible for the greatest cytotoxicity is the lesion at 06-guanine (06-MeG). However, this accounts for only 5% of the alkylating lesions induced by temozolomide. N7-methylguanine is the most frequent lesion (70%) followed by N3-methyladenine (9%). The latter lesions are efficiently repaired by a network of proteins forming the base excision repair (BER) pathway. Two other important DNA repair systems are involved in the cellular defence against lesions caused by temozolomide [19]. 06-methylguanine-DNA methyltransferase (MGMT), discussed in detail below, can directly remove the methyl group at 06-guanine. In the absence of MGMT, unrepaired methylated guanine lesions incorrectly pair with thymine triggering the mismatch repair pathway (MMR). However, an intact MMR, paradoxically, leads to futile attempts at repair as MMR targets newly synthesised DNA, resulting in incorrect reinsertion of thymine opposite the 06-MeG lesion. Ultimately, cycles of ineffectual repair lead to DNA-strand breaks and cell cycle arrest, which is followed either by activation of the apoptotic cascade or a senescence-like state [19,21].

Therapeutic use

The original Phase 1 clinical trial was published in 1992 and established the standard temozolomide dosing regimen as 200mg m−2 given daily for 5 days every 28 days [22]. Phase 3 trials have established temozolomide as standard of care in the management of glioblastoma multiforme and advanced melanoma [23,24]. Temozolomide also has activity against low-grade gliomas and malignant neuroendocrine tumours [25,26]. Alternative dosing schedules of temozolomide have been shown to be effective in patients with progressive or recurrent glioma [27]. Common alternate regimens include ‘dose-dense’ protocols, such as 150mg m−2 days 1–7 and days 14–21 of a 28 day cycle, and ‘metronomic’ protocols using continuous daily low-dose (50–75mg m−2) temozolomide. Metronomic dosing of temozolomide has been suggested to have additive anti-angiogenic properties [28]. In the landmark Phase 3 trial for newly diagnosed glioblastoma multiforme, low-dose continuous temozolomide was given concurrently with radiotherapy for 6 weeks, followed by standard dosing of temozolomide for an additional 6 months [23]. However, other studies have reported improved survival with use of temozolomide for 12–24 months, and recent case reports have highlighted sustained remissions with treatment out to 8 years [29–31].

Side effect profile

When compared to most chemotherapeutic agents, temozolomide has proven to be generally very well tolerated. The most common and dose-limiting toxicity is myelosuppression, with Grade 3 or 4 thrombocytopaenia occurring in 7–17% across studies, and neutropaenia to a lesser extent [18]. There is a predictable nadir that occurs between 21 and 28 days of each cycle with recovery taking 1–2 weeks, and importantly, there is no evidence of cumulative haematological toxicity. Dose-dense regimens cause increased myelotoxicity [18] and lower pretreatment MGMT levels in peripheral blood mononuclear cells appears to be a strong predictor of increased myelotoxicity [32]. Concurrent radiotherapy, dose-dense regimens, corticosteroids and poor performance status increase the risk of opportunistic infection, and in these settings prophylactic trimethoprim-sulfamethoxazole or pentamidine are recommended [18]. There are rare reports of myelodysplastic syndrome, aplastic anaemia and haematological malignancy developing following temozolomide therapy [33–35]. Nonhaematological side effects are usually mild–moderate, most frequently nausea/vomiting (34%) and fatigue (20%). Other less common adverse effects include headache, anorexia, diarrhoea and rash [36]. Rare cases of hypersensitivity pneumonitis [37], Stevens–Johnson syndrome [38] and hearing loss [26] have also been described.

MGMT

Activity and regulation

06-MeG adducts are repaired in the presence of MGMT. Covalent transfer of the methyl group to a cysteine residue on MGMT results in a protein conformational change that targets MGMT for ubiquitin-dependent degradation. Further repair can only occur following MGMT resynthesis de novo, and this may take up to 72 h. MGMT expression in normal cells has significant tissue-specific variation, being highest in the liver and colon and low in brain and haematopoietic progenitor cells. There is also a large degree of interindividual variation in MGMT expression. In human tumour cells, MGMT expression displays even greater variability than normal tissue, but is generally higher. MGMT activity is high in breast cancer, lung and renal carcinomas with lower activity found in brain tumours and melanoma [39].

The regulation of MGMT expression is not fully understood. Tissue variability in expression indicates epigenetic regulation, and tissue-specific transcription factors are likely to be important. The MGMT promoter contains Ap-1 and Sp1 transcription factor binding sites, two glucocorticoid response elements, and a proposed ERα-like domain. MGMT expression is upregulated by corticosteroids [39] and may be related to the oestrogen status of cells [40]. The tumour suppressor protein p53 interacts directly with the MGMT promoter and is required for gene induction following DNA damage. The MGMT promoter is CpG rich and is susceptible to methylation, particularly in the context of tumour formation, resulting in gene silencing. MGMT promoter methylation is seen with greatest frequency in gliomas, occurring in 30–88% of such tumours, but occurs less frequently in melanoma (11%) and rarely in breast or ovarian cancer [41,42]. Loss of heterozygosity at 10q26, the region containing the MGMT gene, is seen in gliomas, and suggests that gene deletion may be another potential cause of loss of MGMT activity; however, MGMT gene mutations are rarely found in tumours [19,41,43].

MGMT status as a biomarker of response to temozolomide

Expression of MGMT represents the major mechanism of resistance to temozolomide and other alkylating agents [44]. The link between low levels of MGMT and temozolomide efficacy has been repeatedly demonstrated in brain tumours, and also seen in melanoma and neuroendocrine tumours [45–49]. Conflicting studies highlight the pitfalls of methods commonly used to assess a tumour’s MGMT status, namely methylation analysis using methylation-specific PCR (MSP) and immunohistochemistry to determine the level of MGMT protein expression [26,50–54]. These studies also suggest there are likely to be other molecular mechanisms contributing to temozolomide responsivity.

Temozolomide experience in pituitary tumours

The first reports detailing the successful use of temozolomide in the management of pituitary carcinomas emerged in 2006 [55,56]. Subsequent case reports documented similarly dramatic responses in patients with locally aggressive pituitary adenomas [57,58]. A rapid uptake in use of temozolomide for these tumours has ensued internationally, with experience in some 40 such cases published at the time of writing (Table 1) [16,17,55–68]. There was a certain degree of positive reporting bias in the initial publications. Recent case series clearly demonstrate a range of responses to temozolomide therapy in aggressive pituitary tumours, and yet it is the first chemotherapeutic agent to show significant activity against such tumours [65–67].

Table 1.   Published cases – temozolomide and aggressive pituitary tumours
ReferencesAgeSexSubtypeKi67Previous treatmentsTMZ regimeLength of therapy (months)Response (Clin/Horm/Rad)Side effectsCourse subsequent TMZ
  1. PRL, prolactin- secreting; ACTH, adrenocorticotrophin-secreting; NF, non-functioning; GH, growth hormone- secreting; NS, not stated; DA, dopamine agonist; SSA, somatostatin analogue; S, surgery; R, radiotherapy (all forms); C, chemotherapy; BA, bilateral adrenalectomy; NS, not stated; ‘+’, not ceased; Y, Yes; CR, complete response; NA, not applicable; PR, partial response (amount of reduction%); NR, no response; PD, progressive disease; S, stable.

[55]72MPRL carcinoma Mets: spineAbsentDA, S(5), R(1)150–200 mg m−2 per day 5 days of 28 days18Y/CR/PRNausea, nasal suffusionStable 24 months after cessation
[56]26MPRL carcinoma Mets: spine10%DA, SSA, S(3), R(2) I131MIBG, C (1)150–200 mg m−2 per day 5 days of 28 days10Y/PR/PRFatigue → cessationProgressive disease 15 months after cessation
[57]46MPRL adenoma40–60%DA, S(6), R(1)150–200 mg m−2 per day 5 days of 28 days7Y/PR/PRNSFurther surgery of soft tumour
[58]52FPRL adenoma< 5%DA, SSA, raloxifene, S(1), R(2)150–200 mg m−2 per day 5 days of 28 days60Y/PR/PRFatigueIncreasing PRL during last year of therapy; some improvement with increased dose but developed pancytopaenia requiring cessation, 1yr pasireotide
[16]48MPRL carcinoma Mets: craniospinal & boneNSDA, SSA, S(6), R(2)150–200 mg m−2 per day 5 days of 28 days29Y/PR/PRNilIncreasing PRL during last months of therapy; death from bowel carcinoma 4 months after cessation TMZ
[68]47MPRL adenoma (MEN1)NSDA, S(1), R(2), C(1)150–200 mg m−2 per day 5 days of 28 days11Y/CR/PRFatigue → cessationHormonal recurrence 1 yr after cessation, then clinical progression; no response 2nd course TMZ
[61]48FPRL carcinoma Mets: cervical lymph nodes5% (10% mets)DA, SSA, S(2), R(1)150–200 mg m−2 per day 5 days of 28 days23Y/CR/PR(62% pit, CR mets)NauseaStable 34 months after cessation
[61]60MPRL adenoma2%DA, SSA, S(1)150–200 mg m−2 per day 5 days of 28 days12Y/CR/PR (80%)Leukopenia, thrombocytopaenia, nauseaStable 12 months after cessation
[65]32MPRL carcinomaNSS (1), R (1)150–200 mg m−2 per day 5 days of 28 days24NS/CR/PR (60% pit, CR mets)NilStable 10 months aftercessation
[65]52MPRL adenoma2%DA, S(1), R(2)150–200 mg m−2 per day 5 days of 28 days8NS/PD/PDThrombocytopaeniaNS
[65]54MPRL carcinoma7%S(4), R(3)150–200 mg m−2 per day 5 days of 28 days5NS/NR/PD (metastases developed on TMZ)NilNS
[65]30FPRL carcinoma30%DA, S(4), R(2)150–200 mg m−2 per day 5 days of 28 days3NS/NR/PD (metastases developed on TMZ)Agranulocytosis → cessationNS
[66]NSNSPRL adenoma> 20%S(2), R(1)75 mg m−2 per day for 21 days, 7 days off11+Y/PR/PR (> 80%)NilNA
[67]62MPRL adenoma9%DA, S(1), R(1)150–200 mg m−2 per day 5 days of 28 days12S/PR/SNSProgressive disease 6 months after cessation; further surgery
[67]57FPRL adenomaNSDA, S(2), R(1)150–200 mg m−2 per day 5 days of 28 days12Y/CR/PRNSNS
[59]64FACTH adenoma (Nelson’s)HighS(1), R(1), BA150–200 mg m−2 per day 5 days of 28 days6Y/PR/PRNauseaProgressive disease 1 year after cessation; further progression 2nd course temozolomide
[60]43FACTH adenomaNSS(3), R(1)150–200 mg m−2 per day 5 days of 28 days16+Y/NS/PRNSNA
[60]60MACTH carcinoma (Nelson’s) Mets: spineNSSSA, S(2), R(1), BA150–200 mg m−2 per day 5 days of 28 days12Y/PR/PRNSProgressive disease 4 months after cessation
[63]46FACTH carcinoma (Nelson’s) Mets: liver3%DA, SSA, pioglitaz150–200 mg m−2 per day 5 days of 28 days36+Y/DR/PR (CR mets)NilNA
[64]50MACTH carcinoma (Nelson’s) Mets: spine & bone31%DA, SSA, rosiglitazone, ketoconazole, metyrapone, S(2), R(2), BA150–200 mg m−2 per day 5 days of 28 days (concurrent capecitabine)5Y/CR/PRNilRapid development progressive disease during 5th month on TMZ; 2 cycles etoposide/cisplatin; death 3 months after cessation TMZ
[65]31MACTH carcinoma (Nelson’s)20%S(3), R(2), BA150–200 mg m−2 per day 5 days of 28 days (carmustine added cycle 8) (concurrent capecitabine)14NS/NR/NRFatigueNS
[65]49MACTH adenoma20%S(3), R(1)150–200 mg m−2 per day 5 days of 28 days (carboplation added cycle 4)7NS/NR/NRFatigueNS
[65]38MACTH carcinoma Mets:spine10%S(4), R(1), mitotane150–200 mg m−2 per day 5 days of 28 days6+Y/PR/PRNSNA
[65]42FACTH adenoma0·50%S(2), R(2)150–200 mg m−2 per day 5 days of 28 days4+Y/PR/PRThrombocytopaeniaNA
[66]NSNSACTH adenoma18%S(3), R(2)75 mg m−2 per day for 21 days, 7 days off11Y/CR/PR (> 80%)Sensorineural hearing loss, fatigue, headachesStable 6 months after cessation
[67]64MACTH adenomaNSDA, SSA, S(2), R(2)150–200 mg m−2 per day 5 days of 28 days3PD/NR/PDNSSalvage radiotherapy, 1 yr cisplatin/capecitabine,4 months gemcitabine, 8 months etoposide; death 27 months following cessation TMZ
[67]52MACTH carcinoma (Nelson’s) Mets: intracranial1%S(3), R(2)150–200 mg m−2 per day 5 days of 28 days12Y/CR/CR (mets)NSStable 28 months after cessation
[67]55FACTH adenoma (Nelson’s)5%SSA, S(2), R(2), BA150–200 mg m−2 per day 5 days of 28 days12S/S/SNSProgressive disease 6months after cessation; further surgery, trial SOM- 230
[67]53FACTH adenoma2·50%S(2), R(2)150–200 mg m−2 per day 5 days of 28 days6S/PD/PDNSDeath 5 months after cessation
[56]38MNF carcinoma (gonadotroph) Mets: spine, bone1%SSA, S(2), R(3)150–200 mg m−2 per day 5 days of 28 days12Y/NA/PRLymphocytopaeniaStable 16 months after cessation
[17]41MNF adenoma (silent ACTH)NSNSNSNSNS/NA/NRNSNS
[60]28FNF adenoma (incidental)NSNilNS10+NA/NA/PRNSNS
[61]20MNF adenoma (null cell)2%DA, SSA, S(6), R(1)150–200 mg m−2 per day 5 days of 28 days15+Y/NA/PR (55%)NauseaNA
[62]70MNF adenoma (gonadotroph)2–6%S(3), R(1)150–200 mg m−2 per day 5 days of 28 days5NS/NA/PRNSDeath due to massive PE whilst on temozolomide thought unrelated
[66]NSNSNF adenoma (null cell)< 3%S(1)75 mg m−2 per day for 21 days, 7 days off10+S/NA/SFatigue, headachesNA
[66]NSNSNF adenoma (gonadotroph)< 3%S(3), R(1)75 mg m−2 per day for 21 days, 7 days off13+S/NA/SFatigue, headachesNA
[66]NSNSNF adenoma (null cell)6%S(2)75 mg m−2 per day for 21 days, 7 days off 10+S/NA/SFatigue, dry mouthNA
[66]NSNSNF carcinoma (null cell)> 20%S(3), R(2)75 mg m−2 per day for 21 days, 7 days off (concurrent thalidomide)2NA/NA/SFatigue, severe dizziness → cessationDeath due to unrelated cause
[66]NSNSNF carcinoma (null cell)> 20%S(2), R(1)75 mg m−2 per day for 21 days, 7 days off7PD/NA/PDFatigue, headachesDeath due to progressive disease
[16]48MGH adenomaNSDA, SSA, S(6), R(1)150–200 mg m−2 per day 5 days of 28 days3PD/NR/NRNilDeath due to progressive disease

Patient characteristics

Temozolomide has been most commonly used as salvage therapy for aggressive prolactin (PRL)-secreting (seven carcinomas, eight adenomas) or ACTH-secreting pituitary tumours (six carcinomas, nine adenomas), probably reflecting the predominance of these tumour types amongst cases of aggressive pituitary tumours. Treatment in the setting of Nelson’s syndrome accounted for 6 of the 15 ACTH-secreting tumours. One of the aggressive prolactinomas occurred in a patient with MEN1 [68]. Use of the agent in clinically nonfunctioning adenomas (NFPAs) has also been reported, comprising three carcinomas and seven adenomas. Five of the NFPAs were ‘null cell’ tumours, three of gonadotrophin lineage and one a silent corticotroph adenoma. There has been only one case of temozolomide used in the context of an aggressive GH-secreting adenoma.

Treatment with temozolomide has been reported more often in male patients with aggressive pituitary tumours (23/35). The average patient age is 46·5 years, range 20–72 years. Prior to temozolomide therapy, patients have typically required multimodal therapy, in many cases over several years, but tumour growth has ultimately proven relentless. Most patients have undergone 2–3 neurosurgical procedures, some up to six, and in addition, radiotherapy or radiosurgery has been administered on at least one occasion in the majority of cases. These aggressive tumours all display eventual resistance to medical therapies such as dopamine agonists and somatostatin analogues. Where qualitatively reported, 19 of 28 cases (68%) had Ki-67 values ≥ 3% (Table 1), in keeping with an aggressive cohort, with other atypical features also commonly seen (but not uniformly collected or reported), including p53 positivity and mitotic activity.

Temozolomide regimes

On the whole, these patients are chemotherapy naïve, with only two patients having been treated unsuccessfully with previous chemotherapeutic regimes [56,68]. Temozolomide was administered as monotherapy in all but three cases. In two cases, dual therapy (temozolomide/interferon and temozolomide/thalidomide) resulted in significant side effects, and in the third case, temozolomide was administered with capecitabine for 5 months and was well tolerated [56,64,66]. Apart from a handful of patients, the standard temozolomide regime of 150–200 mg m−2 per day for 5 days every 28 days was used. Bush et al. used a dose-dense regime (75 mg m−2 per day for 21 days with 7 days off), whilst in the patient receiving concurrent capecitabine, temozolomide was used at a dose of 100–200 mg m−2 per day on day 10–14 following capecitabine on days 1–14 [64,66]. The duration of temozolomide therapy has been highly variable. Progressive disease has necessitated cessation as early as 2 or 3 months, whilst other cases demonstrating good response continue to receive temozolomide beyond 3 years [58,63,65,67]. In one case, treatment was extended to bimonthly in the second year of therapy and trimonthly in the third year [63]. Most frequently, patients are completing 12 cycles of therapy as a prescribed treatment course. Various other chemotherapeutic agents have been tried in a few cases following demonstrable progressive disease, and as historically indicated, any response is usually partial and short-lived. Carmustine was added to the last few cycles of temozolomide in one case of nonresponse, and carboplatin in another, with no apparent benefit [65].

Adverse events

On the whole, temozolomide therapy has been well tolerated. Common side effects include fatigue, nausea and headache. In patients who have developed myelosuppression, a reduction in dose or extension in dosing interval permitted continuation of therapy [61,65]. With regard to serious adverse events, there has been one case of severe agranulocytosis developing after three cycles of therapy necessitating treatment discontinuation [65] and one report of a patient in whom irreversible sensori-neural hearing loss developed [66].

Outcomes

Table 1 illustrates responses (clinical, hormonal and radiological) by tumour type. Table 2 shows overall response by tumour type. Overall, 24 patients (60%) demonstrated a response to temozolomide therapy. Responsivity was seen equally amongst carcinomas and adenomas (46% and 54% respectively). Amongst the prolactinoma cohort, 11/15 (73%) responded; of the ACTH-secreting tumours, 9/15 (60%) showed a response, with no differential response apparent between Nelson’s and typical Cushing’s disease cases. As a group, the NFPAs did not respond as well, with 4/10 responders (40%). However, 2/3 of the gonadotrophin-derived tumours responded compared with 1/5 null cell tumours.

Table 2.   Published cases- outcome and MGMT status
ReferencesSubtypeGradeOutcomeIHC resultMethylation result
  1. PRL, prolactin- secreting; ACTH, Adrenocorticotrophin-secreting; NF, non-functioning; GH, growth hormone-secreting; ND, Not done.

  2. *Poor internal control.

[57]PRLAdenomaResponseNegativeND
[58]PRLAdenomaResponseNDND
[68]PRLAdenomaResponseNegativeMethylated
[61]PRLAdenomaResponseNegativeND
[66]PRLAdenomaResponseLowUnmethylated
[67]PRLAdenomaResponseLow*Unmethylated
[67]PRLAdenomaStableLowUnmethylated
[65]PRLAdenomaProgressive diseaseIntermediateND
[55]PRLCarcinomaResponseNDND
[56]PRLCarcinomaResponseNDND
[16]PRLCarcinomaResponseNegativeMethylated
[61]PRLCarcinomaResponseNegativeND
[65]PRLCarcinomaResponseNDND
[65]PRLCarcinomaProgressive diseaseLowMethylated (8·5%)
[65]PRLCarcinomaProgressive diseaseHighUnmethylated
[59]ACTHAdenomaResponseNegativeND
[60]ACTHAdenomaResponseNegativeND
[65]ACTHAdenomaResponseLowMethylated (9·8%)
[66]ACTHAdenomaResponseLowUnmethylated
[67]ACTHAdenomaStableHighUnmethylated
[65]ACTHAdenomaNo responseLowUnmethylated
[67]ACTHAdenomaProgressive diseaseHighUnmethylated
[67]ACTHAdenomaProgressive diseaseNDND
[60]ACTHCarcinomaResponsePositiveND
[63]ACTHCarcinomaResponseLowND
[64]ACTHCarcinomaResponseNDND
[65]ACTHCarcinomaResponseIntermediateUnmethylated
[67]ACTHCarcinomaResponseLowND
[65]ACTHCarcinomaNo responseIntermediateMethylated (2·6%)
[60]NFAdenomaResponseNDND
[61]NFAdenomaResponsePositiveND
[62]NFAdenomaResponsePositiveND
[66]NFAdenomaStableIntermediateUnmethylated
[66]NFAdenomaStableIntermediate/HighUnmethylated
[66]NFAdenomaStableIntermediate/HighUnmethylated
[17]NFAdenomaNo responsePositiveND
[56]NFCarcinomaResponseNDND
[66]NFCarcinomaNo responseIntermediate/HighUnmethylated
[66]NFCarcinomaProgressive diseaseLowUnmethylated
[16]GHAdenomaProgressive diseasePositiveND

Invariably, clinical response occurred in concert with subsequent radiological response, and in the functioning tumours, a hormonal response as well. Response is most commonly seen within the first 3 months, clinical and hormonal responses typically occurring earlier and overall to a greater degree than radiological response. However, maximal response was often seen after 10–12 cycles of treatment [61,67]. Overall, complete hormonal responses (normalisation) were seen in 50% and partial responses (> 50% decrease) in 50%. Applying traditional RECIST radiological criteria to document degree of tumour shrinkage (partial response > 30% decrease in tumour diameter; complete response disappearance of tumour), all responders showed a partial response. However, in many cases, this underestimated the degree of tumour shrinkage, often being > 50% and in some cases > 80% [61,64,66]. Furthermore, three patients had a complete response in metastatic lesions [63,65,67]. Morphological evidence of response has also been reported in one case, with tumour tissue following temozolomide therapy, demonstrating increased differentiation, a decreased Ki-67 (40–60% to 5%) and mitotic index [69].

Perhaps most impressively, a number of patients have had a sustained response following cessation of temozolomide therapy, with no tumour regrowth out to as long as 3 years [55,61,67]. However, these tumours appear to ultimately regrow, usually heralded by rising hormonal levels (personal communication with R.Lechan, W.Braund, R.Ross, W. Drake).

In addition to the responders, there have been five patients (13%) who are best described as demonstrating stable disease whilst on temozolomide therapy [66,67]. These tumours have shown clear arrest of tumour growth or minor reductions in hormone levels and/or tumour size.

In contrast, seven patients (18%) have shown clear tumour progression during treatment with temozolomide: four patients died as a consequence of their pituitary tumour during the follow-up period (one carcinoma, three adenomas) [16,66,67], two patients developed metastatic disease [65], one patient had an increase in tumour size and lack of hormonal control [65].

Additionally, four cases are best described as overall nonresponders. In these cases, either treatment duration was not long enough to determine response or it is not apparent from the information provided whether patients had stable or progressive disease during therapy [17,65,66].

In vitro work provides evidence in support of temozolomide activity against pituitary tumours. Sheehan et al. demonstrated inhibition of cell proliferation and induction of apoptosis across three pituitary adenoma cell lines (MMQ, GH3 and AtT20). A reduction in prolactin was also seen in the prolactinoma cell line [70].

MGMT status and response to temozolomide in pituitary tumours

MGMT immunohistochemistry to determine the level of MGMT protein expression has been performed retrospectively now on a number of the published pituitary tumour cases in whom temozolomide has been used. In a significant proportion of these tumours, MGMT promoter methylation status has also been assessed (Table 2). Currently, it is unclear as to whether determination of the MGMT status of a pituitary tumour aids in predicting a response to temozolomide therapy. In drawing conclusions on this subject, there are several factors to consider. These include the limitations of techniques used for MGMT testing, the inherent problems with studying small numbers of patients retrospectively across different centres, and attaching realistic expectations to the study of a single potential biomarker.

MGMT Immunohistochemistry

Immunohistochemistry is an attractive technique for determining the MGMT expression level of tumour tissue. It can be applied to paraffin-embedded tissue, and the equipment and technical expertise is available in most laboratories. The experience with MGMT immunohistochemistry in gliomas has not been altogether satisfactory, as numerous nonneoplastic elements present within tumour tissue can make reliable assessment difficult [51]. Applied to pituitary tumours, composed of a predominantly homogeneous population of tumour cells, the technique does appear to hold more promise.

However, the clinical value of utilising MGMT immunohistochemistry to predict response to temozolomide therapy is currently obscured by the absence of standardised scoring across studies. Earlier cases were reported as either ‘positive’ or ‘negative’, indicating the presence or absence of nuclear staining for MGMT. However, the extent of nuclear staining amongst the positive cases varied enormously, from ‘a few positive nuclei’ through to diffusely strong staining [16,61]. More recent cases have been reported semi-quantitatively. However, there is no consensus about percent nuclear MGMT staining that constitutes low, intermediate or high MGMT expression [63,65–67,71]. In this review, low refers to absent or ≤ 10% MGMT expression, intermediate 10–50%, intermediate/high 50–90% and high ≥ 90% MGMT expression. Bush et al. have also highlighted the complexities of immunohistochemistry reporting by adding scores for staining intensity and demonstrating significant interobserver variation in scoring results [66]. In addition, use of different antibody clones and tissue preservation methods may contribute to difficulties in comparison with results between studies.

Furthermore, biological variables are likely to have a significant influence on the clinical response to temozolomide therapy. Marked heterogeneity in MGMT expression is commonly encountered within tumour samples [66], and it is possible that primary and metastatic tumours may have differing levels of MGMT expression. Mohammed et al. recently examined MGMT expression in a double pituitary adenoma, and found positive immunostaining in the GH adenoma, whilst the prolactin adenoma component was negative [72]. Furthermore, MGMT expression levels in tumour tissue may change over time [62], although Lau et al. reported stable MGMT expression in the majority of primary and recurrent tumour samples [73]. There is also uncertainty about the effects of radiotherapy on tumour levels of MGMT expression. Limited data to date does not indicate a significant impact of radiotherapy treatment on MGMT expression in pituitary tumours [66,73].

MGMT expression in pituitary tumours

Whilst some of the difficulties described above may interfere with the reliability of immunohistochemistry as a predictive tool, there are some clear inferences to be gleaned from the combined results in these early studies (Table 2). First, low (< 10%) or negative MGMT expression appears to predict a high chance of response to temozolomide therapy. Thirteen of 17 cases (76%) with negative or low MGMT expression demonstrated a response. However, there are patients with low levels of staining who show no response or even progressive disease. This may be because of technical or biological factors already mentioned, but there are likely to be other resistance mechanisms operating apart from MGMT in these patients. Deficiency in the mismatch repair pathway in gliomas can cause temozolomide resistance irrespective of MGMT status [74].

Conversely, intermediate/high (> 50%) or high (> 90%) MGMT expression has not yet been described in a patient who has demonstrated a response. Stable disease has been seen in 3/6 (50%) of patients with intermediate/high or high MGMT expression, along with one nonresponder and two cases of progressive disease. On review of the earlier cases, there were five cases with positive staining; however, the three responding cases would be better classified as showing intermediate staining. Cases with intermediate staining appear equally likely to demonstrate a response, stable disease, no response or progressive disease.

Additional work has been carried out examining the spectrum of MGMT expression across larger cohorts of pituitary tumours in patients who have not received temozolomide therapy. In an unselected group of 88 surgically treated pituitary tumours, we previously found 13% of tumours with low MGMT expression (defined as < 10%), and 28% and 59% with intermediate (10–90%) and high MGMT expression (> 90%) respectively [16]. We also found prolactinomas were significantly more likely than other subtypes to have low MGMT expression, but did not find significant differences in expression between invasive and noninvasive tumours, or between recurrent and nonrecurrent tumours. Lau et al. have recently examined a cohort of 60 pituitary tumours, comprising 30 pituitary carcinomas and 30 adenomas [73]. Their results also suggest prolactin-producing tumours have a greater propensity for low MGMT expression, with 80% of the prolactin-secreting carcinomas showing low MGMT levels. They found an overall higher frequency of low MGMT expression (54%), defined in their series as < 25% nuclear staining. There was no difference in the frequency of low expression between pituitary carcinomas and invasive adenomas (57% vs 60% respectively), and 40% of noninvasive adenomas also displayed low levels of MGMT. Other studies have reported a higher incidence of low MGMT expression amongst more aggressive subtypes of pituitary tumours. A high frequency of low MGMT expressing tumours has been reported amongst invasive Crooke’s cell adenomas and silent subtype 3 pituitary adenomas [63,75]. Widhalm et al. studied 45 patients with NFPAs and found 38% of tumours had low MGMT expression at primary surgery, using a much higher threshold of < 50% nuclear staining [71]. They observed a higher occurrence of low MGMT expression in primary surgical specimens of tumours showing subsequent tumour regrowth. In addition, primary tumours with low MGMT levels had a shorter interval to second surgery.

MGMT methylation analysis

Although the presence of MGMT promoter methylation appears to be a strong biological marker of response to temozolomide therapy in glioblastomas, this is not the case for pituitary tumours. In contrast to glioblastomas, pituitary tumours exhibit a much lower frequency of MGMT promoter methylation, found in 9–23% of tumours [16,76]. MGMT methylation analysis together with immunohistochemistry has been reported in 20 of the published pituitary cases (Table 2). Two different techniques have been employed across the combined group to detect promoter methylation, being MSP and pyrosequencing. The result of MSP analysis is qualitative, and given its high sensitivity, even a small fraction of methylated DNA will be amplified. The level of promoter methylation appears to be important in determining the amount of MGMT expression, with one study reporting a threshold of > 50% required for loss of protein expression [42,77]. Another disadvantage with MSP is that only a small number of CG dinucleotides are interrogated for the presence of methylation. However, the region amplified in the pituitary tumour cases is a recognised ‘hot spot’ of the MGMT promoter where methylation is associated with gene silencing [42]. MSP is often utilised, as it is a relatively easy and inexpensive method and can be established in most molecular laboratories. It can also be applied to paraffin-embedded tissue, although bears a significant risk of false-positive or false-negative results when DNA quality is poor. Pyrosequencing is a more sophisticated method, which provides a quantitative result, revealing the percentage of methylation at each CG dinucleotide interrogated.

MGMT methylation in pituitary tumours

As illustrated in Table 2, the presence of promoter methylation does not correlate with response to temozolomide therapy in the combined pituitary tumour cohort. There were five methylated tumours and 15 unmethylated tumours. Whilst a therapeutic response was seen in three of the methylated cases, four unmethylated cases also showed a response. Furthermore, nonresponders were seen in the context of either methylated or unmethylated tumours. Interestingly, methylated tumours in this cohort were associated with intermediate or low/negative MGMT protein expression, even though pyrosequencing revealed a very low percentage of methylation (2·6%, 8·5% and 9·8%) [65]. Six of the unmethylated tumours in this cohort were also associated with low/negative MGMT expression. We have previously found high MGMT expression in a tumour with promoter methylation, and also a significant number of unmethylated pituitary tumours associated with low MGMT expression [16]. These results strongly suggest promoter methylation is not the predominant mechanism causing low MGMT expression in pituitary tumours. The primary mechanism responsible for silencing of MGMT gene expression in pituitary tumours remains unclear. We previously failed to detect MGMT gene mutations in pituitary tumours with low MGMT expression [16]. Loss of heterozygosity at 10q26, the region containing the MGMT gene, has been described in 15% of invasive pituitary adenomas [78]; we were unable to detect LOH at 10q26 in a small number of low MGMT expressing pituitary tumours, although this warrants further investigation [16]. Alternate epigenetic mechanisms have been reported to cause MGMT gene silencing in the absence of promoter methylation [79].

Future directions

Whilst temozolomide can be used effectively as monotherapy, the next chapter in clinical exploration is to determine whether dual therapies may offer additional benefit, particularly in nonresponding cases or tumours with high MGMT expression. MGMT inhibitors, such as 06-benzylguanine or lomeguatrib, in combination with temozolomide, have had limited success in resistant gliomas [80]. Early data suggest temozolomide in combination with bevacizumab may have increased effect [81]. Optimal timing of temozolomide therapy in relation to radiotherapy needs to be determined particularly given temozolomide has demonstrated radiosensitising properties [19]. The mechanism(s) responsible for loss of MGMT expression in pituitary tumours remain to be elucidated, along with explanations for subtype differences in expression. Finally, collection of long-term follow-up data on patients treated with temozolomide is going to become very important as the use of temozolomide increases.

Conclusions

Temozolomide is the first chemotherapeutic agent to show substantial activity in the management of aggressive pituitary tumours. Overall, there has been a 60% response rate amongst the published cases to date, with an additional 13% demonstrating stable disease. Almost universally, if a response is seen, it will be evident in the first 3 months of therapy, and this appears a good time frame for a therapeutic trial. However, a formal prospective Phase II clinical trial is necessary to more accurately determine the efficacy of this agent in this tumour type.

MGMT immunohistochemistry shows promise as a predictive tool, but its role needs to be assessed further in the context of a prospective clinical trial, using standardised criteria. Data collected thus far show a high likelihood of response (76%) in the presence of low or negative MGMT immunostaining, whilst a response has not yet been shown in a patient with documented high MGMT expression. Patients with high expression of MGMT may benefit in the future from agents that inhibit MGMT activity, such as 06-benzylguanine or lomeguatrib. Assessment of MGMT promoter methylation status does not appear to be clinically useful in predicting response to therapy. MGMT promoter methylation does not occur frequently in pituitary tumours and does not appear to be the primary mechanism responsible for MGMT gene silencing.

There has been an exciting exponential growth of studies in this area; however, much work still needs to be carried out. Multicentre collaborations and formal clinical trials seem an ideal platform to expedite collection of comprehensive patient data. This will allow the development of guidelines to facilitate effective use of temozolomide in the management of aggressive pituitary tumours.

Acknowledgements

R. Lechan, W. Braund, R. Ross, W. Drake for their correspondence. Funding sources for Ann McCormack – NHMRC Medical Postgraduate Scholarship and Cancer Institute of NSW Research Scholar Award.

Address

Cancer Genetics Unit, Hormones and Cancer Group, Kolling Institute of Medical Research, Royal North Shore Hospital, Sydney, NSW 2065, Australia (A. I. McCormack); Department of Endocrinology, Oxford Centre for Diabetes, Endocrinology and Metabolism, Churchill Hospital, Oxford OX3 7LJ, UK (A. I. McCormack, J. A. H. Wass); Department of Endocrinology, William Harvey Research Institute, Barts and the London School of Medicine, London EC1A 7BE, UK (A. B. Grossman).

Ancillary