Molecular Pathology Shows P16 Methylation in Nonadenomatous Pituitaries from Patients with Cushing’S Disease

1780 Vol. 10, 1780–1788, March 1, 2004 Clinical Cancer Research

Molecular Pathology Shows p16 Methylation in Nonadenomatous Pituitaries from Patients with Cushing’s Disease

David J. Simpson,1 Anne M. McNicol,2 methylated, particularly in those cases of apparently normal David C. Murray,2 Adil Bahar,1 pituitary, is the most likely explanation for the lack of Helen E. Turner,3 John A. H. Wass,3 association between this change and loss of cognate protein in these cases. Margaret M. Esiri,4 Richard N. Clayton,1 and 1 Conclusions: To our knowledge this is the first report William E. Farrell that describes an intrinsic molecular change, namely meth- 1 Institute of Science and Technology in Medicine, School of ylation of the p16 gene CpG island, common to all three Medicine, Keele University, Stoke on Trent, Staffordshire; 2University Department of Pathology, Glasgow Royal Infirmary, histological patterns associated with Cushing’s disease. Glasgow; and 3Departments of Endocrinology and 4Neuropathology, Thus, the use of molecular pathology reveals abnormalities Radcliffe Infirmary Oxford, Oxford, United Kingdom undetected by routine pathological investigation. In cases of “apparently” normal pituitaries it is not possible to determine whether the change is associated with adenoma cells ABSTRACT “scattered” throughout the gland, albeit few in number, or Purpose: The majority of cases of Cushing’s disease are with the ancestor-clonal origin of these tumor cells. due to the presence of a corticotroph microadenoma. Less frequently no adenoma is found and histology shows either corticotroph hyperplasia, or apparently normal pituitary. In INTRODUCTION this study we have used molecular pathology to determine Cushing’s disease is a rare disorder with an annual inci- whether the tissue labeled histologically as “normal” is in- dence estimated to be between 0.7 and 2.4 cases per million (1, deed abnormal. 2). Most cases are due to the presence of a corticotroph mi- Experimental Design: Tissue from 31 corticotroph ade- croadenoma, and selective removal of the adenomatous tissue nomas and 16 nonadenomatous pituitaries were subject to results in correction of the biochemical abnormalities and clin- methylation-sensitive PCR to determine the methylation sta- ical remission of the disease (3–6). However, several reports tus of the p16 gene CpG island. The proportion of methyl- have described patients with Cushing’s disease in whom remis- ated versus unmethylated CpG island was determined using sion was achieved where no adenoma could be identified in the combined bisulphite restriction analysis. Methylation status surgically resected sample (7–9). In these cases, and in the was correlated with immunohistochemical detection of p16. absence of an adenoma, pituitary histology revealed either nor- Results: Seventeen of 31 adenomas (54.8%), 4 of 6 cases mal pituitary tissue or was interpreted as corticotroph hyperpla- of corticotroph hyperplasia, and 7 of 10 apparently normal sia. ؍ pituitaries showed p16 methylation. Ten of 14 (71%; P The evidence for a fundamental pituitary origin for tumors 0.01) adenomas and 2 of 3 cases of corticotroph hyperplasia, that arise within this gland, including those of the corticotroph which were methylated, failed to express p16 protein. How- lineage, is persuasive implying an intrinsic molecular defect ever, only 2 of 7 apparently normal pituitaries that were within these cells and has been the subject of several recent methylated failed to express p16 protein. Quantitative anal- reviews (10, 11). The assumption has been that pituitary tumors, ysis of methylation using combined bisulphite restriction in common with other tumor types, arise by de novo molecular analysis showed only unmethylated CpG islands in postmortem normal pituitaries; however, in adenomas 80–90% of changes that confer a selective growth advantage and result in a the cells within a specimen were methylated. The reverse monoclonal expansion of a single cell to produce a discrete was true for corticotroph hyperplasia and apparently nor- adenoma. However, the role of hypothalamic hormones and/or mal pituitaries where only 10–20% of the cells were meth- growth factors as initiators, facilitators, or promoters of tumor ylated. Thus, the decreased proportion of cells that were growth is also widely accepted (for recent reviews see Refs. 10, 12, 13). Indeed, in those cases where disease is associated with abnormal trophic and/or secretory activity, for example in nonadenomatous tissue associated with Cushing’s disease, the role of hypothalamic or rarely ectopic factors would seem intuitively Received 7/31/03; revised 11/17/03; accepted 11/19/03. attractive but lacking formal proof. Moreover, recent investiga- The costs of publication of this article were defrayed in part by the tions support this view and have challenged the concept of payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to invariant monoclonality (14, 15). These authors have proposed indicate this fact. that a monoclonal expansion might, in some cases, arise on a Requests for reprints: William E. Farrell, Centre for Cell and Molec- background of cell subtype-specific hyperplasia consistent with ular Medicine, School of Postgraduate Medicine, Keele University, involvement of extra- or intrapituitary factors. Indeed, animal North Staffordshire Hospital, Stoke-on-Trent ST4 7QB, United King- dom. Phone: 44-1782-555225; Fax: 44-1782-747-319; E-mail: models show this to be the case (16–18). The nearest human [email protected]. counterpart to the animal models is Cushing’s disease where

apparently normal pituitary tissue, hyperplasia, or adenoma may Table 1 Methylation and expression status of p16 in Cushing’s be associated with the disease. disease adenomatous and nonadenomatous pituitaries Methylation-associated gene silencing is a frequent finding Methylation IHCa in numerous tumor types (reviewed in Ref. 19) including those Patient Tumor grade status CDKN2A status p16 of pituitary origin. We showed recently methylation of the 11 MϪ tumor suppressor genes CDKN2A (p16) and RB1 CpG islands 2 1 U N/A that was associated significantly with loss of their cognate 3 1 U N/A 41 Mϩ proteins in nonfunctioning pituitary tumors and somatotrophi- 51 Mϩ nomas, respectively (20, 21). In nonfunctioning tumors p16- 6 1 M N/A associated methylation occurred early in pituitary tumorigenesis 71 Uϩ ϩ (20), and in other tumor types this epigenetic change has been 81 U 91 MϪ described in the preceding preneoplastic tissue (22). In addition, 10 1 U ϩ a mechanistic role for p16 in pituitary tumorigenesis is sug- 11 1 U N/A gested by the findings that reintroduction of this gene into a 12 1 U ϩ Ϫ corticotroph cell line AtT20, in which the endogenous gene is 13 1 M 14 1 M Ϫ homozygous deleted, inhibits cells proliferation and is associ- 15 1 M ϩ ated with a G1 arrest (23). Taken together these findings 16 1 M N/A prompted us to investigate methylation-associated p16 gene 17 1 U N/A Ϫ silencing in adenomas and nonadenomatous tissue from patients 18 1 M 19 1 M Ϫ with Cushing’s disease, to determine whether molecular pathol- 20 2 U N/A ogy would reveal pathogenetic changes, especially in non- 21 2 U N/A adenomatous tissues, which cannot be revealed immunohisto- 22 2 U ϩ ϩ logically. 23 2 M 24 2 M Ϫ 25 2 M Ϫ 26 2 U N/A MATERIALS AND METHODS 27 3 U ϩ Patient Characteristics. All of the patients had pituitary 28 4 M Ϫ dependent Cushing’s disease as defined by: (a) clinical features; 29 4 U N/A 30 4 M Ϫ (b) sustained increased urine free cortisol excretion; (c) loss of 31 4 M N/A plasma cortisol diurnal rhythm; (d) failure to suppress plasma 32 Hyperplasia U ϩ cortisol with low-dose dexamethasone (0.5 mg dexamethasone 33 Hyperplasia M Ϫ Ϫ every 6 h ϫ48 h); and (e) Ͼ50% suppression of plasma cortisol/ 34 Hyperplasia M 35 Hyperplasia M NI adrenocorticorticotropic hormone (ACTH) with high-dose dex- 36 Hyperplasia U ϩ amethasone (2 mg dexamethasone every 6 h ϫ48 h). Brief 37 Hyperplasia M ϩ patient details are shown in Table 1. 38 Normal M ϩ ϩ Pituitary imaging was carried out by computed tomography 39 Normal M 40 Normal M ϩ or magnetic resonance imaging scan and the designation of 41 Normal U ϩ adenoma size used a modified Hardy classification (24) where 42 Normal U ϩ grade 1 microadenomas were Ͻ10 mm diameter; grade 2 mac- 43 Normal M ϩ Ϫ roadenomas were Ͼ10 mm diameter but without extra/parasel- 44 Normal M 45 Normal M Ϫ lar extension; grade 3 macroadenomas were as grade 2 with 46 Normal M ϩ extra/parasellar extension; and grade 4 metastatic tumors were 47 Normal U N/A within the central nervous system. The majority of patients were 1–6 Postmortem normal (X6) U ϩ young women. Clinical and biochemical status was assessed a IHC, immunohistochemistry; M, methylated; U, unmethylated; within 2–3 months postoperatively and remission defined as ϩ, IHC positive; Ϫ, IHC negative; NI, noninformative/not interpretable; normalization of cortisol production (normal urine free cortisol NA, insufficient material for analysis; Grades: 1, microadenoma; 2, intrasellar macroadenoma; 3, extrasellar macroadenoma; 4, carcinoma. excretion, normal plasma cortisol diurnal rhythm, and dexamethasone suppressibility; low dose). The patients were from two centers, Oxford and Stoke-on- Trent, United Kingdom, where surgery was performed by a Pituitary Histology. Pituitary tissue removed at opera- single experienced pituitary surgeon in each center. The opinion tion was examined by one experienced neuropathologist of the surgeon regarding tumor visualization (though not size) (M. M. E.) and was subject to systematic examination. Adjacent was recorded in most instances. As in previous reports there was sections were stained with: (a) H&E; (b) Gordon and Sweet’s poor correlation between preoperative tumor visualization and reticulin stain; and (c) polyclonal antibodies to ACTH, PRL, what was observed at operation (reviewed in Ref. 25). Similarly, ␤ human chorionic gonadotropin, ␤ thyrotropin, ␤ follicle- surgical opinion of abnormal pituitary tissue correlated poorly stimulating hormone, and ␤ luteinizing hormone (Dako). ␣ with histological reports (vis-a`-vis: normal, hyperplasia, and Human chorionic gonadotropin, (Biogenex) and growth hor- adenoma). mone (kind gift from Dr. Andrew F. Parlow, National Hormone

Table 2 PCR primers for sodium bisulphite converted DNA Oligonucleotide set Sense 5Ј-3Ј Antisense 5Ј-3Ј Prod size (bp) p16M TTATTAGAGGGTGGGGCGGATCGCa GACCCCGAACCGCGACCGTAAa 150 p16U TTATTAGAGGGTGGGGTGGATTGTa CAACCCCAAACCACAACCATAAa 151 RB1M GGGAGTTTCGCGGACGTGACa ACGTCGAAACACGCCCCGa 172 RB1U GGGAGTTTTGTGGATGTGATa ACATCAAAACACACCCCCAa 172 p16COBRA GGGAGTAGTATGGAGTTb CCCTCTACCCACCTAAATb 147 a MS-PCR, methylation-sensitive PCR. b COBRA, combined bisulfite restriction analysis.

and Pituitary Program, University of California Los Angeles, each primer), 1.5 mM MgCl2, and 200 ng of modified template Los Angeles, CA). Histological classification was as follows: DNA. For the p16 and RB1 genes CpG islands, reactions were (a) corticotroph adenoma, uniform ACTH-immunoreactive cells hot started (96°C for 5 min) before the addition of 1 unit with absence of, or fragmented, reticulin staining; (b) cortico- TaqDNA polymerase, and 200 ␮M each of dATP, dGTP, dTTP, troph hyperplasia, expanded clusters comprising mainly ACTH- and dCTP. PCR was carried out for 30 cycles (55°C for 30 s, immunoreactive cells and that distend the reticulin baskets; and 72°C for 30 s, and 94°C for 1 min). (c) apparently normal pituitary tissue, ACTH-immunoreactive PCR products were run on 8% nondenaturing polyacryl- cells showing a similar frequency and distribution to that seen in amide gels, fixed in 10% methylated spirit/0.5% acetic acid for postmortem normal pituitary gland, occupying normal-sized 6 min, and then incubated in 0.1% aqueous silver nitrate for 15 reticulin baskets. min. After two brief washes in distilled water, products were Tissue and DNA Preparation. Ten 5 ␮m unstained sec- visualized by development in 1.5% sodium hydroxide/0.1% tions were taken from the pituitary tissue described above to- formaldehyde. gether with 6 postmortem normal pituitaries processed in the Combined Bisulphite Restriction Analysis. To quanti- same manner and obtained within 12 h of death. Before molec- tate methylated and unmethylated p16 CpG island sequences ular and immunohistochemical analysis histological classifica- contained in DNA extracted from archival specimens we used tion was confirmed in a preceding and subsequent section to combined bisulphite restriction analysis (COBRA; Ref. 28). ensure analysis corresponding to the histology described. For Because the oligonucleotides were designed to specifically am- the molecular analysis sections were subject to microdissection plify the p16 CpG island (see Table 2) and contain no CpG and DNA extraction as described previously (20), and stored dinucleotides, both methylated and unmethylated sequences are at 4°C. coamplified by a single primer pair after bisulphite modifica- Sodium Bisulphite Modification. DNA (ϳ2 ␮g) was tion. Amplification is achieved at an annealing temperature of denatured with NaOH (final concentration 0.2 M) in a total 60°C using PCR parameters as described above except that PCR volume of 30 ␮l for 10 min at 37°C. Urea/sodium bisulphite cycles were reduced to 24 to ensure amplification within the solution (final concentration 0.5 mM hydroquinone, 5.36 M urea, linear range. and 3.44 M sodium bisulphite) at pH 5, freshly prepared, was The bisulphite conversion of the DNA results in the added, mixed, and incubated at 55°C, in the absence of light for methylation-dependent retention of a BstUI (CGCG) restriction 4 h (26). site within the p16 CpG island. Five-␮l of each PCR product DNA samples were purified using the GeneCleanII purifi- was digested with the BstUI restriction enzyme and the resulting cation kit according to the manufacturer’s protocol (Bio 101 Vista 101, San Diego, CA) and eluted in 50 ␮l of water. Modification was completed by desulphonation with NaOH (final concentration 0.3 M) for 15 min at 37°C. DNA was ethanol Table 3 Association between methylation and expression of p16 precipitated and resuspended in 10 ␮l of water. Samples were protein in Cushing’s disease stored at Ϫ20°C. Unmethylated Methylated Methylation Status of CDKN2A/p16 and RB1 by Meth- Adenomas ylation-Sensitive PCR (MS-PCR). Oligonucleotides specific Total cohort (n ϭ 31) 14/31 (45.2%) 17/31 (54.8%) for the p16 CpG island were as described previously (27). The p16 expressed 6/6 (100%) 4/14 (29%) specificity of the methylation target (p16) was determined by p16 not expressed 0/6 (0%) 10/14 (71%)* Hyperplasia assessing the methylation status of the RB1 gene CpG island as Total cohort (n ϭ 6) 2/6 (ϳ33%) 4/6 (ϳ66%) described previously (21). Oligonucleotide sequences are shown in p16 expressed 2/2 1/3 Table 2. To confirm the PCR amplification of methylated p16/RB1 p16 not expressed 0/2 2/3 CpG islands after bisulphite modification we in vitro methylated Apparently normal Total cohort (n ϭ 10) 3/10 (ϳ30%) 7/10 (ϳ70%) genomic DNA with the CpG methylase enzyme SssI. This DNA p16 expressed 2/2 5/7 was then subjected to urea/sodium bisulphite modification as de- p16 not expressed 0/2 2/7 scribed above and served as a positive control. Postmortem normal (n ϭ 6) 6/6 PCR reactions contained oligonucleotide sets specific for p16 expressed 6/6 methylated or unmethylated p16/RB1 CpG islands (2 pmol of * P ϭ 0.01.

fragments (147, 104, and 43 bp) run on 8% polyacrylamide gels Representative examples of the histological findings in and visualized as described above (see Fig. 3). The percentage these three categories associated with Cushing’s disease to- of fully methylated BstUI sites in the DNA sample can then be gether with postmortem normal pituitary are shown in Fig. 1. calculated from the ratio between the BstUI cleaved (methyl- There were no significant differences in the H&E and reticulin ated) and the undigested (unmethylated) PCR product relative to staining patterns between postmortem normal pituitary (Fig. 1, the product generated in the same reaction not subject to restric- A–C) and apparently normal pituitary associated with Cushing’s tion digest. Quantitation was carried out using a GS-700 scan- disease (Fig. 1, J–L). In both cases the H&E and reticulin ning densitometer (Bio-Rad, High Wycomb, United Kingdom). staining shows normal pituitary morphology with intact reticulin p16 Immunostaining. Sections were rehydrated and pre- baskets surrounding well-defined cell nests. All of the postmor- treated by combined microwaving and pressure-cooking for 8 tem normal pituitaries were immunopositive for p16 protein min at full power in a 600-W microwave oven. They were then showing clear nuclear positivity in 50–60% of cells with low stained using a labeled streptavidin-biotin system, with a pri- level cytoplasmic staining, which was considered irrelevant. Of mary mouse monoclonal antibody to p16 (CDKN2A; Clone 8 evaluable cases of apparently normal pituitary 6 stained pos- G175–1239; PharMingen, Cambridge Bioscience, Cambridge, itively for p16 protein (Table 1). In the representative example United Kingdom). The secondary antibody was biotinylated of corticotroph hyperplasia (Fig. 1, G–I) the H&E staining antirabbit/mouse immunoglobulins, followed by streptavidin- shows enlarged cell nests with increased numbers of hyper- horseradish peroxidase (Large volume LSAB kit; Dako Ltd, trophic corticotrophs interspersed with smaller numbers of non- Bucks, United Kingdom), with diaminobenzidine as chromo- corticotroph cells. The reticulin baskets were expanded. Of 5 gen. Negative controls were omission of the primary antibody. evaluable cases of corticotroph hyperplasia 3 stained positively Sections of a malignant melanoma were used as positive con- for p16 protein and 2 were negative (Table 1). As with normals, trols. Adjacent “mirror” sections were also compared with de- only a subpopulation of cells were positive. In the adenomatous termine corticotroph (ACTH positive) expression of p16 relative tissue (Fig. 1, D–F), H&E staining shows a complete lack of to methylation status. organized structure, with loss of the normal reticulin and only Tumor immunopositivity was classified using criteria iden- small irregular fragments between tumor cells. In the example tical to that described by Geradts et al. (29). According to their shown p16 staining was negative. Overall, of 20 evaluable criteria and our own investigations of p16 expression in pituitary adenomas 10 (50%) did not express p16 protein (summarized in tumors, only nuclear staining was defined as positive (20, 29, Table 1). In some cases insufficient pituitary tissue from patients 30). Where no staining of nuclei was observed in normal pitu- with Cushing’s disease was available for immunohistochemistry itary tissue surrounding/within the tumor section as an inbuilt (IHC) analysis or was not interpretable (see Table 1) control, the case was deemed inconclusive. Methylation Status of the p16 Gene. MS-PCR analysis Statistical Analysis. Statistical analysis was performed revealed 17 of 31 (54.8%) corticotroph adenomas to harbor using the Stata v.5 statistical package (Stata Corp. TX). ␹2 tests methylation of the p16 gene CpG island; however, all 6 of the were used to compare variables. Significance was taken at the postmortem normal pituitaries were unmethylated (Fig. 2; sum- 5% level. marized in Table 1). In 3 cases of adenoma, where the p16 gene CpG was found to be methylated by MS-PCR, sufficient juxta- posed normal pituitary was available for analysis after micro- RESULTS dissection. In none of these cases did we detect methylation of Morphological and Immunohistochemical Findings. this CpG island. MS-PCR analysis of Cushing’s associated We examined pituitary tissue derived from 47 patients with corticotroph hyperplasia showed 4 of 6 to be methylated proven Cushing’s disease. The total cohort comprised 31 corti- (ϳ66%) and in apparently normal pituitary associated with this cotroph adenomas and an additional 16 in which no adenoma- disease 7 of 10 (ϳ70%) were methylated (Fig. 2; summarized in tous tissue was identified either at surgery and/or during routine Table 1). Of the total cohort investigated for methylation of histological examination (Table 1). In addition as controls we CDKN2A 20 of 47 were analyzed for methylation of the RB1 included 6 postmortem normal pituitaries from patients with no gene CpG island by MS-PCR as described previously (21). history of endocrine disease or steroid therapy. As an outcome None of the specimens analyzed harbored methylation of RB1 measure postsurgical remission rate was not significantly dif- irrespective of p16 methylation status (data available on re- ferent (P ϭ 0.5) between those patients in whom an adenoma quest). was identified and those in whom disease was associated with In the corticotroph adenomas that were methylated MS- nonadenomatous pituitaries (Table 1). The nonadenomatous PCR only detected sequence corresponding to a methylated p16 specimens were subject to systematic examination of all of the CpG island and did not detect unmethylated sequence. In con- available pituitary tissue to confirm initial observations. This tradistinction, the postmortem normal pituitaries only displayed examination confirmed the absence of adenomatous tissue and an unmethylated CpG island. However, in the MS-PCR analysis revealed in 6 of the 16 cases evidence of corticotroph hyperpla- of corticotroph hyperplasia and apparently normal pituitaries, sia and in the remaining 10, apparently normal pituitary tissue. where we detected methylation by MS-PCR, we also codetected The specimens of nonadenomatous pituitary associated with unmethylated sequence from the same specimen with primers Cushing’s disease all showed various degrees of Crooke’s hy- specific for unmethylated DNA (Fig. 2). Thus, in these cases, aline change. This change, reflecting high circulating glucocor- these results suggest that not all of the cells in the specimens ticoid levels, was confined to the corticotroph cell population harbored methylation of the p16 gene. (data not shown). In the experiments described thus far, we were unable to

Fig. 1 Immunohistochemical analysis of postmortem normal pituitary and pituitary tissue associated with Cushing’s disease. A–C, normal pituitary obtained at autopsy. D–F, pituitary adenoma in Cushing’s disease. This case showed methylation of CDKN2A. G–I, corticotroph hyperplasia in Cushing’s disease. This case shows methylation of CDKN2A. J–L, normal pituitary in Cushing’s disease. This case shows methylation of CDKN2A. A, D, G, and J, medium power view of H&E stained slides to demonstrate overall morphology. Both normal glands (A and J) show a similar pattern. There is complete loss of organized structure in the adenoma (D). There are obvious large cell nests in the hyperplastic gland (G). B, E, H, and K, medium power view of reticulin stained slides. This highlights the organization of the tissue by identifying the connective tissue component. In the normal glands (B and K) the cells are arranged in small groups surrounded by connective tissue. In the adenoma (E), there are only small irregular fragments of reticulin between tumor cells, the only organization being seen in relation to blood vessels. In the hyperplastic gland (H) this stain emphasizes the large irregular cell nests. C, F, I, and L, high-power view of immunostaining for p16 protein, to allow better appreciation of nuclear staining. Positive cells have brown nuclei, whereas negative nuclei show only the blue hematoxylin counterstain. Cytoplasmic staining is viewed as nonspecific (20, 29, 30). In both normal glands only a subpopulation of nuclei show positivity. The adenoma (F) is completely negative. The hyperplastic nests (I) show a similar pattern to the normal with a subpopulation of positive nuclei.

exclude the possibility that methylation might represent a sec- Association between Methylation and Absence of p16 ondary change due to the cortisol excess that characterizes this Protein. We and others have shown previously an association disease. To additionally explore this possibility we used MS- between methylation of the p16 gene and absence of cognate PCR for the p16 gene CpG island to determine methylation of protein in pituitary tumors (20, 30–33). Therefore, we looked pituitary glands obtained at autopsy from 2 patients on long- for such associations in the pituitary tissue associated with term glucocorticoid therapy and from 1 patient with ectopic Cushing’s disease. In the corticotroph adenomas sufficient tu- ACTH syndrome. All 3 showed Crooke’s hyaline change. None mor was available from 20 of 31 adenomas for p16 IHC analysis of the 3 cases showed methylation of this gene CpG island (data (Table 2 and summarized in Table 1). Ten of 14 (71%) adeno- not shown). mas that were methylated did not express p16, and the remain-

Fig. 2 Methylation-sensitive PCR (MS-PCR) of DNA extracted from pituitary tissue associated with Cushing’s disease Shown (from left to right) are representative examples of postmortem normal pituitary, corticotroph adenoma, corticotroph hyperplasia, and apparently normal pituitary. An amplicon in the lane denoted M or U indicates identification of a methylated or unmethylated CDKN2A/p16 CpG island, respectively. Because the products of these reactions are generated using different sets of primers they cannot be compared quantitatively. Specimen numbers are given above each gel. In vitro methylated (IVM) is an in vitro methylated control. Postmortem normal and adenoma show either a band representing either unmethylated (U) or methylated (M) sequence respectively. In the specimens representing either hyperplasia or apparent normal amplicons indicative of methylated and unmethylated CpG island are revealed.

ing 4 (29%) showed nuclear positivity for this protein. The 6 positive methylation signal, between 80 and 90% of the speci- adenomas that were unmethylated all expressed p16 protein. men was methylated and the remainder unmethylated. None of Methylation was significantly associated with gene silencing in the adenomas in which MS-PCR predicted an unmethylated adenomas (P ϭ 0.01). CpG island were methylated by the COBRA technique. In Sufficient material from 5 of 6 cases of corticotroph hy- apparently normal pituitary and corticotroph hyperplasia the perplasia was available for p16 IHC analysis. Of 3 samples that ratio of methylated versus unmethylated CpG island was re- were methylated 2 did not express p16 protein and 1 did (Table versed from that seen in adenomas. In these cases, between 80 2 and summarized in Table 1). The 2 specimens that were and 90% of individual specimens were found to be unmethyl- unmethylated at the sites examined within this CpG island ated and the remainder methylated. Only those cases found to expressed p16 protein. The few specimens available for analysis give a positive display for methylation by MS-PCR were found precluded statistical analysis. to be methylated by the COBRA methodology. Thus, in those In those cases of apparently normal pituitary associated cases where methylation is not associated with loss of p16 with Cushing’s disease 8 of 10 were available for IHC analysis. protein a likely explanation is that a significant proportion of Of 7 specimens that were methylated 2 failed to express p16 cells within the specimen do not harbor this change, at least at protein; however, the remaining 5 stained positively for p16 in the sites investigated by this technique. a proportion of cells (Table 2 and summarized in Table 1). Two specimens unmethylated by MS-PCR expressed p16 protein. The number of specimens available again precluded statistical DISCUSSION analysis; however, the data would suggest that methylation is In this study we have shown that molecular analysis can not associated with gene silencing in this Cushing’s associated detect changes intrinsic to pituitary tissue derived from patients histology. with ACTH-dependent Cushing’s disease even where conven- COBRA. We reasoned that our failure to detect a clear tional histopathology appears normal. The patients showed all association between methylation and loss of p16 protein expres- three histological patterns reportedly associated with Cushing’s sion in apparently normal pituitary associated with Cushing’s disease, namely corticotroph adenoma, corticotroph hyperpla- disease might reflect the proportion of cells harboring this sia, and apparently normal pituitary (7, 8, 34). MS-PCR analysis epigenetic aberration versus those that did not. To determine the of the p16 gene CpG island showed this to be frequently proportion of cells within a specimen that were methylated at methylated irrespective of the histological findings. the CDKN2A/p16 CpG island we used the quantitative COBRA This particular tumor suppressor gene was chosen because for simultaneous detection of a methylated and unmethylated of its known association with pituitary tumorigenesis (reviewed

CpG island. Fig. 3 shows this analysis in representative speci- in 12) as well as being a regulatory component of the RB1 G1-S mens. In all 6 of the postmortem pituitary tissues we only cell cycle pathway. Although RB1 knockout mice develop neu- detected sequence corresponding to an unmethylated CpG is- rointermediate lobe tumors of the corticotroph lineage, RB1 land. From corticotroph adenomas, where MS-PCR gave a gene mutations are uncommon in human pituitary tumors. How-

Fig. 3 Combined bisulphite restriction analysis (COBRA). Representative examples of specimens examined for coamplification of methylated and unmethylated CDKN2A/p16 CpG island are shown. Specimens are as describe in Fig. 2. To the left of the gels the position of the unmethylated (147) and methylated (104 bp) amplicons are indicated. To aid clarity the 43 bp fragment generated by cleavage of methylated sequence is not shown. For individual specimens the left lane represents amplification of methylated and unmethylated sequence before digestion (U). The right lane represents the same PCR product after digestion with BstU1 (C). In postmortem pituitary only an unmethylated CpG island is present. In the adenomas 10–20% was unmethylated and 80–90% methylated. The reverse was found in apparently normal and hyperplasia where 10–20% was methylated and 80–90% unmethylated. For each specimen the ratio of BstU1 cleaved (methylated) versus BstU1 resistant (unmethylated) was calculated relative to the product generated in the same reaction that was not subject to restriction digest.

ever, loss of pRB has been demonstrated frequently in human sion, the methylation pattern and density is faithfully replicated somatotrophinomas as a consequence of or associated with in each of the progeny cells. However, in cases of corticotroph methylation of its CpG island (21). Accordingly, we sought hyperplasia and apparently normal pituitary, associated with specificity for p16 gene methylation by examining RB1 meth- Cushing’s disease, this reflects a predominantly polyclonal popu- ylation and found in all of the instances this to be unmethylated. lation. Thus, the proportion of cells that harbor this epigenetic This suggests that methylation did not simply reflect nonspecific change combined with the methylation density within individual methylase activity within a specimen. It also suggests that in cells will impinge on expression status as assessed by immuno- corticotroph adenomas, as in somatotrophinomas and nonfunc- histochemistry. tional tumors, methylation-associated loss of p16 and pRb are There is increasing evidence that corticotroph hyperplasia mutually exclusive events (30). and a finding of apparently normal pituitary despite a thorough We now show a similar frequency of p16 methylation in histological analysis are associated with Cushing’s disease in corticotrophinomas (55%) to that seen in nonfunctioning tumors between 15% and 30% of cases, and removal of apparently (70%) that is also significantly associated with loss of cognate normal pituitary tissue results in clinical and biochemical cure protein by IHC analysis. However, in those patients in whom a of Cushing’s disease (7, 8, 36–38). The key issue is, therefore, discrete adenoma could not be identified, methylation of this whether this apparently normal pituitary tissue harbors patho- CpG island, whilst frequent, was not associated with complete logical change, which cannot be revealed by conventional his- loss of p16 protein. Furthermore, and in contrast to corticotroph tology and/or immunohistochemistry. Our studies provide adenomas, MS-PCR showed coamplification of sequence that strong support for the fact that what is removed in cases deemed represented an unmethylated CpG island. to represent nonadenomatous tissue is indeed abnormal. In other Several studies have shown that extinction of gene expression, tumor types CpG island methylation has been described in through CpG island methylation, is density dependent (35). How- preneoplastic tissue (22), and our data are compatible with these ever, the MS-PCR technique does not reflect methylation density findings. However, and perhaps more contentious, is the defi- but reports on the status of specific residues within the CpG island nition of true preadenomatous tissues. Thus, identification of (27). We reasoned that the lack of association between p16 meth- p16 methylation in nonadenomatous pituitary tissue may simply ylation and loss of protein expression in tissue that represented reflect identification of cells that are actually “adenomatous” either hyperplasia or apparently normal pituitary that only a pro- albeit very few and “scattered” throughout the gland. Our find- portion of cells within a specimen harbored this epigenetic change. ings are consistent with this explanation because the MS-PCR Indeed, quantitative COBRA analysis showed this to be the case, technique is reported to reliably detect 0.1% methylated DNA because only 10–20% of the cells within the specimen harbor a present in an otherwise unmethylated sample (27). However, methylated p16 CpG island. The reverse was true in adenomas and equally plausible, is that we have detected changes in cells where, and in marked contrast, 80–90% of cells were methylated. that are the ancestral-clonal origins of these tumor cells, where, Thus, in the adenomas, presumed to reflect a monoclonal expan- in this case, p16 methylation does indeed represent one of the

earliest identified changes preceding adenoma formation. Which 11. Dahia, P. L., and Grossman, A. B. The molecular pathogenesis of of these two explanations is correct we are at present unable to corticotroph tumors. Endocr. Rev., 20: 136–155, 1999. determine. Regardless of the precise mechanistic interpretation 12. Farrell, W. E., and Clayton, R. N. Molecular pathogenesis of of p16 methylation in the context of Cushing’s disease the pituitary tumors. Front. Neuroendocrinol., 21: 174–198, 2000. molecular pathological approach described herein identifies ab- 13. Clayton, R. N., and Farrell, W. E. Clonality of pituitary tumours: normal cells, irrespective of histological findings. Given the more complicated than initially envisaged? Brain Pathol., 3: 313–327, 2001. high frequency of this change in Cushing’s associated nonade- ϳ 14. Levy, A. Is monoclonality in pituitary adenomas synonymous with nomatous pituitaries ( 70%) this relatively simple technique neoplasia? Clin. Endocrinol., 52: 393–407, 2000. could provide a useful addition to conventional investigations in 15. Clayton, R. N., Pfeifer, M., Atkinson, A. B., Belchetz, P., Wass, circumstances of doubtful pituitary histopathology. J. A., Kyrodimou, E., Vanderpump, M., Simpson, D., Bicknell, J., and To our knowledge this is the first report that shows a Farrell, W. E. Different patterns of allelic loss (loss of heterozygosity) molecular pathological change, namely, methylation of the p16 in recurrent human pituitary tumors provide evidence for multiclonal gene CpG island common to the three histological patterns of origins. Clin. Cancer Res., 6: 3973–3982, 2000. corticotrophs in the pituitary in Cushing’s disease. Our data 16. Asa, S. L., Kovacs, K., Stefaneanu, L., Horvath, E., Billestrup, N., Gonzalez-Manchon, C., and Vale, W. Pituitary mammosomatotroph strongly suggest that there is an intrinsic abnormality within the adenomas develop in old mice transgenic for growth hormone-releasing pituitary tissue removed from patients with Cushing’s disease hormone. Proc. Soc. Exp. Biol. Med., 193: 232–235, 1990. even when no adenoma is identified, as witnessed by the meth- 17. Asa S. L. Kelly, M. A., Grandy, D. K., and Low, M. J. Pituitary ylation of p16. This does not seem to be a secondary effect of lactotroph adenomas develop after prolonged lactotroph hyperplasia in the disease state as it was not identified in association with dopamine D2 receptor-deficient mice. Endocrinology, 140: 5348–355, glucocorticoid therapy or in ectopic ACTH syndrome. Methyl- 1999. ation appears to affect only a subpopulation of cells within these 18. Heaney, A. P., Horwitz, G. A., Wang, Z., Singson, R., and Melmed, glands. This might reflect the presence of a small clone of S. Early involvement of estrogen-induced pituitary tumor transforming gene and fibroblast growth factor expression in prolactinoma pathogen- neoplastic corticotrophs not yet big enough to have the recog- esis. Nat. Med., 5: 1317–1321, 1999. nized architecture of an adenoma or the ancestor-clonal origins 19. Baylin, S. B., Herman, J. G., Graff, J. R., Vertino, P. M., and Issa, of these tumor cells. J. P. Alterations in DNA methylation: a fundamental aspect of neoplasia. Adv. Cancer Res., 72: 141–196, 1998. ACKNOWLEDGMENTS 20. Simpson, D. J., Bicknell, J. E., McNicol, A. M., Clayton, R. N., and We thank Nicola Sullivan for expert technical assistance. Farrell, W. E. Hypermethylation of the p16/CDKN2A/MTSI gene and loss of protein expression is associated with nonfunctional pituitary adenomas but not somatotrophinomas. Genes Chromosomes Cancer, REFERENCES 24: 328–336, 1999. 1. Extabe, J., and Vazquez, J. A. Morbidity and Mortality in Cushings- 21. Simpson, D. J., Hibberts, N. A., McNicol, A. M., Clayton, R. N., disease - an epidemiologic approach. Clin. Endocrinol., 40: 479–484, and Farrell, W. E. Loss of pRb expression in pituitary adenomas is 1994. associated with methylation of the RB1 CpG island. Cancer Res., 60: 2. Bochicchio, D., Losa, M., and Buchfelder, M. Factors influencing the 1211–1216, 2000. immediate and late outcome of Cushing’s disease treated by transsphe- 22. Belinsky, S. A., Nikula, K. J., Palmisano, W. A., Michels, R., noidal surgery: a retrospective study by the European Cushing’s Disease Saccomanno, G., Gabrielson, E., Baylin, S. B., and Herman, J. G. Survey Group. J. Clin. Endocrinol. Metab., 80: 3114–3120, 1995. Aberrant methylation of p16(INK4a) is an early event in lung cancer and 3. Salassa, R. M., Laws, E. R., Jr., Carpenter, P. C., and Northcutt, R. C. a potential biomarker for early diagnosis. Proc. Natl. Acad. Sci. USA, Transsphenoidal removal of pituitary microadenoma in Cushing’s dis- 95: 11891–11896, 1998. ease. Mayo Clin. Proc., 53: 24–28, 1978. 23. Frost, S. J., Simpson, D. J., Clayton, R. N., and Farrell, W. E. 4. Tyrrell, J. B., Brooks, R. M., Fitzgerald, P. A., Cofoid, P. B., Transfection of an inducible p16/CDKN2A construct mediates reversi-

Forsham, P. H., and Wilson, C. B. Cushing’s disease. Selective trans- ble growth inhibition and G1 arrest in the AtT20 pituitary tumour cell sphenoidal resection of pituitary microadenomas. N. Engl. J. Med., 298: line. Mol. Endocrinol., 13: 1801–1810, 1999. 753–758, 1978. 24. Bates, A. S., Farrell, W. E., Bicknell, E. J., McNicol, A. M., Talbot, 5. Bigos, S. T., Somma, M., Rasio, E., Eastman, R. C., Lanthier, A., A. J., Broome, J. C., Perrett, C. W., Thakker, R. V., and Clayton, R. N. Johnston, H. H., and Hardy, J. Cushing’s disease: management by Allelic deletion in pituitary adenomas reflects aggressive biological transsphenoidal pituitary microsurgery. J. Clin. Endocrinol. Metab., 50: activity and has potential value as a prognostic marker. J. Clin. Endo- 348–354, 1980. crinol. Metab., 82: 818–824, 1997. 6. Boggan, J. E., Tyrrell, J. B., and Wilson, C. B. Transsphenoidal 25. McCance, D. R., Besser, M., and Atkinson, A. B. Assessment of microsurgical management of Cushing’s disease. Report of 100 cases. cure after transsphenoidal surgery for Cushing’s disease. Clin. Endocri- J. Neurosurg., 59: 195–200, 1983. nol. (Oxf)., 44: 1–6, 1996. 7. Burke, C. W., Adams, C. B., Esiri, M. M., Morris, C., and Bevan, 26. Paulin, R., Grigg, G. W., Davey, M. W., Piper, A. A. Urea J. S. Transsphenoidal surgery for Cushing’s disease: does what is improves efficiency of bisulphite-mediated sequencing of 5Ј-methylcy- removed determine the endocrine outcome? Clin. Endocrinol. (Oxf)., tosine in genomic DNA. Nucleic Acids Res., 26: 5009–5010, 1998. 33: 525–537, 1990. 27. Herman, J. G., Graff, J. R., Myohanen, S., Nelkin, B. D., Baylin, 8. Kruse, A., Klinken, L., Holck, S., and Lindholm, J. Pituitary histol- S. B. Methylation-specific PCR: a novel PCR assay for methylation ogy in Cushing’s disease. Clin. Endocrinol., 37: 254–259, 1992. status of CpG islands. Proc. Natl. Acad. Sci. USA, 93: 9821–9826, 9. Yap, L. B., Turner, H. E., Adams, C. B., and Wass, J. A. Undetect- 1996. able postoperative cortisol does not always predict long-term remission 28. Xiong Z., and Laird, P. W. COBRA: a sensitive and quantitative in Cushing’s disease: a single centre audit. Clin. Endocrinol. (Oxf.), 56: DNA methylation assay. Nucleic Acids Res., 15: 2532–2534, 1996. 25–28, 2002. 29. Geradts, J., Kratzke, R. A., Niehans, G. A., and Lincoln, C. E. 10. Asa, S. L, and Ezzat, S. The cytogenesis and pathogenesis of Immunohistochemical detection of the cyclin-dependent kinase inhibitor pituitary adenomas. Endocr. Rev., 19: 798–827, 1998. 2/multiple tumor suppressor gene 1 (CDKN2/MTS1) product

p16INK4A in archival human solid tumors: correlation with retinoblas- tumors by hypermethylation is more common in null cell adenomas. toma protein expression. Cancer Res., 55: 6006–6011, 1995. Endocr. Pathol., 12: 281–289, 2002. 30. Simpson, D. J., Frost, S. J., Bicknell, J. E., Broome, J. C., McNicol, 34. Bloodworth, J. M. B., Jr. Assessment of the pituitary hyperplasia/ A. M., Clayton, R. N., and Farrell, W. E. Aberrant expression of G(1)/S neoplasia interface. Path. Res. Pract., 183: 626–630, 1988. regulators is a frequent event in sporadic pituitary adenomas. Carcino- 35. Jones, P. A. DNA methylation errors and cancer. Cancer Res., 56: genesis (Lond.), 22: 1149–1154, 2001. 2563–2467, 1996. 31. Woloschak, M., Yu, A., and Post, K. D. Frequent inactivation of the p16 gene in human pituitary tumors by gene methylation. Mol. Car- 36. Horvarth E. Pituitary hyperplasia. Path. Res. Pract., 183: 631–635, cinog., 19: 221–224, 1997. 1988. 32. Seemann, N., Kuhn, D., Wrocklage, C., Keyvani, K., Hackl, W., 37. McCance, D. R., Gordon, D. S., Fannin, T. F., Hadden, D. R., Buchfelder, M., Fahlbusch, R., and Paulus, W. CDKN2A/p16 inactiva- Kennedy, L., Sheridan, B., and Atkinson, A. B. Assessment of endo- tion is related to pituitary adenoma type and size. J. Pathol., 193: crine function after transsphenoidal surgery for Cushing’s disease. Clin. 491–497, 2001. Endocrinol. (Oxf), 38: 79–86, 1993. 33. Ruebel, K. H., Jin, L., Zhang, S., Sceithauer, B. W., and Lloyd, 38. Horvath, E., Kovacs, K., and Scheithauer, B. W. Pituitary hyper- R. V. Inactivation of the p16 gene in human pituitary nonfunctioning plasia. Pituitary, 1: 169–179, 1999.

Downloaded from clincancerres.aacrjournals.org on September 27, 2021. © 2004 American Association for Cancer Research. Molecular Pathology Shows p16 Methylation in Nonadenomatous Pituitaries from Patients with Cushing's Disease

David J. Simpson, Anne M. McNicol, David C. Murray, et al.

Clin Cancer Res 2004;10:1780-1788.

Updated version Access the most recent version of this article at: http://clincancerres.aacrjournals.org/content/10/5/1780

Cited articles This article cites 34 articles, 5 of which you can access for free at: http://clincancerres.aacrjournals.org/content/10/5/1780.full#ref-list-1

Citing articles This article has been cited by 3 HighWire-hosted articles. Access the articles at: http://clincancerres.aacrjournals.org/content/10/5/1780.full#related-urls

E-mail alerts Sign up to receive free email-alerts related to this article or journal.

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications Subscriptions Department at [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://clincancerres.aacrjournals.org/content/10/5/1780. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.