Journal of PathologyJ Pathol 2004; 203: 814–821Published online in Wiley InterScience (www.interscience.wiley.com). DOI: 10.1002/path.1562

Original Paper

Possible linkage between specific histological structuresand aberrant reactivation of the Wnt pathway inadamantinomatous craniopharyngioma

Keisuke Kato,1* Yukio Nakatani,2 Hiroshi Kanno,3 Yoshiyuki Inayama,2 Rieko Ijiri,1 Noriyuki Nagahara,4

Tetsumi Miyake,1 Mio Tanaka,1 Yumi Ito,5 Noriko Aida,6 Katsuhiko Tachibana,7 Ken-ichi Sekido8

and Yukichi Tanaka1

1Division of Pathology, Kanagawa Children’s Medical Center, Yokohama, Japan2Department of Pathology, Yokohama City University School of Medicine, Yokohama, Japan3Department of Neurosurgery, Yokohama City University School of Medicine, Yokohama, Japan4Department of Hygiene, Nihon Medical School, Tokyo, Japan5Department of Oral Pathology, Kanagawa Dental College, Yokosuka, Japan6Division of Radiology, Kanagawa Children’s Medical Center, Yokohama, Japan7Division of Endocrinology, Kanagawa Children’s Medical Center, Yokohama, Japan8Division of Neurosurgery, Kanagawa Children’s Medical Center, Yokohama, Japan

*Correspondence to:Keisuke Kato, Division ofPathology, Kanagawa Children’sMedical Center, Yokohama,232-8555, Japan.E-mail: p-cmck@gd5.so-net.ne.jp

Received: 23 July 2003Revised: 3 December 2003Accepted: 19 January 2004

Abstract

This study concerns the significance of nuclear/cytoplasmic expression of beta-catenin andmutation of the beta-catenin gene in craniopharyngiomas. Fourteen adamantinomatoustype and one squamous papillary type craniopharyngiomas were studied. Histologically,13 of 14 adamantinomatous type craniopharyngiomas showed typical features, ie mixturesof ‘palisading cells’, ‘stellate cells’, and ‘ghost cells’. In addition, ‘whorl-like arrays’ ofepithelial cells were frequently observed in the areas of stellate cells. On immunohistochem-istry, all typical adamantinomatous type craniopharyngiomas showed nuclear/cytoplasmicexpression of beta-catenin predominantly in cohesive cells within the whorl-like arrays andin cells transitional towards ghost cells, where immunoreactivity for Ki-67 was almostabsent. The cohesive cells in the whorl-like arrays also demonstrated loss of cytoker-atin isoform expression. Using direct sequencing of amplified nucleic acids, nine of the13 typical adamantinomatous type craniopharyngiomas with nuclear beta-catenin accumu-lation showed heterozygous one-base substitution mutation of the beta-catenin gene. Theother unusual adamantinomatous type and squamous papillary type craniopharyngiomasshowed no obvious nuclear/cytoplasmic beta-catenin immunoreactivity and no mutation ofthe beta-catenin gene, suggesting molecular heterogeneity. These findings suggest that thepathogenesis of typical adamantinomatous type craniopharyngioma is associated with abnor-malities of Wnt signalling that act as a morphogenetic signal towards whorl-like arrays andghost cells rather than as simple proliferation stimuli.Copyright 2004 Pathological Society of Great Britain and Ireland. Published by JohnWiley & Sons, Ltd.

Keywords: Wnt signal; beta-catenin; craniopharyngioma

Introduction

Craniopharyngioma (CP) is an intrasellar/suprasellardysembryonal neoplasm accounting for 10% of allpaediatric intracranial tumours [1]. Histologically, CPis benign, but it has a tendency to recur with significantpost-treatment sequelae [1]. CPs are classified into twohistopathological subtypes with distinct clinical fea-tures: adamantinomatous type (A-type) and squamouspapillary type (SP-type) [1]. Although the histogen-esis is unknown, A-type CPs are considered to bederived from an enamel anlage in Rathke’s pouch andSP-type CPs are assumed to originate in metaplas-tic squamous cell nests of the adenohypophysis [1].

The former hypothesis comes from the histologicalsimilarity between A-type CPs and odontogenic neo-plasms [2].

The Wnt pathway is implicated in the developmentalprocess [3]. Disturbance of Wnt signalling duringorganogenesis has been shown to result in severe struc-tural abnormality [3]. The Wnt pathway is also impli-cated in the development of several neoplasms, includ-ing colon cancer and pilomatricoma [3–7]. Familialadenomatous polyposis (FAP) results from mutationof the adenomatous polyposis coli gene (APC ), whichencodes a negative regulator of the Wnt pathway,and patients with FAP can develop Wnt pathway-associated neoplasms such as desmoid tumours [3].

Copyright 2004 Pathological Society of Great Britain and Ireland. Published by John Wiley & Sons, Ltd.

Significance of nuclear/cytoplasmic expression of beta-catenin in craniopharyngiomas 815

Activating mutations of CTNNB1 encoding proteinbeta-catenin (CTNNB1), which plays a pivotal rolein the Wnt pathway, have been found in these neo-plasms [3]. These CTNNB1 mutations affect serinephosphorylation of CTNNB1 at residue 33 by glyco-gen synthase kinase 3 beta (GSK-3 beta), which iscritical for CTNNB1 degradation, resulting in translo-cation of CTNNB1 to the nucleus. This in turn formsa complex with the T-cell factor/lymphoid enhancerbinding factor family, acting as a transcription factorto reactivate the Wnt pathway [3].

The emergence of solid cell nests that mimicmulberries (‘morules’) has been reported in severalneoplasms, including thyroid carcinoma, low-gradeadenocarcinoma of fetal lung type, and pancreato-blastoma, although the designations are variable [6,7].Some nuclei in morules are optically clear and filledwith biotin-rich material [6,7]. Recent investigationshave revealed that some neoplasms with morules haveactivating mutations of CTNNB1 [6,7].

The biology, histogenesis, and molecular pathogen-esis of CP have not been extensively investigated [1].We found cytoplasmic/nuclear immunoreactivity forCTNNB1 to be present predominantly in whorl-likearrays that mimic morules in A-type CPs. Recently,

Sekine et al have identified CTNNB1 mutation in A-type CPs but not in SP-type CPs [8]. Moreover, theydescribed nuclear CTNNB1 accumulation in the mes-enchymal component. However, they did not demon-strate the biological significance of CTNNB1 mutationin A-type CP. In this study, we have investigatedCTNNB1 mutation further and have analysed its rela-tionship to histological findings in CPs.

Materials and methods

Fifteen CPs were studied with informed consent. Nopatient had a history of FAP. The tumour samples werefixed in 10% buffered-formalin, acid-decalcified, andembedded in paraffin wax. The clinical information islisted in Table 1.

Immunohistochemistry

Immunohistochemistry was performed using an indi-rect immunoperoxidase method using the antibodieslisted in Table 2, according to previously reportedmethods [5,7]. For double immunohistochemistry,after development with diaminobenzidine tetrachlo-ride, the sections underwent microwave treatment to

Table 1. Clinical features

PatientNo Age/sex

Tumoursize (cm) Surgery

Recurrence/regrowth

Secondrecurrence/regrowth

Outcome/period afterlast operation

1 2y 10m/F 1.5 × 1.5 Subtotal Yes No NOD/22y 10m2 13y 3m/M 3.0 × 4.0 × 4.0 Partial Yes No Died of complications due to

chemotherapy/10m3 3y 10m/F 2.0 × 3.0 Subtotal No — NOD/16y 8m4 5y 7m/M 3.2 × 2.8 × 2.5 Partial Yes No NOD/20y 5m5 1y/F 2.5 × 2.0 Subtotal No — NOD/8y6 6y 2m/M 3.0 × 2.5 × 3.0 Subtotal Yes Yes AWD/7y 10m7 3y 11m/M 11.4 × 9.9 × 7.7 Subtotal No — NOD/7y8 6y 8m/F 2.5 × 2.5 × 4.3 Partial Yes Yes NOD/2m9 3y 3m/M 3.5 × 3.5 Total No — NOD/4y7m

10 7y 3m/M 2.8 × 3.8 × 3.6 Total No — NOD/1y 4m11 6y 10m/M 2.1 × 4.5 × 3.2 Subtotal No — NOD/6m12 28y/M 4.0∗ Partial No — NOD/7y8m13 61y/M 2.0∗ Partial No — NOD/5y2m14 66y/F 2.0∗ Total Yes — DOD/2y15 48y/M 2.5∗ Partial No — NOD/2y5m

y = years; m = months; M = male; F = female; NOD = no evidence of disease; AWD = alive with disease; DOD = died of disease. ∗ Maximumdiameter.

Table 2. List of the antibodies employed

Antibody Clone DilutionAntigenretrieval Manufacturer

CTNNB1 14 1/100 MW Transduction Laboratories, Lexington, KY, USAKi-67 MIB-1 1/100 MW Immunotech, Marseille, FranceCytokeratin (Nos 10, 14–16, and 19) AE-1 1/400 MW ICN Biomedicals, Aurora, OH, USACytokeratin (Nos 1–8) AE-3 1/400 MW ICN BiomedicalsCytokeratin (No 8) CAM5.2 1/10 MW Becton Dickinson, San Jose, CA, USACytokeratin (No 7) OV-TL 12/30 1/1 MW DAKO A/S, Glostrup, DenmarkBiotin BK-1/39 1/20 None DAKO A/S

MW = microwave treatment in 0.01 M citrate buffer (pH 6.0) using a 150 W oven.

J Pathol 2004; 203: 814–821

816 K Kato et al

inactivate the antibodies and were then subjected toan indirect immunoperoxidase method using alkalinephosphatase (AP)-conjugated anti-mouse immuno-globulin Fab fragments and a new Fuchsin reagent kit(Nichirei, Tokyo, Japan).

CTNNB1 mutation analysis

Molecular investigation was performed according topreviously reported methods [5,7]. Genomic DNAwas obtained by proteinase K digestion and phe-nol/chloroform extraction from fresh-frozen samplesor from paraffin wax-embedded tissue. For paraf-fin wax-embedded tissue, the tumour areas and non-tumour areas (as normal control) were dissected usinga sterile blade. In cases 6, 7, and 11, both fresh-frozen samples and paraffin wax-embedded tissuewere available. For paraffin wax-embedded tissue, a95 bp sequence containing the phosphorylation sitefor GSK-3 beta in exon 3 of CTNNB1 was ampli-fied: a 222 bp fragment was amplified from fresh-frozen tissue. The amplified products were submit-ted to direct sequencing using an ABI Big Dye Ter-minator Cycle Sequencing Kit (Applied Biosystems,Foster City, CA, USA) and the same primer pairsused for PCR. Sequence analysis was performed withABI Genetic Analyzer Model 310 (Applied Biosys-tems). Three independent analyses were performed andthree consistent results were considered to be a finalresult.

Results

Histopathology

Fourteen of the 15 CPs studied were of A-type (cases1–13 and 15) and one was of SP-type (case 14).Thirteen (cases 1–13) of the 14 A-type CPs showedtypical histology. The tumour cell nests were com-posed of palisading cells at the periphery and stel-late cells inside the nests. Within the nests were fociof ghost cells frequently associated with dystrophiccalcification (Figure 1A). Of note was that stellatecells aggregated focally as whorl-like arrays, insidewhich the tumour cells became compactly cohesivewith more ample cytoplasm than the usual stellatecells (Figure 1B). Occasionally, these structures con-tained foci of ghost cells with gradual transition fromthe epithelial cells (Figure 1C). Neither optically clearnuclei in the whorl-like array nor a neoplastic mes-enchymal component was evident.

One of the 14 A-type CPs showed slightly unusualhistology (case 15). The tumour cells were arranged inthin cords or trabeculae composed of limiting palisad-ing cells and internal stellate cells with foci of acantho-sis. Occasionally, the cords formed duct-like features,but showed no ghost cells, calcification, or whorl-likearrays. The remaining case (case 14) showed typicalSP-type histology, characterized by proliferation ofstratified squamous epithelial cells (intermediate and

superficial layers) resting on the basal layers. Foci ofcalcification were far less frequently encountered.

Immunohistochemistry

The results of immunohistochemistry are presented inTable 3. Typical A-type CPs showed strong nuclear/cytoplasmic immunoreactivity for CTNNB1 in com-pactly cohesive epithelial cells within the whorl-likearrays and in cells that were transitional towardsghost cells (Figures 2A and 2B). In addition, thepalisading cells showed scattered aberrant CTNNB1expression. Although artificial damage made it dif-ficult to confirm the whorl-like arrays, ancillaryCTNNB1 immunohistochemistry confirmed their pres-ence. Other tumour cells showed diffuse membra-nous immunoreactivity. No obvious mesenchymalcells with aberrant CTNNB1 expression were identi-fied. The tumour cells of the remaining unusual A-typeand SP-type CPs showed membranous immunoreac-tivity for CTNNB1 with no/rare nuclear/cytoplasmicimmunoreactivity.

The palisading cells of the typical A-type CPs weremore frequently Ki-67-positive (average 0.076) thanthe stellate cells (average 0.022). Within the stel-late cells, the cells near the palisading cells and thewhorl-like arrays were more frequently stained thanother stellate cells. By contrast, the compactly cohe-sive cells within the whorl-like arrays, immunoreactivefor CTNNB1 in the nucleus/cytoplasm, were almostnegative and this was confirmed by double immuno-histochemistry for Ki-67 and CTNNB1 (Figure 3A).The neoplastic cells of the unusual A-type CP showedoccasional immunoreactivity for Ki-67 without a spe-cific distribution pattern. Basal and intermediate cellswere stained in the SP-type CP, the rates being 0.22and 0.05, respectively. The surface cells were almostnegative.

Cohesive cells within the whorl-like array of thetypical A-type CPs, which were immunoreactive forCTNNB1 in the nucleus/cytoplasm, were not reactivefor AE-3 (which recognizes cytokeratins 1–8), AE-1 (which recognizes cytokeratins 10, 14–16, and19), OV-TL 12/30 (which recognizes cytokeratin 7),or CAM5.2 (which recognizes cytokeratin 8). Thiswas confirmed by double immunostains for AE-3and CTNNB1 (Figure 3B). The ghost cells wereimmunoreactive only for AE-3. The stellate cells of thetypical A-type CPs were reactive for AE-3, AE-1, OV-TL 12/30, and CAM5.2. Immunostains for AE-1 andCAM5.2 labelled the stellate cells around the whorl-like arrays more intensely than other stellate cells.The palisading cells of the typical A-type CPs werestrongly reactive for AE-3 and not/weakly for AE-1,OV-TL 12/30, and CAM5.2. The neoplastic cells ofthe remaining unusual A-type and SP-type CPs werediffusely immunoreactive for CAM 5.2, AE-1, OV-LT 12/30, and AE-3. Immunohistochemistry for biotin

J Pathol 2004; 203: 814–821

Significance of nuclear/cytoplasmic expression of beta-catenin in craniopharyngiomas 817

A

C

B

PP

S

Figure 1. Histology of typical A-type CPs. (A) General appearance of typical A-type CPs, composed of limiting palisading cells(P), internal stellate cells (S), and ghost cells (G). Note that epithelial cells aggregate in the stellate cells, forming whorl-like arrays(arrow-heads). (B) Epithelial whorl-like array (small arrow-heads). Note the aggregates of compactly cohesive epithelial cells withinthe whorl-like array (large arrow-heads). P = palisading cells; S = stellate cells. (C) Gradual transition from epithelial cells to ghostcells (G) in the epithelial whorl-like array

A

P

P

PS

B

S

Figure 2. Immunohistochemistry for CTNNB1 in typical A-type CPs. (A) Nuclear/cytoplasmic immunoreactivity (arrow-heads)in the compactly cohesive cells within the whorl-like array. P = palisading cells; S = stellate cells. (B) Nuclear/cytoplasmicimmunoreactivity (arrow-heads) in the transitional cells adjacent to the ghost cells (G). S = stellate cells

J Pathol 2004; 203: 814–821

818 K Kato et al

Table 3. Results of immunohistochemistry and molecular study

Patient HistologicalKi-67

No subtype CTNNB1 AE-3 P S WA Mutation of CTNNB1

1 Typical A-type M/NC (WA) +(P/S/G), −(WA) —∗ —∗ — —2 Typical A-type M/NC (WA) +(P/S/G), −(WA) 0.068 0.013 0 —3 Typical A-type M/NC (WA) +(P/S/G), −(WA) 0.093 0.036 0 —4 Typical A-type M/NC (WA) +(P/S/G), −(WA) 0.153 0.033 0 gac → aac Asp 32 Asn5 Typical A-type M/NC (WA) +(P/S/G), −(WA) 0.049 0.037 0 —6 Typical A-type M/NC (WA) +(P/S/G), −(WA) 0.059 0.01 0 acc → atc Thr 41 Ile7 Typical A-type M/NC (WA) +(P/S/G), −(WA) 0.049 0.021 0 gac → tac Asp 32 Tyr8 Typical A-type M/NC (WA) +(P/S/G), −(WA) 0.045 0.013 0 tct → cct Ser 33 Pro9 Typical A-type M/NC (WA) +(P/S/G), −(WA) 0.055 0.011 0 gac → tac Asp 32 Tyr

10 Typical A-type M/NC (WA) +(P/S/G), −(WA) 0.145 0.04 0 tct → cct Ser 33 Pro11 Typical A-type M/NC (WA) +(P/S/G), −(WA) 0.072 0.019 0 tct → cct Ser 33 Pro12 Typical A-type M/NC (WA) +(P/S/G), −(WA) 0.185 0.045 0 tct → tgt Ser 37 Cys13 Typical A-type M/NC (WA) +(P/S/G), −(WA) 0.012 0.002 0 gct → act Ala 43 Thr14 SP-type M + 0.225 (B) 0.05 (I) — —15 Unusual A-type M + almost 0 almost 0 — —

M = membranous; NC = nuclear/cytoplasmic; WA = cohesive cells in the whorl-like array; P = palisading cells; S = stellate cells; G = ghostcells; B = basal cells; I = intermediate cells. ∗ The tissue was too small to be evaluated.

showed no obvious positive signal in the typical andunusual A-type CPs.

CTNNB1 mutation analysis

Direct sequencing of the amplified CTNNB1 productrevealed a heterozygous one-base nucleotide sub-stitution in nine of the 13 typical A-type CPswith nuclear/cytoplasmic CTNNB1 immunoreactivity(Table 3 and Figures 4A and 4B). The nine observednucleotide substitutions were missense mutations ofcodons 32, 33, 37, 41, and 43. In cases 7 and 11,independent analyses of fresh-frozen tissue and paraf-fin wax-embedded tissue showed the same mutations.These CTNNB1 alterations were not observed in thesurrounding stroma in all the tumours where it waspossible to dissect the tumour areas separately fromthe stroma (cases 4 and 7–13). In case 6, CTNNB1was not amplifiable in the analysis of paraffin wax-embedded tumour and non-tumour tissue. However,three independent analyses of the fresh-frozen tissuerevealed a heterozygous one-base substitution in codon41, which encodes threonine, a possible kinase sub-strate; thus, this substitution was not considered to bea polymorphism but rather a significant mutation. Nomutation was identified in the remaining four typicalA-type CPs, in the one unusual A-type CP, or in thesingle SP-type CP.

Discussion

There has been limited investigation of the molecularpathogenesis of CP. Gorski et al pointed out thatchromosomal translocations involving chromosomes2 and 12, not loci of CTNNB1, were observed inboth CPs examined thus far [9]. In addition, Sarubiet al showed that A-type CP is clonal but shows nomutation of the PTCH, Gsalpha, or Gi2alpha genes

responsible for Gorlin syndrome and pituitary tumour,respectively [10]. However, they did not show specificgene mutations in CPs.

Our results using CTNNB1 immunohistochemistryand CTNNB1 mutation analysis indicate reactivationof Wnt signalling in typical A-type CPs. Although fourof our 13 typical A-type CPs showed no CTNNB1mutations, they may have arisen as a result of muta-tion of other genes, such as APC, that encode Wntsignal regulators, or by deletion mutation of CTNNB1.Recently, similar to our results, Sekine et al have iden-tified mutation of CTNNB1 in all of ten A-type CPsbut not in six SP-type CPs [8]. Furthermore, A-typeCP has been reported in a patient with FAP, in addi-tion to other Wnt pathway-associated neoplasms [11].These findings suggest that reactivation of the Wntpathway might be a significant step in the developmentof A-type CPs.

In A-type CP, whorl-like arrays have not beenfully described. Sekine et al described these specificstructures for the first time where aberrant CTNNB1expression is frequently observed in A-type CPs [8].The present study further identified the specific natureof the whorl-like array. It has been suggested thataberrant reactivation of Wnt signalling is a featurecommon to neoplasms with morules [6,7]. Previousstudies have shown that morules frequently demon-strate nuclear/cytoplasmic CTNNB1 immunoreactivity[6,7]. In addition, several authors have pointed outthat morules have a different cytokeratin expressionprofile to that surrounding neoplastic tissue [12]. Wefound histological similarity between the whorl-likearray and morules, and demonstrated nuclear CTNNB1accumulation and loss of cytokeratin isoform expres-sion. These findings suggest that the whorl-like arrayhas a morule-like nature, although optically clear

J Pathol 2004; 203: 814–821

Significance of nuclear/cytoplasmic expression of beta-catenin in craniopharyngiomas 819

A

PS

SP

B

Figure 3. Immunohistochemistry for Ki-67/CTNNB1 and AE-3/CTNNB1 in typical A-type CPs. (A) Double immunostainingfor Ki-67 (brown) and CTNNB1 (red). Note that CTNNB1-immunoreactive compactly cohesive cells (arrow-heads) show noobvious Ki-67 immunoreactivity. The Ki-67 frequency is higher in the palisading cells (P) than in the stellate cells (S). (B) Doubleimmunostains for AE-3 (red) and CTNNB1 (brown). Note the compactly cohesive cells negative for AE-3 but immunoreactive forCTNNB1 in their nucleus/cytoplasm (arrow-heads)

A Bcase 7

the tumour tissue

codon 32

the stroma

GAC to TAC (Asp to Tyr)

T C CT T A A ACC T T TTG GGN

10 20

A C C

T C CT T A A ACC T T TTGG GG10 20

A C C

case 12

codon 37

the tumour tissue

the stroma

TCT to TGT (Ser to Cys)

T T T T T TNA A AA C C C CCGG GG G

20 30

T T T T TTA A AAC C CCC CG G GGG20 30

Figure 4. Electropherogram showing heterozygous one-base substitution of CTNNB1 in typical A-type CPs. (A) Case 7 showsheterozygous G-to-T substitution at codon 32 resulting in replacement of aspartic acid by tyrosine. An arrow-head indicatesheterozygosity for G (black line) and T (red line) in the tumour tissue. (B) Case 12 shows heterozygous C-to-G substitution atcodon 37 resulting in replacement of serine by cysteine. An arrow-head indicates heterozygosity for C (blue line) and G (blackline) in the tumour tissue

J Pathol 2004; 203: 814–821

820 K Kato et al

nuclei and biotin immunoreactivity were not con-firmed. These aspects further suggest the notion thatthe formation of morules or morule-like structures maybe a common manifestation of aberrant reactivation ofthe Wnt pathway in the context of tumourigenesis.

Our present study opens the debate regarding thehistogenesis of typical A-type CP. Although, initially,it was suggested that CPs demonstrate histologicalsimilarity to ameloblastoma, subsequent scrutiny hasdisclosed that most A-type CPs resemble the odon-togenic neoplasm, calcifying odontogenic cyst (COC)[2]. Moreover, A-type CP and COC have been shownto mimic pilomatricoma based on the presence ofghost cells [2]. Pilomatricoma is a benign skin tumourthat shows differentiation towards hair follicles [13].Recent investigation has identified that pilomatricomatumourigenesis is associated with up-regulation of theWnt pathway mainly caused by CTNNB1 mutation [4].The Wnt pathway is one of the main signal transduc-tion systems associated with terminal differentiationin the normal hair shaft as a result of regulation ofcytokeratin expression, and accumulation of CTNNB1has been reported in differentiating central matrix cellsin hair follicles [13]. Notably, A-type CPs have beenfound to show differentiation towards hair folliclesrather than oropharyngeal enamel anlage, accordingto cytokeratin expression profiles [14]. These findingssuggest that A-type CP and pilomatricoma might berelated tumours with histological similarity and spe-cific lines of differentiation induced by aberrant reac-tivation of the Wnt pathway as a common geneticbackground. In this context, it would be of interest toinvestigate cytokeratin isoform and CTNNB1 expres-sion in COC.

Our present study indicates that reactivation of theWnt signal pathway does not act as a simple prolifera-tion signal in typical A-type CPs. Double immunohis-tochemistry disclosed that aberrant CTNNB1 expres-sion in the cohesive cells of the whorl-like array wasinversely related to Ki-67 immunoreactivity. Further-more, such aberrant expression was frequently accom-panied by ghost cell formation, possibly indicating aterminal differentiation process in typical A-type CPs.In some neoplasms, it has been reported that aber-rant reactivation of the Wnt pathway induces a specificmorphology rather than simple proliferation. The accu-mulation of CTNNB1 predominantly in morules is onesuch example [6,7]. In colon cancer, Jung et al havefound that the invasion front shows less frequent Ki-67immunoreactivity but a higher frequency of CTNNB1nuclear accumulation, and they have pointed out thatthe invasive front resembles the first phase of gas-trulation during sea urchin embryogenesis [15]. Thesefindings indicate that reactivation of the Wnt pathwaydoes not merely function as a mitogenic stimulus, butrather causes mis-specification of cells to form distinctstructures in the context of tumourigenesis in typicalA-type CPs.

Our present study did not identify the mesenchymalcomponent advocated by Sekine et al [8]. They found

nuclear/cytoplasmic CTNNB1 immunoreactivity in themesenchymal cells distinct from the epithelial cellnests and the same mutation was present in theepithelial and mesenchymal components in two A-type CPs [8]. Complete resolution of this matter awaitsfurther study.

Although the sample sizes are restricted, our find-ings and the results presented by Sekine et al suggestthat CTNNB1 immunohistochemistry may be usefulfor discriminating A-type from SP-type CPs [1]. How-ever, in our study, one A-type CP showed unusual his-tology and no nuclear/cytoplasmic CTNNB1 immuno-reactivity, suggesting histological heterogeneity. Inthis regard, some authors have advocated that A-typeCP is further subclassified into major COC type andminor types, based on histological similarities withodontogenic tumours [2,16]. According to their clas-sification, our typical A-type CPs would be consid-ered to be of COC type and the unusual A-type CPwould be considered to be of adenomatoid odonto-genic tumour type. These findings suggest that A-typeCP is histologically and genetically heterogeneous.

Significant morbidity and mortality result from thesustained growth and invasion of CP [1] and thereforenew treatment strategies are required for this tumour.Recently, non-steroidal anti-inflammatory drugs andanti-oestrogens have received attention as new treat-ment strategies against some neoplasms with Wnt sig-nal abnormality [17]. Our new insights may open thedebate regarding the biology of CPs and may con-tribute to the development of new treatment modalitiesbased on this biology.

Acknowledgements

We are grateful to Mss Takako Watanabe, Yukari Hasegawa,and Ikuko Nanba for histological technical assistance. Part ofthis work was presented at the 92nd Annual Meeting of theJapanese Society of Pathology, Fukuoka, Japan in April 2003.

References

1. Thapar K, Kovacs K. Neoplasms of the sellar region. In Russell& Rubinstein’s Pathology of Tumors of the Nervous System,Bigner DD, McLendon RE, Bruner JM (eds), vol 2. Arnold:London, 1998; 561–677.

2. Bernstein ML, Buchino JJ. The histologic similarity betweencraniopharyngioma and odontogenic lesions: a reappraisal. OralSurg Oral Med Oral Pathol 1983; 56: 502–511.

3. Polakis P. Wnt signaling and cancer. Genes Dev 2000; 14:1837–1851.

4. Chan EF, Gat U, McNiff JM, Fuchs E. A common human skintumour is caused by activating mutations in beta-catenin. NatureGenet 1999; 21: 410–413.

5. Tanaka Y, Kato K, Notohara K, et al. Frequent beta-catenin muta-tion and cytoplasmic/nuclear accumulation in pancreatic solid-pseudopapillary neoplasm. Cancer Res 2001; 61: 8401–8404.

6. Nakatani Y, Masudo K, Miyagi Y, et al. Aberrant nuclearlocalization and gene mutation of beta-catenin in low-gradeadenocarcinoma of fetal lung type: up-regulation of the Wntsignaling pathway may be a common denominator for thedevelopment of tumors that form morules. Mod Pathol 2002; 15:617–624.

J Pathol 2004; 203: 814–821

Significance of nuclear/cytoplasmic expression of beta-catenin in craniopharyngiomas 821

7. Tanaka Y, Kato K, Notohara K, et al. Significance of aberrant(cytoplasmic/nuclear) expression of beta-catenin in pancreatoblas-toma. J Pathol 2003; 199: 185–190.

8. Sekine S, Shibata T, Kokubu A, et al. Craniopharyngiomas ofadamantinomatous type harbor beta-catenin gene mutations. AmJ Pathol 2002; 161: 1997–2001.

9. Gorski GK, McMorrow LE, Donaldson MH, et al. Multiplechromosomal abnormalities in a case of craniopharyngioma.Cancer Genet Cytogenet 1992; 60: 212–213.

10. Sarubi JC, Bei H, Adams EF, et al. Clonal composition of humanadamantinomatous craniopharyngiomas and somatic mutationanalyses of the patched (PTCH), Gsalpha and Gi2alpha genes.Neurosci Lett 2001; 310: 5–8.

11. Link MJ, Driscoll CL, Giannini C. Isolated, giant cerebellopontineangle craniopharyngioma in a patient with Gardner syndrome: casereport. Neurosurgery 2002; 51: 221–226.

12. Okamoto Y, Kashima K, Daa T, et al. Morule with biotin-containing intranuclear inclusions in thyroid carcinoma. Pathol Int1995; 45: 573–579.

13. Moreno-Bueno G, Gamallo C, Perez-Gallego L, et al. Beta-catenin expression in pilomatricomas. Relationship with beta-catenin gene mutations and comparison with beta-cateninexpression in normal hair follicles. Br J Dermatol 2001; 145:576–581.

14. Tateyama H, Tada T, Okabe M, et al. Different keratin profilesin craniopharyngioma subtypes and ameloblastomas. Pathol ResPract 2001; 197: 735–742.

15. Jung A, Schrauder M, Oswald U, et al. The invasion front ofhuman colorectal adenocarcinomas shows co-localization ofnuclear beta-catenin, cyclin D1, and p16INK4A and is a regionof low proliferation. Am J Pathol 2001; 159: 1613–1617.

16. Paulus W, Stockel C, Krauss J, et al. Odontogenic classificationof craniopharyngiomas: a clinicopathological study of 54 cases.Histopathology 1997; 30: 172–176.

17. Samuels BL. Management of recurrent desmoid tumor aftersurgery and radiation: role of cytotoxic and non-cytotoxictherapies. Surg Oncol 1999; 8: 191–196.

J Pathol 2004; 203: 814–821