1 23

Endocrine Pathology ISSN 1046-3976Volume 23Number 4 Endocr Pathol (2012) 23:221-231DOI 10.1007/s12022-012-9222-y

Biomarkers of Parathyroid Carcinoma

Boban M. Erovic, Luke Harris, MinaJamali, David P. Goldstein, JonathanC. Irish, Sylvia L. Asa & Ozgur Mete

1 23

Your article is protected by copyright and

all rights are held exclusively by Springer

Science+Business Media, LLC. This e-offprint

is for personal use only and shall not be self-

archived in electronic repositories. If you

wish to self-archive your work, please use the

accepted author’s version for posting to your

own website or your institution’s repository.

You may further deposit the accepted author’s

version on a funder’s repository at a funder’s

request, provided it is not made publicly

available until 12 months after publication.

Biomarkers of Parathyroid Carcinoma

Boban M. Erovic & Luke Harris & Mina Jamali &David P. Goldstein & Jonathan C. Irish &

Sylvia L. Asa & Ozgur Mete

Published online: 22 September 2012# Springer Science+Business Media, LLC 2012

Abstract The diagnosis of parathyroid carcinoma can bechallenging, and adjuvant therapies such as radiotherapyand chemotherapy are not particularly beneficial in themanagement of this disease, creating a challenge whendealing with unresectable recurrent and metastatic malig-nancy. We investigated the expression profile of biomarkersthat represent potential markers of malignancy or targets fornovel therapies in this disease. We constructed a tissuemicroarray of parathyroid carcinomas from 10 patients aswell as parathyroid adenomas from 25 patients and stainedthe slides for 34 proteins involved in angiogenesis (platelet-derived growth factor receptor (PDGFR)-α, PDGFR-β, vas-cular endothelial growth factor receptor-2 (VEGFR-2), andepidermal growth factor receptor (EGFR)), inflammation(cyclooxygenase (COX)-1 and COX-2), cell adhesion(matrix metalloproteinase (MMP)-1, CD9, and keratin 7),cell cycle (Cdc2p34, cyclin D1, retinoblastoma (Rb), p27,p21, parafibromin, Bmi-1, 14-3-3σ, and p53), and apoptosis(Bcl-2a, Mcl-1, Bcl-xL, and glutathione-S-transferase-isoen-zyme π (Gst-π)) along with some markers of the sonichedgehog (Smo, SHH, Gli-1, Gli-2, Gli-3, and patched),mTOR (AKT, mammalian target of rapamycin (mTOR),and Forkhead box O (FoxO)-1), and WNT (Wisp-1, Wisp-

2, and β-catenin) signal transduction pathways. Proteinexpression was determined using computerized image anal-ysis software (Spectrum Plus©, Aperio). Bcl-2a, parafibro-min, Rb, and p27 were significantly decreased to variabledegrees in all parathyroid carcinomas. COX-1/2, CD9,MMP-1, FoxO-1, VEGFR-2, PDGFR-α/β, Gst-π, Gli-1,Gli-2, Gli-3, and patched were expressed in the majority ofbenign and malignant tumor cells. These results indicate thatthe use of a panel that includes Bcl-2a, parafibromin, Rb,and p27 may be helpful in the assessment of atypical para-thyroid neoplasms. Although the majority of other markersstudied are also expressed in both benign and malignantparathyroid neoplasms, we have identified several potentiallyimportant target proteins related to angiogenesis and cellproliferation along with COX-1/2, Gst-π and members ofsonic hedgehog pathway that may be therapeutic targets inparathyroid carcinoma. While these results are preliminary, asuccessful outcome of a clinical trial directed against thesenovel targets would provide much needed systemic adjuvanttreatment for patients with metastatic parathyroid carcinoma.

Keywords Parathyroid carcinoma . Angioinvasion .

Bcl-2a . Sonic hedgehog . Parafibromin .

Targeted anticancer therapy

Introduction

Parathyroid carcinoma is a rare and highly aggressive ma-lignant tumor that accounts for less than 1 % of all cases ofprimary hyperparathyroidism [1]. The histopathological di-agnosis of malignancy can be challenging in borderlinecases. Once diagnosed, the initial treatment of patients withparathyroid carcinoma is en bloc resection of the parathy-roid gland with ipsilateral hemithyroidectomy and centralneck dissection [2]. However, even with current treatment

B. M. Erovic : L. Harris :D. P. Goldstein : J. C. IrishDepartment of Otolaryngology-Head and Neck Surgery/SurgicalOncology, Wharton Head and Neck Program,Toronto, Ontario, Canada

M. Jamali : S. L. Asa :O. MeteDepartment of Pathology,University Health Network,Toronto, Ontario, Canada

O. Mete (*)Department of Pathology, University Health Network,200 Elizabeth Street, 11th Floor,Toronto, ON M5G 2C4, Canadae-mail: ozgur.mete2@uhn.ca

Endocr Pathol (2012) 23:221–231DOI 10.1007/s12022-012-9222-y

Author's personal copy

options, parathyroid carcinoma is usually associated with ahigh incidence of locoregional recurrence and distant me-tastasis [2]. Epidemiological studies have shown that duringthe past 30 years, there has been little change in the stage atwhich the disease presents, and little improvement in theprognosis of patients with parathyroid carcinoma [3, 4].

The biochemical disturbance due to severe hyperparathy-roidism is the primary cause of mortality in patients withrecurrent or metastatic parathyroid carcinoma. Moreover,there is some evidence that parathyroid carcinoma is lesssensitive to adjuvant radiotherapy and chemotherapy thanother forms of head and neck cancer [5, 6]. Given the highrecurrence rate and despite aggressive surgical treatment, aneffective systemic adjuvant therapy would represent a majorbreakthrough in the management of these patients. Earlyattempts were made by Baeta and colleagues to show thatimmune targeting of parathyroid cancer cells with bovineparathyroid hormone (PTH) fragments, modified humanPTH fragments, and intact human PTH led to decreased serumlevels of PTH and calcium in one patient with metastaticparathyroid carcinoma [7]. In addition, Schott and colleaguesinvestigated the role of tumor lysate and parathyroid hormone-pulsed dendritic cells [8, 9]; however, these attempts resultedin limited success. In the current study, we aimed to identifythe expression profile of several proteins that may representbiomarkers of parathyroid neoplasms and can be subject tonovel adjuvant targeted therapies in the management ofpatients with parathyroid carcinoma.

Materials and Methods

Specimens and Inclusion Criteria

Patients with parathyroid carcinoma managed at the Prin-cess Margaret Hospital between 1998 and 2010 were iden-tified through the hospital registry. Only newly diagnosedand previously untreated parathyroid carcinoma patients thatwere reviewed by the University Health Network EndocrinePathology site group were included. Patients with incom-plete medical records and those who did not pursue curativetreatment were excluded. Approval was obtained from theinstitutional Research Ethics Board.

A pathologic diagnosis of parathyroid carcinoma wasmade only when a parathyroid neoplasm exhibited un-equivocal evidence of vascular invasion, malignant in-vasion into adjacent structures (local gross invasion) ordocumented metastatic disease. A total of ten parathy-roid carcinomas had sufficient tissue to allow for theconstruction of a tissue microarray (TMA). A separateTMA containing 25 parathyroid adenomas was alsoconstructed for comparison. All tumors were reviewedby two endocrine pathologists (OM and SLA).

TMA Construction

Representative blocks were chosen for each case. TMAswere constructed with three cores from each case; each coremeasured 0.6 mm in diameter. The cores were mounted inlinear arrays in one paraffin TMA block. Tissue cores fromnormal parathyroid and normal liver were included for con-trol purposes. A character map was constructed delineatingthe various cases and their order in the TMA.

Immunohistochemistry and Automated Analysis

The immunohistochemical expression for 34 proteins involvedin angiogenesis (platelet-derived growth factor receptor(PDGFR)-α, PDGFR-β, vascular endothelial growth factorreceptor (VEGFR)-2, inflammation (cyclooxygenase (COX)-1 and COX-2), cell adhesion (matrix metalloproteinase(MMP)-1, CD9, and keratin 7), cell cycle (Cdc2p34, cyclinD1, retinoblastoma (Rb), p27, p21, parafibromin, Bmi-1, and14-3-3σ), and apoptosis (Bcl-2a, Mcl-1, Bcl-xL, p53, andglutathione-S-transferase-isoenzyme π (Gst-π)) along with epi-dermal growth factor receptor (EGFR) and somemarkers of thesonic hedgehog (Smo, SHH, Gli-1, Gli-2, Gli-3, and patched),AKT (Forkhead box O (FoxO)-1, AKT, and mammalian targetof rapamycin (mTOR)), and WNT (Wisp-1, Wisp-2, and β-catenin) signal transduction pathways were studied (Table 1).

Formalin-fixed paraffin-embedded sections (4 μm) weredewaxed in five changes of xylene and rehydrated throughgraded alcohols. Negative and positive control tissues wereselected based on previous publications where these anti-bodies were applied. Multiple control experiments wereundertaken to optimize each antibody, including absorptionof primary antibody or antiserum with purified antigen.Antigen retrieval or unmasking procedures were applied ifcontrol experiments indicated that this improved specificityor sensitivity (Table 1). Endogenous peroxidase wasblocked using 3 % hydrogen peroxide. After blocking for5 min with casein, sections were incubated with the appro-priate primary antibodies, and the reactions were detectedusing the Biogenex Super Sensitive Kit (cat. No. HK518-YAK) Super Enhancer followed with the Polymer HRPreagent and color development with NovaRed solution(Vector labs, cat. No. SK-4800). Sections were counter-stained lightly with Mayer’s hematoxylin, dehydrated inalcohols, cleared in xylene, and mounted with Permount(Fisher, cat. No. SP15-500). Staining for COX-1, COX-2,and patched used the avidin/biotin blocking kit (Lab Vision,cat. No. TA-015-BB) followed by primary antibody incuba-tion and biotinylated secondary antibody (Vector labs) for30 min and horseradish peroxidase-conjugated ultrastrepta-vidin labeling reagent (ID labs) for 30 min.

Immunostained slides of TMA sections were scannedusing a ScanScopeXT© (Aperio, Vista, CA) scanner and

222 Endocr Pathol (2012) 23:221–231

Author's personal copy

analyzed with Spectrum Plus© automated image analysessoftware (Aperio, Vista, CA). For cytoplasmic staining,Color Deconvolution V9.0 algorithm was used, with theresulting variables being average weak, moderate, andstrong cytoplasmic intensity scores. Nuclear and membranestaining algorithms were also used. The variables analyzedwere the percent positive cells and intensity scores (weak,moderate, and strong) for each individual core and theaverage percent positive cells and intensity are listed inTables 2 and 3. Although computerized automated analyseshave been validated to be reliable by several investigators,immunostained slides were also checked by conventionallight microscopy.

Results

Clinicopathological Features of Parathyroid Adenomasand Carcinomas

The study included ten patients with parathyroid carci-noma (four men and six women). The mean age at thetime of diagnosis was 55 years (range, 36–71 years). Atinitial presentation, no patients had clinical or radio-graphic evidence of regional nodal metastases, and noneof the patients were suspected of having parathyroidcarcinoma preoperatively. Three patients underwent fineneedle aspiration biopsy; on analysis, the aspirates

Table 1 Materials and methods for immunohistochemical staining

Antibody Source Cat. No. Pretreatment Dilution (incubation)

Parafibromin Santa Cruz Sc-33638 Flex TRS 1/50 (30 min)

14-3-3Sigma Thermo Scientific/Neomarkers MS-1185-PO TE9 1/400 (1 h)

Bcl-2 alpha Thermo Scientific/Neomarkers MS-123-PO TE9 1/100 (1 h)

Bcl-xl Thermo Scientific/Neomarkers MS-1334-PO TE9 1/1,000 (1 h)

Bmi-1 Millipore 05-637 TE9 1/400 (overnight)

Catenin beta(E5) Santa Cruz Sc-7963 Citrate 1/50 (overnight)

Cdc2p34 Santa Cruz Sc-54 Citrate 1/1,000 (overnight)

CD9 Novocastra NCL-CD9 TE9 1/200 (1 h)

Cytokeratin 7 Thermo Scientific/Neomarkers MS-1352 PO Trypsin 1/300 (overnight)

EGFR Invitrogen 28-0005 Pepsin 1/100 (1 h)

Gli-1 Santa Cruz Sc-20687 Citrate 1/100 (overnight)

Gli-2 Aviva ARP31885_T100 TE9 1/100 (overnight)

Gli-3 Santa Cruz Sc-20688 TE9 1/100 (overnight)

Mcl-1 Neomarkers MS-681 PO Trypsin 1/100 (overnight)

MMP 1 Lab Vision/Neomarkers RB-9225-PO citrate 1/800 (overnight)

PDGFR-α Abcam Ab61219 TE9 1/300 (overnight)

PDGFR-β Thermo Scientific/Neomarkers RB-1692-PO TE9 1/100 (overnight)

SHH Santa Cruz Sc-9024 Citrate 1/400 (overnight)

Smo Santa Cruz Sc-13943 Trypsin 1/300 (overnight)

VEGFR-2 Santa Cruz Sc-6251 TE9 1/1,000 (1 h)

Wisp-1 Santa Cruz Sc-25441 Citrate 1/100 (overnight)

Wisp-2 Santa Cruz Sc-25442 Trypsin 1/500 (overnight)

pAKT Cell Signaling 3787S Citrate 1/50 (overnight)

pmTOR ser2448 Cell Signaling 4976 Citrate 1/50 (overnight)

FoxO-1 Novus Biological NBPI-19825 Citrate 1/100 (overnight)

p27 Transduction Lab 610242 Citrate 1/1,000 (1 h)

p21 Pharmingen 556431 TE9 1/200 (overnight)

p53 Roche (Ventana) DO7, 790-2912 CC1 standard 32 min

Cyclin D1 Lab Vision RM 9104 TE9 1/200 (overnight)

Rb Pharmingen 554136 Citrate 1/300 (1 h)

Cox-1 Santa Cruz (Goat) Sc-1754 TE9 1/100 (overnight)

Cox-2 Santa Cruz (Goat) Sc-1747 Citrate 1/1,000 (overnight)

Patched Santa Cruz (Goat) Sc-6149 Trypsin 1/100 (overnight)

Gst-π Novocastra NCL-Gstpi Citrate 1/800 (1 h)

Endocr Pathol (2012) 23:221–231 223

Author's personal copy

confirmed a parathyroid lesion. All patients were man-aged with primary surgery, and eight (80 %) underwentpostoperative radiotherapy. The extent of surgical man-agement was resection of the parathyroid gland in fourpatients, parathyroidectomy and ipsilateral hemithyroi-dectomy in four patients, and parathyroidectomy withtotal thyroidectomy, thymectomy, and paratracheal neckdissection in two patients.

The mean diameter of the parathyroid specimens was3.4 cm (range, 1.5–7.5 cm), and the mean weight of theparathyroid glands was 19.15 g. Capsular invasion wasobserved in 70 % (n07) of the specimens, and invasion into

the adjacent soft tissue was observed in 50 % (n05) of thespecimens. Positive resection margins were observed in five(50 %) surgical specimens. Unequivocal evidence ofangioinvasion characterized by thrombus adherent to intra-vascular tumor cells [10] was observed in 70 % (n07) ofspecimens (Fig. 1). Three tumors that lack angioinvasionrevealed gross local invasion into adjacent structures.

For comparison, 25 adenomas were obtained from 10men and 15 women, mean age at diagnosis 51.84 years(range, 17–79 years). These patients all underwent limitedsurgical resection of the lesion and were cured of hyperpara-thyroidism post-operatively.

Table 2 Immunohistochemicalresults of a parathyroid carcino-ma TMA

Protein Average of percent ofpositive tumor cells (%)

Distribution of intensity in positivecells (%)

Weak Moderate Strong

Growth factor receptors EGFR 16.65 9.6 6.7 0.32

PDGFR-α 89.7 28.64 56.06 4.99

PDGFR-β 72.26 46.42 25.11 0.73

VEGFR-2 78.19 33.52 42.01 2.65

Inflammation Cox-1 82.83 36.9 43.68 2.25

Cox-2 78.06 29.3 47.02 1.73

Apoptosis Bcl-2a 39.45 24.82 13.99 0.63

Mcl-1 58.11 37.37 20.1 0.63

Bcl-xL 69.56 40.26 35.25 3.03

Gst-π 88.08 19.39 62.14 6.54

WNT pathway Wisp-1 70.96 34.16 34.92 1.87

Wisp-2 49.05 29.76 18.32 0.96

β-catenin 72.94 30.11 40.88 1.93

SHH pathway Smo 26.06 21.74 4.15 0.16

SHH 42.26 31.95 10.12 0.18

Gli-1 71.96 45.33 26.41 0.21

Gli-2 83.13 32.3 44.19 6.62

Gli-3 73.94 27 40.11 6.81

Patched 86.83 32.84 52.85 1.14

Cell to cell Keratin 7 44.51 23.34 18.64 2.55

MMP-1 89.26 20.01 66.4 2.84

CD9 95.1 40.5 39.28 13.58

Cell cycle Cyclin D1 34.45 18.08 9.75 6.61

Cdc2p34 45.6 33.71 11.64 0.23

14-3-3-σ 63.63 26.47 36.19 0.97

p21 56.01 22.49 20.77 12.74

Parafibromin 28.02 26.01 2.01 0

Rb 37.92 26.94 8.54 2.43

p27 58.78 26.18 21.67 10.92

p53 3.9 0 1 2.9

Bmi-1 69.94 33.49 29.06 7.38

AKT pathway pmTOR 20.78 17.27 3.36 0.13

pAKT 27.57 22.15 5.33 0.09

FoxO-1 80.73 33.36 46.68 0.67

224 Endocr Pathol (2012) 23:221–231

Author's personal copy

Immunohistochemical Features

The results showing the average percent positive tumor cellsand intensity for each marker are summarized in Tables 2and 3, for carcinomas and adenomas, respectively.

Angiogenesis

All benign and malignant tumors expressed markers relatedto angiogenesis. PDGFR-α (Fig. 2a) and PDGFR-β andVEGFR-2 (Fig. 2b) were present in the majority of tumorcells in all tumors, with predominantly weak to moderate

intensity. No significant differences were observed betweenadenomas and carcinomas with respect to markers related toangiogenesis. While the average percent positive tumor cellsfor PDGFR-α and PDGFR-β, and VEGFR were 90, 72, and78 % in carcinomas, respectively; they were 92, 80, and86 % in adenomas, respectively.

Epidermal Growth Factor Receptor

EGFR was detectable in a minority of tumor cells in alltumors with predominantly weak intensity in both carcino-mas and adenomas. While the average percent of tumor cells

Table 3 Immunohistochemicalresults of a parathyroid adenomaTMA

Protein Average of percent ofpositive tumor cells (%)

Distribution of intensity in positivecells (%)

Weak Moderate Strong

Growth factor receptors EGFR 8.97 6.95 1.98 0.04

PDGFR-α 92.12 33.85 57.42 0.85

PDGFR-β 80.5 42.76 36.88 0.86

VEGFR-2 85.58 31.03 51.8 2.75

Inflammation Cox-1 83.43 38.32 43.71 1.4

Cox-2 77.48 44.07 33.2 0.21

Apoptosis Bcl-2a 88.66 1.61 23.13 61.92

Mcl-1 52.2 40.86 11.2 0.14

Bcl-xL 90.27 31.74 57.9 0.63

Gst-π 89.35 24.86 59.34 5.15

WNT pathway Wisp-1 75.96 47.2 27.75 1.01

Wisp-2 45.43 26.98 17.52 1.93

β-Catenin 75.69 34.69 39.53 1.47

SHH pathway Smo 24.03 22.53 1.45 0.05

SHH 45.92 35.39 9.63 0.9

Gli-1 78.16 48.91 28.89 0.36

Gli-2 83.14 40.28 41.58 1.28

Gli-3 87.6 31.03 51.07 5.5

Patched 63.29 39.82 22.71 0.76

Cell to cell Keratin 7 53.29 21.23 28.28 3.78

MMP-1 91.95 21.24 65.53 5.08

CD9 94.25 48.95 38.57 6.73

Cell cycle Cyclin D1 35.93 22.91 8.4 4.62

Cdc2p34 67.79 36.6 30.55 0.64

14-3-3-σ 88.18 40.59 47.18 0.41

p21 78.02 34.57 28.88 14.57

Parafibromin 90.49 15.67 37.4 37.42

Rb 82.55 38.87 35.06 8.62

p27 81.63 26.69 33.38 21.56

p53 0.6 0.28 0.32 0

Bmi-1 28.53 24.84 3.61 0.08

AKT pathway pmTOR 57.04 37.21 19.78 0.05

pAKT 87.29 39.82 46.8 0.67

FoxO-1 28.53 24.84 3.61 0.08

Endocr Pathol (2012) 23:221–231 225

Author's personal copy

positive for EGFR was 17 % in carcinomas, it was 9 % inadenomas.

Inflammation

COX-1 (Fig. 2c) and COX-2 expressions were present withpredominantly moderate to weak intensity in the majority oftumor cells in all benign and malignant tumors. No signif-icant differences were observed between adenomas andcarcinomas with respect to markers related to inflammation.The average percent of tumor cells positive for COX-1 andCOX-2 was 83 and 78 % in carcinomas and 83 and 77 % inadenomas, respectively.

Apoptosis

All benign and malignant tumors exhibited variable positivityfor markers related to apoptosis. Except for anti-apoptotic Bcl-2a, no significant differences were observed between adeno-mas and carcinomas. However, while Bcl-2a was significantlyexpressed in the majority of parathyroid adenomas with amoderate to strong staining (Fig. 3b), parathyroid carcinomasshowed significant loss of staining (Fig. 3a) with a predomi-nantly weak pattern of positivity. The average percent oftumor cells positive for Bcl-2a was 89 and 39 % in adenomasand carcinomas, respectively. Anti-apoptotic Gst-π expressionwas present in the majority of benign and malignant tumorcells, with predominantly moderate staining (Fig. 2d). The

average percentages of tumor cells positive for Gst-π, Mcl-1, and Bcl-xL were 89, 52, and 90 % in adenomas and 88, 58,and 70 % in carcinomas, respectively.

WNT pathway

Wisp-1 (Fig. 2e), Wisp-2, and β-catenin expressions werepresent in all adenomas and carcinomas with predominantlyweak to moderate staining. No significant differences wereobserved between adenomas and carcinomas with respect tomarkers related to the WNT pathway. While the averagepercentages of tumor cells positive for Wisp-1, Wisp-2, andβ-catenin were 71, 49, and 73 % in carcinomas comparedwith 76, 45, and 76 % in adenomas.

Sonic Hedgehog Homolog Pathway

Members of sonic hedgehog homolog (SHH) pathway werevariably expressed in all benign and malignant tumors. Nosignificant differences were observed between adenomas andcarcinomas with respect to markers related to SHH pathway.Among them, Gli-1, Gli-2 (Fig. 2f), Gli-3, and patched werepresent in themajority of tumor cells, with predominantly weakto moderate intensity. While the average percentages of tumorcells positive for Gli-1, Gli-2, Gli-3, patched, Smo, and SHHwere 72, 83, 74, 87, 26, and 42 % in carcinomas, they were 78,83, 87, 63, 24, and 46 % in adenomas, respectively.

Cell Adhesion

All markers related to cell adhesion were expressed in themajority of benign and malignant tumor cells with predom-inantly moderate intensity. No significant differences wereobserved between adenomas and carcinomas with respect tomarkers related to cell adhesion. While CD9 revealed thehighest average percent of positive tumor cells at 95 and94 % in carcinomas and adenomas, respectively, keratin 7and MMP-1 revealed only 44 and 89 % positivity in carci-nomas and 53 and 92 % in adenomas, respectively.

Cell Cycle

Except for Cdc2p34 and 14-3-3σ that were localized to thecytoplasm, all other markers belonging to this group revealednuclear positivity. All parathyroid carcinomas showed vari-able loss of expression for markers related to cell cycle withthe exception of p53. Importantly, all carcinomas revealedsignificant loss of p27 (Fig. 3c), parafibromin (Fig. 3e), andRb (Fig. 3g); the average percentages of tumor cells positivefor p27, Rb, and parafibromin were 59, 38, and 28 % incarcinomas with predominantly weak staining, compared with82, 83, and 90 % in parathyroid adenomas (Figs. 3d, f, and h)with predominantly moderate to strong staining. The percent

Fig. 1 a Angioinvasion is characterized by thrombus adherent tointravascular tumor cells. b CD34 highlights tumor cells invadingthrough a vessel wall and thrombus adherent to intravascular tumor

226 Endocr Pathol (2012) 23:221–231

Author's personal copy

of tumor cells positive for parafibromin was as low as 5 %,and in the lowest parathyroid carcinomas, almost all stainingwas localized to vascular endothelium and other stromal cells(Fig. 3e). When a >50 % cut-off value was applied for com-bined loss of parafibromin and Rb, this panel provided 80 %sensitivity, 100 % specificity, 100 % positive predictive value,and 93 % negative predictive value for the diagnosis ofparathyroid carcinoma.

The percentages of tumor cells positive for p53 ranged from0 to 35 % in carcinomas and 0 to 5 % in adenomas; however,the averages of positivity for p53 were 0.6 and 3.9 % inadenomas and carcinomas, respectively. The average percen-tages of tumor cells positive for 14-3-3σ, Bmi, p21, cdc2p34,and cyclin D1 were 63, 70, 56, 45, and 34% in carcinomas and88, 28, 78, 68, and 36 % in adenomas, respectively.

AKT Pathway

Members of AKT pathway were variably expressed in alltumors. Except for pAKT, no significant differences wereobserved between adenomas and carcinomas with respect tomarkers related to AKT pathway. While the average percent

of tumor cells positive for FoxO-1, pmTOR, and pAKTwere 81, 21, and 28 % in carcinomas with predominantlyweak to moderate staining, these were 28, 57, and 87 % inadenomas with predominantly weak to moderate staining,respectively.

Discussion

Translational research focusing on the genomic and proteo-mic features of biological pathways involved in tumorigen-esis lead to real time applications of targeted treatmentmodalities in order to provide the best personalized patientcare. Adjuvant therapies such as radiotherapy and chemo-therapy are not particularly beneficial in the management ofparathyroid carcinoma; thus, there is a need for effectivesystemic adjuvant therapy in the management of recurrent ormetastatic parathyroid carcinoma.

In the current study, we aimed to identify the expressionprofile of proteins that can be to the targets of novel adju-vant therapies, specifically those involved in angiogenesis,inflammation, cell adhesion, cell cycle, and apoptosis along

Fig. 2 a PDGFR-α is presentin the majority of tumor cells; aparathyroid carcinoma is illus-trated in this example. bVEGFR-2 is seen in the major-ity of tumor cells, with pre-dominantly weak to moderateintensity; a parathyroid carci-noma is illustrated in this ex-ample. c COX-1 expression ispresent with predominantlymoderate intensity in the ma-jority of tumor cells; a parathy-roid carcinoma is illustrated inthis example. d Gst-π expres-sion is present in the majority oftumor cells, with predominantlymoderate staining; a parathy-roid carcinoma is illustrated inthis example. e WNT pathwaymembers are identified withpredominantly weak to moder-ate staining in parathyroid neo-plasms. In this picture, Wisp-1expression is illustrated in aparathyroid carcinoma. f Mem-bers of SHH pathway are vari-ably expressed in parathyroidneoplasms. Among them Gli-2reveals cytoplasmic and nuclearreactivity in the majority ofparathyroid carcinoma cells

Endocr Pathol (2012) 23:221–231 227

Author's personal copy

with some markers of the SHH, AKT, and WNT signaltransduction pathways. Tyrosine kinase receptor inhibitorshave shown efficacy in patients with squamous cell carci-nomas of the head and neck [11], as well as lung [12] andcolonic adenocarcinomas [13]. Unlike EGFR expression,our results highlight that VEGFR-2, PDGFR-α, and

PDGFR-β were predominantly expressed in benign andmalignant parathyroid neoplasms. These findings may pro-vide the rationale for the use of future targeted therapies,such as sunitinib, bevacizumab and pazopanib which havebeen shown to be effective in the treatment of kidney, lung,and breast malignancies [14–16].

Fig. 3 Parathyroid carcinomasreveal loss of expression of Bcl-2a (a), p27 (c), parafibromin(e), and Rb (g). The percent oftumor cells positive for parafi-bromin was as low as 5 % andin the lowest cases, almost allstaining was localized to vas-cular endothelium and otherstromal cells (e). In contrast, themajority of tumor cells in para-thyroid adenomas were positivefor Bcl-2a (b), p27 (d), parafi-bromin (f), and Rb (h)

228 Endocr Pathol (2012) 23:221–231

Author's personal copy

The high percent of tumor cells in all parathyroid carcino-mas in this series that were positive for both COX-1 andCOX-2 may be of clinical interest. Multiple studies havedemonstrated anti-tumor effects of nonsteroidal anti-inflammatory drugs [17–19]. Our preliminary results suggestthat there may be value in determining whether the adminis-tration of nonsteroidal anti-inflammatory drugs may provide abenefit in patients with metastatic parathyroid carcinoma.

It is well-known that the hedgehog signaling cascade isone of the main pathways involved in human embryogene-sis and organogenesis. Mutations that inappropriately reac-tivate this normally quiescent pathway have been linked tothe development of skin, colon, and lung carcinoma [20].Manipulation of the hedgehog pathway by selective agonistsand antagonists, such as cyclophosphamide, may alter thecourse of malignancy [20]. New research in prostate cancerhas shown that blocking Gli-2 expression with antisenseoligonucleotides in vitro leads to a significant delay in tumorprogression and an increase in paclitaxel chemosensitivity[21]. Similar results were achieved with antisense oligonu-cleotides in the treatment of mantle cell lymphoma. Block-ing Gli-2 and Patched leads to decreased cell proliferation inmantle cell lymphoma cells and also to a significant increasein tumor cell susceptibility to doxorubicin [22]. Our resultsindicate that markers related to SHH pathway are variablyexpressed in benign and malignant parathyroid neoplasms.Although there is no significant difference between adeno-mas and carcinomas, the high average percent of tumor cellspositive for Gli-2 and patched in our series may be of furtherclinical interest to target this active pathway.

FoxO transcription factors are downstream targets of thePI3K and AKT pathway. Activation of FoxOs by AKT resultsin inhibition of transcriptional roles of FoxOs and contributesto cell survival, growth and proliferation. Emerging evidencesuggests involvement of FoxOs in diverse intracellular path-ways in a number of physiological as well as pathologicalconditions including cancer [23]. Based on current knowledgeand studies from the past decade, the development of novelagents which specifically activate FoxO members could beuseful in the prevention and treatment of epithelial malignan-cies in particular [24]. Although the average percent of posi-tive tumor cells was as high as 81 % in parathyroidcarcinomas, tumors with a low percent of positive tumor cellsmay be important for the selection of patients who mightbenefit from agents that activate FoxO-1.

CD9 and MMP-1 were also significantly expressed in themajority of both benign and malignant tumor cells. CD9 andMMP-1 are involved in cell motility, invasion, and metas-tasis [25–27]. Selective targeting of CD9 and MMP-1 withspecific antibodies and small interfering RNA has beendemonstrated to trigger apoptosis in chemoresistant lungcarcinoma cells [26] and can suppress the invasion of tumorcells in a three-dimensional collagen gel [27].

Human Gst-π plays a role in the detoxification of platinum-based cytotoxic agents and can promote resistance to plati-nums. Gst-π has also been shown to modulate the activity ofJNKs by inhibiting apoptosis and by causing cell proliferationand tumor growth [28]. Furthermore, transfection of Gst-πantisense cDNA increased the sensitivity of a cancer cell lineto various anticancer agents including adriamycin, etoposide,and melphalan [28]. A new targeted agent TLK-286 hasdemonstrated cytotoxic activity on its own [29], but throughinhibition of Gst-π, it can enhance the activity of cisplatin incancer cells [30]. Our results highlight that Gst-π is expressedin the majority of benign and malignant parathyroid prolifer-ations; thus, a Gst-π inhibitor such as TLK-286 may holdpromise in patients with parathyroid carcinoma.

There has been considerable progress in elucidating thepathogenesis of parathyroid carcinomas [31–33]. Inactiva-tion of the Rb gene and allelic loss and abnormal p53 proteinexpression in parathyroid carcinomas implicates the role ofRb and p53 in the pathogenesis of a subset of these tumors[32, 33]. Germline mutations in the HRPT2 (also known asCDC73) tumor suppressor gene, which encodes the proteinparafibromin, are associated with the hyperparathyroidism-jaw tumor syndrome (HPT-JT), in which the afflicted mem-bers often present with parathyroid carcinoma [31, 34]. Inaddition, somatic HRPT2 mutations are also found in 67–100 % of sporadic parathyroid carcinomas [31, 34]. Thus,loss of parafibromin is regarded as a diagnostic marker ofparathyroid carcinoma. Usually, mutation results in globalloss of parafibromin expression, which is regarded as astrong indicator of a diagnosis of parathyroid carcinoma,however, there is no consensus on the extent of parafibro-min loss required for the diagnosis of parathyroid carcino-ma. Moreover, parafibromin expression can still becompatible with HRPT2 mutation if the mutation does notaffect the antigenicity or level of expression of the protein.Gill et al. reported complete absence of nuclear staining in75 % of HPT-JT-related tumors and 73 % of sporadicparathyroid carcinomas with focal weak staining in theremaining 25 % of HPT-JT tumors and 18 % of sporadicparathyroid carcinomas [34]. In their study, only one para-thyroid carcinoma exhibited diffuse strong nuclear expres-sion of parafibromin [34]. Furthermore, Krebs et al. reportedthat HRPT2 inactivation is not an important participant inthe pathogenesis of typical sporadic parathyroid adenomas[35]. p27kip1 (p27) is a cyclin-dependent kinase inhibitorthat helps regulate the transition from the G1 to the S-phaseof the cell cycle [36]. Loss of p27 expression has also beenimplicated in the pathogenesis of parathyroid carcinomas[36]. In our series, all cases fulfilled conventional WHOcriteria of parathyroid carcinoma [37], and revealed variableloss of parafibromin. While the percent of tumor cells pos-itive for parafibromin was as low as 5 % in our series, thetumors with the lower numbers exhibited staining mainly in

Endocr Pathol (2012) 23:221–231 229

Author's personal copy

stromal elements of the tumor. In addition, all cases revealedvariable loss of Rb and p27 expression along with parafi-bromin. Moreover, our results indicate that when a >50 %cut-off value is applied for combined loss of parafibrominand Rb, this panel has 80 % sensitivity, 100 % specificity,100 % positive predictive value, and 93 % negative predic-tive value for the diagnosis of parathyroid carcinoma. An-other interesting finding in this study is the significant lossof Bcl-2a expression in parathyroid carcinomas while para-thyroid adenomas are strongly positive for Bcl-2a. Althoughthe underlying cellular mechanisms of this finding remain tobe explained, the loss of Bcl-2a expression in parathyroidcarcinomas is of interest.

Our results indicate that the use of a panel that includesBcl-2a, parafibromin, Rb, and p27 may be helpful in theassessment of atypical parathyroid neoplasms that lack un-equivocal angioinvasion, perineural invasion, and gross lo-cal invasion into adjacent organs or metastasis. To ourknowledge, this is also the first time that sonic hedgehogpathway proteins and COX-1/2 have been reported to beexpressed in parathyroid neoplasms. Although the majorityof other markers studied are also expressed in both benignand malignant parathyroid neoplasms, we have identifiedseveral potentially important target proteins related to an-giogenesis and cell proliferation along with COX-1/2, Gst-π, and members of sonic hedgehog pathway that may betherapeutically targeted in parathyroid carcinoma. We rec-ognize that these results are preliminary and further in vitroand in vivo data are required, a successful outcome of aclinical trial directed against these novel targets would pro-vide much needed systemic adjuvant treatment for patientswith metastatic parathyroid carcinoma.

Acknowledgments The preliminary results of this paper were pre-sented in part at the 101st United States and Canadian Academy ofPathology Annual Meeting, 17–23 March 2012, Vancouver, BC,Canada.

References

1. Hundahl SA, Fleming ID, Fremgen AM, Menck HR (1999) Twohundred eighty-six cases of parathyroid carcinoma treated in theU.S. between 1985 and 1995: a National Cancer Data Base Report.The American College of Surgeons Commission on Cancer andthe American Cancer Society. Cancer 86:538–544.

2. Erovic BM, Goldstein DP, Kim D, Mete O, Brierly J, Tsang R,Freeman JL, Asa SL, Rotstein L, Irish JC (2012). Parathyroidcancer: outcome analysis of 16 patients treated at the PrincessMargaret Hospital. Head Neck 2012, doi:10.1002/hed.22908.

3. Lee PK, Jarosek SL, Virnig BA, Evasovich M, Tuttle TM (2007)Trends in the incidence and treatment of parathyroid cancer in theUnited States. Cancer 109:1736–1741.

4. Sandelin K, Auer G, Bondeson L, Grimelius L, Farnebo LO (1992)Prognostic factors in parathyroid cancer: a review of 95 cases.World J Surg 16:724–731.

5. Chow E, Tsang RW, Brierley JD, Filice S (1998) Parathyroidcarcinoma—the Princess Margaret Hospital experience. Int JRadiat Oncol Biol Phys 41:569–572.

6. Munson ND, Foote RL, Northcutt RC, Tiegs RD, Fitzpatrick LA,Grant CS, van Heerden JA, Thompson GB, Lloyd RV (2003)Parathyroid carcinoma: is there a role for adjuvant radiation ther-apy? Cancer 98:2378–2384.

7. Betea D, Bradwell AR, Harvey TC, Mead GP, Schmidt-Gayk H,Ghaye B, Daly AF, Beckers A. (2004) Hormonal and biochemicalnormalization and tumor shrinkage induced by anti-parathyroidhormone immunotherapy in a patient with metastatic parathyroidcarcinoma. J Clin Endocrinol Metab 89:3413–3420.

8. Schott M, Feldkamp J, Schattenberg D, Krueger T, Dotzenrath C,Seissler J, Scherbaum WA (2000) Induction of cellular immunityin a parathyroid carcinoma treated with tumor lysate-pulsed den-dritic cells. Eur J Endocrinol 142:300–306.

9. Schott M, Feldkamp J, Schattenberg D, Seissler J, Scherbaum WA(1999) Dendritic cell immuno-therapy in disseminated parathyroidcarcinoma. Lancet 353:1188–1189.

10. Mete O, Asa SL (2011) Pathological definition and clinical signif-icance of vascular invasion in thyroid carcinomas derived fromfollicular epithelium. Modern Pathology 24:1545–1552.

11. Machiels JP, Schmitz S (2011) Molecular-targeted therapy of headand neck squamous cell carcinoma: beyond cetuximab-based ther-apy. Curr Opin Oncol 23:241–248.

12. Cataldo VD, Gibbons DL, Pérez-Soler R, Quintás-Cardama A(2011) Treatment of non-small-cell lung cancer with erlotinib orgefitinib. N Engl J Med 364:947–955.

13. Chee CE, Sinicrope FA (2010) Targeted therapeutic agents forcolorectal cancer. Gastroenterol Clin North Am 39:601–613.

14. Hsu JY, Wakelee HA (2009) Monoclonal antibodies targetingvascular endothelial growth factor: current status and future chal-lenges in cancer therapy. BioDrugs 23:289–304.

15. Chow LQ, Eckhardt SG (2007) Sunitinib: from rational design toclinical efficacy. J Clin Oncol 25:884–896.

16. Bukowski RM (2010) Pazopanib: a multikinase inhibitor withactivity in advanced renal cell carcinoma. Expert Rev AnticancerTher 10:635–645.

17. Perrone MG, Scilimati A, Simone L, Vitale P (2010) SelectiveCOX-1 inhibition: a therapeutic target to be reconsidered. CurrMed Chem 17:3769–805.

18. Kao J, Sikora AT, Fu S (2009). Dual EGFR and COX-2 inhibitionas a novel approach to targeting head and neck squamous cellcarcinoma. Curr Cancer Drug Targets 9:931–937.

19. Moreira L, Castells A (2010). Cyclooxygenase as a target for colo-rectal cancer chemoprevention. Curr Drug Targets 12:1888–1894.

20. Peukert S, Miller-Moslin K (2010) Small-molecule inhibitors ofthe hedgehog signaling pathway as cancer therapeutics. ChemMedChem 5:500–512.

21. Narita S, So A, Ettinger S, Hayashi N, Muramaki M, Fazli L, Kim Y,GleaveME (2008)GLI2 knockdown using an antisense oligonucleotideinduces apoptosis and chemosensitizes cells to paclitaxel in androgen-independent prostate cancer. Clin Cancer Res 14:5769–5777.

22. Hegde GV, Munger CM, Emanuel K, Joshi AD, Greiner TC,Weisenburger DD, Vose JM, Joshi SS (2008) Targeting of sonichedgehog-GLI signaling: a potential strategy to improve therapyfor mantle cell lymphoma. Mol Cancer Ther 7:1450–1460.

23. Fu Z, Tindall DJ. FOXOs, cancer and regulation of apoptosis(2008) Oncogene 27:2312–2319.

24. Reagan-Shaw S, Ahmad N. The role of Forkhead-box Class O(FoxO) transcription factors in cancer: a target for the managementof cancer (2007) Toxicol Appl Pharmacol 224:360–368.

25. Erovic BM, Pammer J, Hollemann D, Woegerbauer M, Geleff S,Fischer MB, Burian M, Frommlet F, Neuchrist C (2003) Motility-related protein-1/CD9 expression in head and neck squamous cellcarcinoma. Head Neck 25:848–857.

230 Endocr Pathol (2012) 23:221–231

Author's personal copy

26. Kohmo S, Kijima T, Otani Y, Mori M, Minami T, Takahashi R,Nagatomo I, Takeda Y, Kida H, Goya S, Yoshida M, Kumagai T,Tachibana I, Yokota S, Kawase I (2010) Cell surface tetraspaninCD9 mediates chemoresistance in small cell lung cancer. CancerRes 70:8025–8035.

27. Takino T, Saeki H, Miyamori H, Kudo T, Sato H (2007) Inhibitionof membrane-type 1 matrix metalloproteinase at cell-matrix adhe-sions. Cancer Res 67:11621–11629.

28. Thévenin AF, Zony CL, Bahnson BJ, Colman RF (2011) GST pimodulates JNK activity through a direct interaction with JNKsubstrate, ATF2. Protein Sci 20:834–848.

29. Edelman MJ (2006) Novel cytotoxic agents for non-small cell lungcancer. J Thorac Oncol 1:752–755.

30. Singh S, Okamura T, Ali-Osman F (2010) Serine phosphorylationof glutathione S-transferase P1 (GSTP1) by PKCα enhancesGSTP1-dependent cisplatin metabolism and resistance in humanglioma cells. Biochem Pharmacol 80:1343–1355.

31. Shattuck TM, Välimäki S, Obara T, Gaz RD, Clark OH, ShobackD, Wierman ME, Tojo K, Robbins CM, Carpten JD, Farnebo LO,Larsson C, Arnold A (2003) Somatic and germ-line mutations ofthe HRPT2 gene in sporadic parathyroid carcinoma. N Engl J Med349:1722–1729.

32. Cryns VL, Rubio MP, Thor AD, Louis DN, Arnold A (1994) p53abnormalities in human parathyroid carcinoma. J Clin EndocrinolMetab 78:1320–1324.

33. Cryns VL, Thor A, Xu HJ, Hu SX, Wierman ME, Vickery AL Jr,Benedict WF, Arnold A (1994) Loss of the retinoblastoma tumor-suppressor gene in parathyroid carcinoma. N Engl J Med 330:757–761.

34. Gill AJ, Clarkson A, Gimm O, et al. (2006) Loss of nuclearexpression of parafibromin distinguishes parathyroid carcinomasand hyperparathyroidism-jaw tumor (HPT-JT) syndrome-relatedadenomas from sporadic parathyroid adenomas and hyperplasias.Am J Surg Pathol 30:1140–1149.

35. Krebs LJ, Shattuck TM, Arnold A (2005) HRPT2 mutationalanalysis of typical sporadic parathyroid adenomas. J Clin Endo-crinol Metab 90:5015–5017.

36. Erickson LA, Jin L, Wollan P, Thompson GB, van Heerden JA,Lloyd RV (1999). Parathyroid hyperplasia, adenomas, and carci-nomas: differential expression of p27Kip1 protein. Am J SurgPathol 23:288–295.

37. DeLellis RA, Lloyd RV, Heitz PU, editors (2004) World HealthOrganization Classification of Tumours. Pathology and genetics oftumours of endocrine organs. IARC Press, Lyon.

Endocr Pathol (2012) 23:221–231 231

Author's personal copy