Utility of brachyury in distinction of chordoma from cytomorphologic mimics in fine-needle aspiration and core needle biopsy

ORIGINAL ARTICLES

Utility of Brachyury in Distinctionof Chordoma FromCytomorphologic Mimics inFine-Needle Aspiration and CoreNeedle BiopsyVickie Y. Jo, M.D.,* Jason L. Hornick, M.D., PhD., andXiaohua Qian, M.D., PhD.

Chordoma is a neoplasm of notochordal differentiation that typi-cally occurs in the axial skeleton. Accurate diagnosis is thera-peutically important but can be challenging, especially in fine-needle aspiration (FNA) and core needle biopsy (CNB). Immu-nohistochemistry for the transcription factor brachyury (T) hasrecently proven diagnostically useful in whole-tissue sections.Our aim was to compare brachyury performance with conven-tional markers (S-100, EMA, keratin) and to evaluate its utilityin distinguishing chordoma from cytomorphologic mimics. Bra-chyury immunohistochemistry was performed on chordoma (8FNA, 12 CNB), chondrosarcoma (10 FNA), and metastaticmucinous adenocarcinoma (12 FNA). Immunohistochemistryperformed at the time of diagnosis was also reviewed. Bra-chyury was positive in 17 (85%) cases of chordoma and typi-cally showed moderate-to-strong nuclear staining. Of five sets ofconcurrent FNA and CNB, four pairs were positive for bra-chyury in both samples and one pair was positive for brachyuryin the CNB and negative in the cell block. S-100, EMA, and ker-atin stains were available for 13 chordomas: 9 (69%) cases(including the 3 negative for brachyury) were positive for S-100and keratin or EMA; 4 cases were keratin positive but S-100negative. No nuclear brachyury staining was seen in chondro-sarcoma or adenocarcinoma, though two adenocarcinomasshowed cytoplasmic staining. Brachyury separates chordomafrom cytomorphologic mimics with high sensitivity and specific-

ity in small biopsies. As a single test, brachyury has higher sen-sitivity than a combined panel of S-100 and epithelial markers.When added to the conventional panel, brachyury increases sen-sitivity to 100% without sacrificing specificity. Diagn.Cytopathol. 2014;42:647–652. VC 2014 Wiley Periodicals, Inc.

Key Words: chordoma; brachyury; fine-needle aspiration;immunohistochemistry

Chordoma is a neoplasm of notochordal differentiation

that most often occurs in the axial skeleton and typically

affects adults (older than 50 years) with a slight male

predilection. It is considered to have low-to-intermediate

malignant potential, and shows both locally aggressive

behavior and a propensity for recurrence and distant

metastasis. For skull-based chordomas, the 5-year and 10-

year survival rates are 70% and 45%, respectively.1

Separation from mimics such as chondrosarcoma and

metastatic mucinous adenocarcinoma is therapeutically

important but diagnostically challenging, especially with

the rising use of minimally invasive imaging-guided

biopsy methods such as fine-needle aspiration (FNA) and

core needle biopsy (CNB).2

Chordoma is characterized by epithelioid cells with

bland round nuclei and small or inconspicuous nucleoli,

with cytoplasm ranging from eosinophilic and granular to

abundant and vacuolated (so-called physaliphorous cells).

The tumor cells are arranged in sheets and cords or are

present singly, embedded in an abundant chondromyxoid

extracellular matrix (Fig. 1). The relative proportion of

epithelioid cells and stroma may be variably present both

within a given lesion and between tumors, which can lead

These results were presented in part at the 101st Annual Meeting of theUS-Canadian Academy of Pathology, Vancouver, March 2012.

Department of Pathology, Brigham and Women’s Hospital and Har-vard Medical School, Boston, Massachusetts

*Correspondence to: Vickie Jo, MD, Department of Pathology, Brig-ham and Women’s Hospital, 75 Francis Street, Boston, MA 02115,USA. E-mail: [email protected]

Received 3 May 2013; Revised 5 October 2013; Accepted 9 January2014

DOI: 10.1002/dc.23100Published online 19 February 2014 in Wiley Online Library

(wileyonlinelibrary.com).

VC 2014 WILEY PERIODICALS, INC. Diagnostic Cytopathology, Vol. 42, No 8 647

to diagnostic challenges, especially in needle biopsies.2,3

Owing to its notochordal differentiation, tumor cells show

expression of S-100 and epithelial markers (cytokeratin

and EMA). Brachyury (T), a nuclear transcription factor

required during embryogenesis for posterior mesoderm

formation and axial development, is expressed in chor-

doma4–6; while studies report chromosomal aberrations of

the brachyury gene locus 6q27 in sporadic and hereditary

chordomas, the tumorigenic role of brachyury remains

unclear.6 Immunohistochemistry for brachyury has

recently proven to be diagnostically useful in whole-

tissue sections (or tissue microarray (TMA) constructed

from whole-tissue sections) with sensitivity ranging from

90 to 100% and specificity of 100%.7–9 To date, the sen-

sitivity and specificity of brachyury for chordoma in small

biopsy samples such as FNA and CNB has not been eval-

uated. Because positive staining for brachyury can be

focal, the utility of this marker in small biopsies needs to

be validated.

The aim of this study is to compare brachyury staining

in chordoma with the conventional immunohistochemical

panel (S-100, EMA, and keratin) in FNA and CNB sam-

ples, and to assess the utility of brachyury in distinguishing

chordoma from its cytomorphologic mimics, chondrosar-

coma, and metastatic mucinous adenocarcinoma.

Materials and Methods

Cases were retrieved from cytology, surgical pathology,

and consult archives of the Department of Pathology,

Brigham and Women’s Hospital, Boston, MA. All FNA

cases had available cell blocks for immunohistochemical

studies. In total, 20 chordoma cases (8 FNA and 12

CNB), 10 FNAs of chondrosarcoma, and 12 FNAs of

metastatic mucinous adenocarcinoma were selected for

study. All accompanying immunohistochemical studies

performed at the time of diagnosis were reviewed. Of the

chordoma cases, five had concurrent FNA and CNB

specimens. All cases of chordoma and chondrosarcoma

were confirmed by correlation with radiologic studies

and/or resection specimens. All cases of metastatic muci-

nous adenocarcinoma had confirmed primary malignan-

cies and were anatomically distributed as follows: upper

gastrointestinal (2), lower gastrointestinal (6), lung (2),

and pancreatobiliary (2).

All FNA and CNB performed at our institution were

percutaneous and guided via computed tomography.

Both alcohol-fixed, Papanicolaou stained and air dried,

Romanowsky stained smears were prepared from FNA.

Needle rinses and/or dedicated passes were placed in

Roswell Park Memorial Institute (RPMI) medium, after

which a cell block was made by the addition of throm-

bin and plasmin and subsequently fixed in 10% forma-

lin. CNB were placed in 10% formalin, and both cell

blocks and CNB were paraffin-embedded for H&E

stains.

Immunohistochemistry was performed on all cases

using a rabbit polyclonal antibody to brachyury (Santa

Cruz Biotechnology, Santa Cruz, CA; 1:400 dilution)

after antigen retrieval in citrate buffer (pH 6.0) using a

pressure cooker. Nuclear staining (of any intensity) in at

Fig. 1. Cytologic features of chordoma on aspirate smear: (a) epithelioid cells with eosinophilic cytoplasm embedded in a chondromyxoid matrix,403; (b) epithelioid cells with abundant cytoplasmic vacuolization, 403. (Romanowsky stained, air dried smears).

Diagnostic Cytopathology DOI 10.1002/dc

JO ET AL.

648 Diagnostic Cytopathology, Vol. 42, No 8

least 5% of tumor cells was recorded as a positive result.

Immunohistochemistry (S-100, EMA, pan-keratin, AE1/

AE3) performed at the time of diagnosis was also

reviewed. Immunohistochemistry for S-100 (polyclonal;

Dako, Carpinteria, CA; 1:3,000 dilution with no pretreat-

ment) and AE1/AE3 (Dako; 10 min proteinase digestion;

1:200 dilution) was performed on all chondrosarcoma

cases.

Results

The chordoma cases included 15 axial skeleton primary

tumors and 5 metastases (spine, pelvis, lung). Nuclear

brachyury staining was observed in 85% (17/20) of chor-

doma cases (5/8 cell blocks; 12/12 CNB). Fourteen chor-

domas showed multifocal nuclear staining of moderate to

strong intensity (Figs. 2a and b), and three showed weak

staining. There were two cell blocks that were sparsely

cellular in which brachyury staining of moderate intensity

was observed (Figs. 2c and d). There were five sets of

concurrent FNA and CNB. Four pairs were brachyury

positive in both samples, and one pair was brachyury

positive in the CNB and negative in the cell block. S-

100, EMA, and keratin stains were available for 13 cases;

9 cases (69%) were positive for S-100 and keratin and/or

EMA, including the 3 brachyury-negative chordomas.

Four cases were keratin positive but S-100 negative.

Table I summarizes the immunohistochemical results in

the chordoma cases, and Table II summarizes the results

according to specimen type (cell block and CNB).

Brachyury was negative in all cases of chondrosar-

coma (0/10), and as expected S-100 was positive in all

cases and AE1/AE3 was uniformly negative (Fig. 3).

For metastatic mucinous adenocarcinoma, no nuclear

brachyury staining was observed (0/12), though two

cases showed cytoplasmic staining (Fig. 4). A compari-

son of the immunohistochemical results in chordoma,

chondrosarcoma, and metastatic mucinous adenocarci-

noma are presented in Table III. The positive predictive

values for brachyury and the conventional panel were

both 100%; however the negative predictive value was

higher for brachyury (88%) in comparison to the con-

ventional panel (85%).

Fig. 2. Brachyury shows strong nuclear staining in chordoma. H&E stained cell block section of chordoma (a, 403) showing strong nuclear brachyurystaining (b, 403). Tumor cells in a more sparsely cellular cell block (c, 203) showed weak-to-moderate nuclear staining for brachyury (d, 403).


USE OF BRACHYURY TO IDENTIFY CHORDOMA

Diagnostic Cytopathology, Vol. 42, No 8 649

Discussion

We have confirmed the diagnostic utility of immunohisto-

chemistry for brachyury for chordoma in FNA and CNB,

similar to previous findings using whole-tissue sections

(or TMAs derived from whole sections). As a single test,

brachyury has greater sensitivity (85%) than the conven-

tional panel of S-100 and keratin and/or EMA (69%).

Brachyury is highly specific for chordoma, and, when

added to the conventional panel, brachyury increases sen-

sitivity to 100%. No nuclear staining was observed in

chondrosarcoma or metastatic mucinous adenocarcinoma,

which are two other chief diagnostic considerations in

small biopsies.

Nuclear brachyury staining in chordoma was observed

to be moderate-to-strong in intensity, even in two sparsely

cellular samples, and showed comparable sensitivity

(85%) to prior studies.7–9 Using the conventional immu-

nohistochemical panel of S-100 and epithelial markers,

we found a lower sensitivity for chordoma in comparison

to brachyury. While keratin and/or EMA was positive in

all chordoma cases tested, S-100 was negative in 31% of

cases (4/13). Lower frequency of S-100 positivity in chor-

doma has been previously reported, which ranges in sen-

sitivity from 85.7 to 100%.7,8,10,11 Similar to prior

reports, we found higher sensitivity for the epithelial

markers, with reported sensitivities ranging from 97 to

100% for keratin7,8,10,11 and 95 to 100% for EMA.7,8,10,11

Only one prior study using whole-tissue sections assessed

brachyury in addition to the conventional markers,7 which

reported 98% sensitivity and 100% specificity for the

combined panel of brachyury, S-100, and AE1/AE3; nota-

bly in their study all cases were S-100 positive and AE1/

AE3 negativity accounted for the decreased sensitivity.

Our findings support that brachyury separates chordoma

from chondrosarcoma in small biopsies, similar to prior

studies using whole-tissue sections.5,7,10 An immunohisto-

chemical panel including brachyury and the conventional

Table I. Comparison Between Brachyury Expression and ConventionalMarkers in Chordoma

BrachyuryS-100 and

cytokeratin/EMABrachyury or S-100

and cytokeratin/EMA

No. Positive 17/20 9/13 20/20Sensitivity 85% 69% 100%

Table II. Immunohistochemical Results for FNA/Cell Block and CoreNeedle Biopsy Samples

Specimen Brachyury S-100Cytokeratin

or EMA

FNA/cell block (positive/total) 5/8 4/5 5/5Core needle biopsy (positive/total) 12/12 7/8 8/8

Fig. 3. Chondrosarcoma can show morphologic overlap with chordoma. Aspirate smear of chondrosarcoma showing abundant chondromyxoid matrixwith embedded chondrocytes showing cytoplasmic vacuolization (a, Romanowsky stain, 403). Cell block (b, 403). Brachyury staining was absent inall cases of chondrosarcoma (c, 403).


JO ET AL.


markers can be useful, as chondrosarcomas are typically

S-100 positive and generally negative for keratins (though

focal staining can be observed in surgical resection speci-

mens). In our study, mucinous adenocarcinoma showed

no nuclear staining for brachyury, similar to previous

findings which included small numbers of adenocarci-

noma cases.8,10 Adding brachyury to the conventional

panel is advantageous given that a proportion of chordo-

mas are S-100 negative, and keratin staining would not

distinguish chordoma from adenocarcinoma. Attention to

clinical history and diligent interpretation of staining pat-

terns is mandatory, as we found cytoplasmic staining for

brachyury in two cases of adenocarcinoma.

Brachyury expression apart from chordoma is limited,

and appears to be absent in the majority of entities that

enter the differential diagnosis. Both conventional clear

cell renal cell carcinoma9,10 and chordoid meningioma8

show absent brachyury expression. Brachyury is also neg-

ative in myxopapillary ependymoma, a diagnostic consid-

eration in sacral locations.5 Myxopapillary ependymoma

is positive for S-100 and GFAP, and can show at least

focal cytokeratin positivity,12 thus brachyury is useful in

the distinction between myxopapillary ependymoma and

chordoma. Myoepithelial tumors of soft tissue typically

show heterogeneous expression of S-100, GFAP, keratin,

EMA, and SMA, and may mimic chordoma when a myx-

oid stroma is predominant. Brachyury staining is absent

in soft tissue mixed tumor/myoepithelioma,13 and careful

histologic examination will also help distinguish myoepi-

thelial neoplasms from chordoma. Lastly, the survey stud-

ies examining brachyury expression in large numbers of

Fig. 4. Metastatic mucinous adenocarcinoma on aspirate smear (a, Romanowsky stain, 403) and cell block (b, 403) can mimic chordoma. Brachyurystaining was absent in the majority of cases (c, 403) but showed cytoplasmic staining in two cases (d, 403).

Table III. Comparison Between Brachyury and Conventional Markersin Chordoma, Chondrosarcoma, and Metastatic MucinousAdenocarcinoma

Brachyury S-100Cytokeratinand/or EMA

S-100 andcytokeratin/EMA

Chordoma 17/20 9/13 13/13 9/13Chondrosarcoma 0/10 10/10 0/10 0/10Metastatic mucinous

adenocarcinoma0a/12 N/A 6/6 N/A

aTwo cases of adenocarcinoma showed cytoplasmic staining.


USE OF BRACHYURY TO IDENTIFY CHORDOMA

Diagnostic Cytopathology, Vol. 42, No 8 651

neoplasms have reported an absence of staining in the

majority of entities tested, including osteosarcoma, syno-

vial sarcoma, chondromyxoid fibroma, chondroblastoma,

extraskeletal myxoid chondrosarcoma, myxoid liposar-

coma, rhabdomyosarcoma, pleomorphic adenoma, mucoe-

pidermoid carcinoma, melanoma, and urothelial, breast,

and ovarian carcinoma.6,10,12 While one study reported bra-

chyury expression in non-neoplastic testis12 and a second

study found positive staining in a small subset of germ cell

tumors,10 a subsequent larger series reported no staining in

111 germ cell tumors and non-neoplastic testis samples.9

In conclusion, brachyury distinguishes chordoma from

cytomorphologic mimics with high sensitivity and specific-

ity in small biopsy samples. As a single test, brachyury has

higher sensitivity (85%) than a combined panel of S-100

and epithelial markers (69%), and may be useful in sam-

ples with sparse cellularity. While both brachyury and the

conventional panel had positive predictive values of 100%,

brachyury had higher negative predictive value (88%) than

the conventional panel (84%). When added to the conven-

tional panel, brachyury increases the sensitivity to 100%;

thus the use of an all-inclusive immunohistochemical panel

of S-100, epithelial markers, and brachyury can be consid-

ered a new gold standard for diagnosis. Cytoplasmic stain-

ing in mucinous metastatic adenocarcinoma is a potential

pitfall when interpreting brachyury in samples with scant

cellularity. The pragmatic use of immunohistochemical

panels directed by the differential diagnosis and clinical

context is recommended.

References

1. Rosenberg AE, Nielsen GP, Keel SB, et al. Chondrosarcoma of thebase of the skull: A clinicopathologic study of 200 cases withemphasis on its distinction from chordoma. Am J Surg Pathol1999;23:1370–1378.

2. Layfield LJ. Cytologic differential diagnosis of myxoid and muci-nous neoplasms of the sacrum and parasacral soft tissues. DiagnCytopathol 2003;28:264–271.

3. Kay PA, Nascimento AG, Unni KK, Salam~ao DR. Chordoma.Cytomorphologic findings in 14 cases diagnosed by fine needleaspiration. Acta Cytol 2003;47:202–208.

4. Schwab JH, Boland PJ, Agaram NP, et al. Chordoma and chondro-sarcoma gene profile: Implications for immunotherapy. CancerImmunol Immunother 2009;58:339–349.

5. Vujovic S, Henderson S, Presneau N, et al. Brachyury, a crucialregulator of notochordal development, is a novel biomarker forchordomas. J Pathol 2006;209:157–165.

6. Nibu Y, Jos�e-Edwards DS, Di Gregario A. From notochord forma-tion to hereditary chordoma: The many roles of Brachyury. BiomedRes Int 2013, in press.

7. Oakley GJ, Fuhrer, Seethala RJ. Brachyury, SOX-9, and podopla-nin, new markers in the skull base chordoma vs chondrosarcomadifferential: A tissue microarray-based comparative analysis. ModPathol 2008;21:1461–1469.

8. Sangoi AR, Dulai MS, Beck AH, Brat DJ, Vogel H. Distinguishingchordoid meningiomas from their histologic mimics: An immuno-histochemical evaluation. Am J Surg Pathol 2009;33:669–681.

9. Sangoi AR, Karamchandani J, Lane B, et al. Specificity of bra-chyury in the distinction of chordoma from clear cell renal cell car-cinoma and germ cell tumors: A study of 305 cases. Mod Pathol2011;24:425–429.

10. Jambhekar NA, Rekhi B, Thorat K, Dikshit R, Agrawal M, Puri A.Revisiting chordoma with brachyury, a “new age” marker: Analysisof a validation study of 51 cases. Arch Pathol Lab Med 2010;134:1181–1187.

11. Cho HY, Lee M, Takei H, Dancer J, Ro JY, Zhai Q. Immunohisto-chemical comparison of chordoma with chondrosarcoma, myxopa-pillary ependymoma, and chordoid meningioma. ApplImmunohistochem Mol Morphol 2009;17:131–138.

12. Lamzabi I, Arvanitis LD, Reddy VB, Bitterman P, Gattuso P.Immunophenotype of myxopapillary ependymomas. Appl Immuno-histochem Mol Morphol 2013;21:485–489.

13. Tirabosco R, Mangham DC, Rosenberg AE, et al. Brachyury expres-sion in extra-axial skeletal and soft tissue chordomas: A marker thatdistinguishes chordoma from mixed tumor/myoepithelioma/para-chordoma in soft tissue. Am J Surg Pathol 2008;32:571–580.


JO ET AL.


Documents

Utility of brachyury in distinction of chordoma from cytomorphologic mimics in fine-needle aspiration and core needle biopsy