Brachyury, SOX-9, and podoplanin, new markers in the skull base chordoma vs chondrosarcoma differential: a tissue microarray-based comparative analysis

Abstract

The distinction between chondrosarcoma and chordoma of the skull base/head and neck is prognostically important; however, both have sufficient morphologic overlap to make delineation difficult. As a result of gene expression studies, additional candidate markers have been proposed to help in separating those entities. We sought to evaluate the performance of new markers: brachyury, SOX-9, and podoplanin alongside the more traditional markers glial fibrillary acid protein, carcinoembryonic antigen, CD24, and epithelial membrane antigen. Paraffin blocks from 103 skull base/head and neck chondroid tumors from 70 patients were retrieved (1969-2007). Diagnoses were made based on morphology and/or whole-section immunohistochemistry for cytokeratin and S100 protein yielding 79 chordomas (comprising 45 chondroid chordomas and 34 conventional chordomas), and 24 chondrosarcomas. A tissue microarray containing 0.6 mm cores of each tumor in triplicate was constructed using a manual array (MTA-1; Beecher Instruments). For visualization of staining, the ImmPRESS detection system (Vector Laboratories) with 2-diaminobenzidine substrate was used. Sensitivities and specificities were calculated for each marker. Core loss from the microarray ranged from 25 to 29% yielding 66-78 viable cases per stain. The classic marker, cytokeratin, still has the best performance characteristics. When combined with brachyury, accuracy improves slightly (sensitivity and specificity for detection of chordoma 98 and 100%, respectively). Positivity for both epithelial membrane antigen and AE1/AE3 had a sensitivity of 90% and a specificity of 100% for detecting chordoma in this study. SOX-9 is apparently common to both notochordal and cartilaginous differentiation, and is not useful in the chordoma-chondrosarcoma differential diagnosis. Glial fibrillary acid protein, carcinoembryonic antigen, CD24, and epithelial membrane antigen did not outperform other markers, and are less useful in the diagnosis of chordoma vs chondrosarcoma. Podoplanin still remains the only positive marker for chondrosarcoma, though its accuracy is less than previously reported.

Conflict of interest statement

Disclosure/Conflict of Interest

The authors have no financial disclosures pertaining to this research to report.

Figures

Figure 1

Whole slide histology of A) hyaline chondrosarcoma, B) myxoid chondrosarcoma, C) chondroid chordoma, D) conventional chordoma and E) chordoma with solid growth pattern (20x).

Figure 2

Observed immunohistochemistry for microarray cores of A) chordoma and B) chondrosarcoma of the skull base using AE1/AE3, brachyury, SOX-9, podoplanin, EMA, CD24, polyclonal CEA and GFAP.

Figure 3

The sole chordoma not positive for AE1/AE3. This tumor was brachyury and S100 protein positive (shown), in addition to SOX-9, CD24, and podoplanin. It was negative for CEA and EMA as well.

Figure 4

Chordoma, positive for AE1/AE3 and podoplanin.

Figure 5

Rhabdoid chordoma, negative for brachyury, but still positive for AE1/AE3.

Figure 6

Kaplan-Meier curve showing a trend to adverse outcome in chordoma cases staining negative for brachyury (p-value 0.099).