Chordomas and chondrosarcomas of the skull base pose a detrimental effect on the quality of life of patients. With a worse 5-year survival rate, particular effort has been made to improve the management of chordomas. This includes the search for molecules that could serve as prognostic marker and/or therapeutic target.[10, 11] In this regard, the assumption of chordomas being of notochordal origin may be crucial to understand the pathophysiology. The role of conversed genes that show similar expression patterns during development could therefore be a key but have not been studied as a large panel thus far.

The present study was designed to evaluate the presence of developmental transcription factors in chordomas and chondrosarcoma in order to differentiate the two entities and to discover potential targets for therapy. In general, most of the developmental factors tested in this study were exclusively expressed in chordoma, although eleven factors were expressed both in chordoma and chondrosarcoma. Confirming previous findings, the classical mesodermal and notochordal transcription factor brachyury was highly and exclusively expressed in chordomas.[3] In vitro studies on chordoma brachyury expression and its clinical consequences still require further determination. However, induced expression in human embryonic stem cells has shown to introduce a mesodermal phenotype.[12] The latter indicates a role for brachyury in cell differentiation, rather than promoting cell proliferation. However, both phenomena have been described in chordoma.[13] In the current study, the lack of expression of other canonical mesodermal factors TBX6 and GSC, suggests that brachyury expression in chordomas might indicate a more notochordal phenotype. This suggestion is in line with the expression of notochordal marker SHH and the novel finding of HoxA5, Chordin, Gli1 and Gli3 expression in chordomas, which are expressed in chordoma only and are also involved in notochordal development.

HoxA5 is a family member of the regulatory Hox-genes, which are known to be involved in a plethora of developmental processes.[14] In cancer, expression of HoxA5 inhibits stemness of cancer cells, resulting in differentiation and subsequent decreased proliferative potential by inhibition of WNT signaling.[15] Induced upregulation of this gene is shown to alter the cell cytoskeleton, a crucial phenomenon witnessed in the pathology of chordoma.[16, 17] In addition, expression of HoxA5 promotes the expression of CD24 generating a more epithelial-like phenotype.[18] CD24 has previously been shown to be upregulated in chordoma compared to soft tissue sarcoma and chondrosarcoma [19]. Even if our cohort only revealed expression of HoxA5 in a third of the cases, these fully overlapped with CD24 expressing samples. So, as chordomas are known epithelial tumors, show cytoskeletal remodeling and as we have also confirmed the abundant expression of CD24 in 75% of the cohort, the expression of HoxA5 poses a promising target for additional examination in these tumors. In addition, the differential expression, in our chordoma cohort, might also explain the different clinical behavior of patients harboring this disease.

In contrast to the expression of HoxA5, the expression of chordin is related to a less differentiated state in cancer cells. Overexpression is known to potently antagonize BMP-2 signaling and if downregulated is shown to be responsible for osteogenic differentiation.[20] Therefore it is not surprising that all chondrosarcoma samples failed to show any expression of this gene. Evidence of chordin’s involvement in tumor biology is missing. As chordin was expressed in 42% of the cases, further investigation of its functional involvement in chordomas is warranted to facilitate our growing knowledge of the pathophysiology.

ACAN, a proteoglycan, is a key component of the extracellular matrix of cells. During the end of adolescence and the beginning of young adulthood, around the same period notochordal remnants are expected to fully regress; a peak increase of ACAN accumulates around the nucleus pulposus cells in the human intervertebral disk.[11, 21] Hereafter, due to age-related alterations, potentially by changes in the content, the glycosaminoglycan structure, and the ability to bind hyaluronic acid, tissue functions are affected. In cancer, high expression of ACAN is associated with the process of cartilage destruction, a higher malignant state and disease progression.[22,23,24] In chordoma, Gottschalk et al.[25] illustrated the presence of abundant ACAN in chordoma and chondrosarcoma immunohistochemistry. However, our results demonstrate for the first time a clear difference in expression at the mRNA level, where chordomas do prominently express ACAN, and none of the three chondrosarcomas show ACAN expression. Further studies are required to unravel the functional impact of ACAN in chordomas.

SOX9 is a gene well recognized for its role in sex-determination.[26] However, it is also involved in notochordal maintenance and chondrogenesis.[27, 28] Both chordomas and chondrosarcomas have previously been described to express SOX9 and higher expression of SOX9 has been linked to a more aggressive phenotype in chordoma.[4, 29] Interestingly, in our samples all four chordomas tested positive, whereas none of three evaluated chondrosarcomas were SOX9 positive. Although these studies [4, 30] investigated Sox9 protein expression in fixed tissues and we evaluated RNA expression in fresh frozen samples, the difference is notable and additional investigations at both levels in a larger cohort may be useful in deciphering these differences.

Finally, the expression of stem cell factors KLF4, SOX2, OCT4 and Nanog was analyzed in this study to indicate degree of differentiation between both tumors. KLF4 was exclusively expressed in chordomas in three out of four samples. However, the expression was low. The other factors, showed increased expression in chondrosarcoma compared to chordoma, indicating that these neoplasms potentially vary in their degree of differentiation with the data suggesting more dedifferentiation in chondrosarcomas compared to chordomas.

In this study expression data is presented at the transcript level. The correlation between mRNA expression levels and protein levels is positive, albeit non-linear. Other factors, e.g. microRNA (miRNA) mediated mRNA degradation, miRNA sponging by long non-coding RNAs, alternative splicing, posttranslational modifications and speed of protein degradation are all of influence. Optimally, a broad set of molecular omics-based analyses would be performed to substantiate these findings. Insufficient material was present for additional immunohistochemical assays to concurrently evaluate protein expression, which is a limitation of this study. Ideally our findings should be corroborated in a larger cohort. A prospective longitudinal cohort study could possibly link expression of the discussed factors to patient prognosis. Furthermore, the data obtained provides useful substrates for the development of functional knock-down assays to further elucidate the biochemical signaling networks down-stream of these developmental factors.