Experimental Models of Chordomas

Today, in 2016, the treatment of chordoma is still challenging despite major improvements in diagnosis, treatment, and care. There are more unknowns than what we know about the biology of chordomas; the exact mechanical link between ecchordosis physaliphora and chordoma has not been established. Similarly, there is compelling evidence for a relatively benign and a more aggressive clinical subset of chordoma, but the mechanisms underlying this difference in biological behavior are still unknown. The radiobiology of chordomas also has many unknowns. Last but not least, we have neither any medical treatment for chordoma, nor diverse experimental models to screen for possible candidates. Definitive treatment for chordomas will be, without a doubt, only possible with understanding of the disease pathology. Therefore, experimental modeling of chordomas in the laboratory is of crucial importance.

Chordomas are not endemic to the human species; spontaneous development of chordomas have been reported in animals such as dogs, cats, ferrets, mink, rats, and mice; however, in all species in which the tumor exists, chordoma is a very rare tumor. ¹^,²^,³^,⁴^,⁵^,⁶^,⁷^,⁸^,⁹^,¹⁰^,¹¹^,¹²^,¹³^,¹⁴^,¹⁵^,¹⁶^,¹⁷^,¹⁸^,¹⁹^,²⁰^,²¹^,²² There are no animals that spontaneously and predictably develop chordomas; therefore, experimental studies rely on experimentally induced tumors and cell cultures. Many experimental models have been created using chemical carcinogens for other cancers, such as skin cancer, lung cancer, or gliomas. Such chemically induced in situ models do not exist for chordomas. There is one single report of metastatic chordoma development in 1 of 100 Fischer 344 rats after oral administration of diarylanilide yellow; in this study, the tumor could not be attributed to the use of the drug and the observation has not been replicated in any other animal model. ¹

Useful and practical chordoma models consist of the early models of mechanically induced notochord tumors, human tumor–derived cell lines (and their xenograft models in small animals), and genetically engineered models. When analyzed systemically, the history of experimental chordoma models can be divided into two eras. Before the molecular biological era, models consisted of mainly animal tumors resembling chordomas that were produced by mechanical–surgical interventions. In the molecular biological era, both in vitro chordoma cultures as well as in vivo animal models were created; these models were meticulously characterized. Most recent studies also signal the emergence of small-animal models that were created by the introduction of human chordoma–specific genetic defects.

8.2 Initial Efforts in the Pre-Molecular Biological Era

In 1846, Rudolph Virchow reported of “gelatinous nestlike formations” within the spheno-occipital synchondrosis and for the first time identified the “physaliphorous” cells, characterized by a large vacuolated nucleus. In his initial description, Virchow speculated that these should be “growth and mucoid metamorphosis of the sphenoocipitalcartilage.” He therefore identified these lesions as “ecchondrosis physalifora spheno-occipitalis” with references to the origin from cartilage and the characteristic vacuolated cells. ²³ In contrast to this theory, in 1858, Heinrich Müller hypothesized that these physaliphorous cell collections were remnants of the chorda dorsalis and suggested the convergence to the term “ecchordosis physaliphora,” with reference to the chorda dorsalis. ²⁴ Müller’s theory was based on the then recent identification by Kölliker that the nucleus pulposus originated from the notochord. ²⁵ This caused controversy whether to define chordomas as “tumors” or “developmental anomalies.” This issue was settled in 1864, when Klebs gave the description of pontine compression due to a chordoma. ²⁶ To support the theory of “origin from chorda dorsalis,” Ribbert created an animal model. The author also successfully produced an experimental model by puncturing the anterior intervertebral ligament and the nucleus pulposus, creating a herniation of the nucleus pulposus in rabbits resulting in tumors that were histologically similar to human chordomas. With this strong support for the theory of notochordal origin, Ribbert christened the tumor “chordoma” and classified it as a “developmental tumor.” ²⁷ His model of iatrogenic ecchordosis was replicated by Congdon in 1952 in a similar rabbit model. ²⁸ Fischer and Steiner ²⁹ created a malignant chordoma model again in rabbits. Both these observations supported Ribbert’s hypothesis.

8.3 Studies on Human Tumors in the Molecular Biological Era

Chordomas are rare and slowly growing primary bone tumors that originate from embryonic remnants of the notochord, a mesoderm-derived structure that plays an important role in neurulation and embryonic development. Chordomas are locally infiltrating tumors. They most frequently arise in the sacrum, followed by cranial and axial skeleton, with an incidence rate of <1 per 1,000,000 people per year. ³⁰ The median age at presentation is around the sixth decade. ³⁰

The current standard regimen for chordomas is surgical resection followed by high-dose radiotherapy. Complete resection is difficult due to location and locally destructive behavior of these tumors. Chordomas are also resistant to chemotherapy. The prognosis is 5-, 10-, and 20-year survival rates of 67, 40, and 13%, respectively. ³⁰ The overall median survival is approximately 6 years. ³⁰ As chordomas are usually low grade, distant metastases may be found in up to 40 to 60% of cases many years after initial diagnosis.

There has been relatively little preclinical research focusing on chordomas. For developing novel and efficacious therapies for the treatment of chordomas, it is necessary to develop and evaluate preclinical experimental model systems. Novel in vitro cell lines and in vivo xenograft models of chordomas from primary tumors allow clinicians to understand chordoma tumorigenesis, and the development and evaluation of new therapeutics.

8.3.1 In Vitro Cell Lines

Immortalized tumor cell lines are an important component of our preclinical cancer research armamentarium. Initial efforts concentrated on explant cultures from surgical specimens. Horten and Montague successfully cultured explants from a sacral chordoma on different culture systems, including collagen-coated glass coverslips and organ cultures. ³¹ The authors reported that the main morphological features of small polygonal shape and large spherical nucleus, abundant endoplasmic reticulum, and Golgi apparatus were consistent in all culture conditions, but the degree of cytoplasmic vacuolization differed, with less vacuoles when grown on coverslips. ³¹ These vacuoles eventually resulted in the physaliphorous appearance of the tumor cells. Primary cultures of chordomas are difficult to achieve due to their slow-growing nature and their propensity to undergo crisis with continued passaging.

The first chordoma cell line, U-CH1, was established in 2001 and derived from a recurrent sacrococcygeal chordoma, which was resected after radiotherapy. ³² This first reported cell line had a long doubling time (7 days), chromosomal rearrangements (der(1)t(1;22), del(4), +del(5), +del(6), +7, del(9), del(10), +der(20)t(10;20), +21) and chromosomal instability. Since then, several chordoma cell lines ( ▶ Table 8.1) have been reported in the literature, including three lines derived from aggressive chordomas of the skull base, a metastatic lesion from a soft tissue mass from the neck, two lines derived from primary sacral chordomas (CCL-3, JHC7), two lines derived from a recurrent sacral chordoma (U-CH2, MUG-Chor1), one line derived from a recurrent chordoma of the lumbar spine (CH8), and a line derived from a scapular tumor (EACH-1). ³²^,³³^,³⁴^,³⁵^,³⁶^,³⁷^,³⁸^,³⁹ When grown in culture, most of these chordoma cell lines grow slowly with long doubling times, just as in primary cultures. Therefore, there are only a few primary chordoma cell lines that have been reported in the literature.

**Table 8.1** Chronological list of established chordoma cell lines
Cell line	Characteristics	Reference	Brachyury expression	CD24 expression
U-CH1	Recurrent sacral chordoma from a 56-year-old male patient	Scheil et al 2001 ³²	Yes	Yes
3 cell lines	Clival chordomas from three patients	Ricci-Vitani et al 2006 ³⁸	N/A	N/A
Metastatic chordoma cell line	Soft tissue mass from neck	Ostroumov et al 2007 ³⁷	N/A	N/A
U-CH2	Recurrent sacral chordoma from a 72-year-old female patient	Brüderlein 2010 ³³	Yes	Yes
EACH-1	Extra-axial scapular	DeComas et al 2010 ³⁴	N/A	N/A
JHC7	Primary sacral chordoma from a 61-year-old female patient	Hsu et al 2011 ³⁵	Yes	Yes
JHH-2009–011	Clival chordoma (low passage culture)	Siu et al 2012 ³⁹	N/A	N/A
MUG-Chor1	Primary sacral chordoma from a 57-year-old female patient	Rinner et al 2012 ⁶⁶	Yes	Yes
CH22	Sacral chordoma	Liu et al 2012 ³⁹	N/A	N/A
UM-Chor1	Primary clival chordoma from a 66-year-old male patient	Not yet published	Yes	Yes
U-CH10	Recurrent sacral chordoma from a 75-year-old male patient	Not yet published	Yes	Yes
U-CH11	Primary sacral chordoma from a 71-year-old male patient	Not yet published	Yes	Yes
DVC-4	Sacral	Karikari et al 2014 ³⁶	Yes	Yes

Among these cell lines, the initially described U-CH1 cell line is the most extensively studied and reported one. ³²^,³³^,³⁶^,⁴⁰^,⁴¹^,⁴²^,⁴³^,⁴⁴^,⁴⁵^,⁴⁶^,⁴⁷^,⁴⁸^,⁴⁹^,⁵⁰^,⁵¹^,⁵²^,⁵³ Using the U-CH1 chordoma cell line, Aydemir et al ⁴⁰ have shown that chordoma cells can be induced to commit to osteogenic line, when treated with an osteogenic differentiation medium. The authors have also identified a CD133- and CD15-positive subpopulation in U-CH1 cells that exhibited stem cell characteristics (such as colony formation in soft agar and self-renewal capacity). ⁴⁰ The interaction of the tumor cell with its microenvironment was also studied using the same U-CH1 cell line. Patel et al ⁴⁶ showed that hypoxia as well as the connective tissue growth factor (CCN2) increased the expression of notochord-associated markers (brachyury, SOX5, SOX6, CD24, and FOXA1) and promoted tumor sphere formation, which is characteristic of progenitor cells. Similarly, a comparison of gene expression of U-CH1 and U-CH2 with that of vertebral disc tissue yielded 65 significantly differentially expressed genes, among which were T Gene, CD24, ECRG4, RARRES2, IGFBP2, RAP1, HAI2, RAB38, SPP1, GalNAc-T3, and VAMP8. The U-CH1 cell line was successfully used to screen for drug effects against chordoma. ⁴¹^,⁵⁴ Schwab et al ⁵⁴