Chordoma and chondrosarcoma of the skull base are frequently amalgamated because of similar anatomic location, clinical presentation, and radiologic findings. The chondroid chordoma variant has been reported to carry a better prognosis. The objective of the current study was to investigate the distinctions between these 3 entities.
The data concerning 109 patients with chordoma, chondroid chordoma, and chondrosarcoma who were treated by a single surgeon with maximum surgical resection and frequently by adjunct proton beam radiotherapy between 1990 and 2006 were analyzed retrospectively. Pathologic distinction was established by cytokeratin and epithelial membrane antigen staining. Clinical, radiologic, pathologic, and cytogenetic studies were analyzed in relation to disease recurrence and death.
The average follow-up was 48 ± 37.5 months (range, 1–191 months). There were no reliable distinguishing clinical or radiologic features noted between the groups. Chondrosarcoma patients had a significantly better outcome compared with chordoma patients with regard to survival and recurrence-free survival (P = .028 and P < .001, respectively), whereas patients with chondroid chordoma had a poor outcome similar to chordoma patients with regard to survival and recurrence-free survival (P = .337 and P = .906, respectively).
Chordoma and chondrosarcoma of the skull base are rare tumors. The combined incidence is reported to be 0.03 per 100,000 persons in the U.S.1 Occupying the same anatomic location, the clinical presentation and radiologic characteristics of chordoma and chondrosarcoma are quite similar, often resulting in their amalgamation in the literature. They are, however, quite different entities.
Chordoma and chondrosarcoma have specific histologic2 and immunohistologic features3 that usually allow for their accurate distinction. Skull base chordomas arise from remnants of the primitive notochord at the spheno-occipital synchondrosis, whereas chondrosarcomas originate from primitive mesenchymal cells or from the embryonic rest of the cartilaginous matrix of the cranium. Although they each have typical pathologic features, they can be differentiated by immunohistochemical staining. Microscopically, chordomas are comprised of uniform cells containing small, oval, eccentric nuclei with dense chromatin and numerous cytoplasmic vacuoles (physaliferous cells). Chondroid chordoma, a variant of chordoma first identified by Heffelfinger et al.,2 also has a cartilaginous component. Chondrosarcoma is comprised of cartilage with pleomorphic chondrocytes. Chordoma and chondroid chordoma are immunopositive for epithelial markers including cytokeratin and epithelial membrane antigen (EMA), whereas chondrosarcoma is negative for both.2
It is apparent that the natural history and clinical outcome are quite different between chordomas and chondrosarcomas.2, 4 However, there is much uncertainty concerning these tumors due to their rarity and thus the small size of existing case series. The current study analyzes the presentation, response to treatment, and outcome in a large series of 109 patients with chordoma, chondroid chordoma, and chondrosarcoma who were uniformly treated and followed by a single surgeon (O.A.).
MATERIALS AND METHODS
Among 168 patients treated for chordoma, chondroid chordoma, and chondrosarcoma by the senior author (O.A.), a series of 109 patients treated between 1990 and 2006 were analyzed retrospectively for outcome, disease recurrence, and complications. The patients were treated for de novo, residual, or recurrent disease (Table 1). Diagnosis was established using histologic criteria and confirmed with cytokeratin and EMA immunohistochemical staining. Before surgery, all patients were evaluated radiographically, including magnetic resonance imaging (MRI), computed tomography, and magnetic resonance angiography and/or magnetic resonance venography. Tumors were classified as large if they measured ≥2.5 cm in any dimension and small if they measured <2.5 cm in all dimensions.
Table 1. Clinical Characteristics of 109 Patients in 3 Groups: Chordoma, Chondroid Chordoma, and Chondrosarcoma
Chordoma (n = 67)
Chondroid chordoma (n = 22)
Chondrosarcoma (n = 20)
38.3 ± 19.3
35.9 ± 19.1
46.4 ± 14.2
Previous surgery and radiotherapy
Radical surgical resection, defined as resection of all visible tumor under the surgical microscope, was achieved in 90 patients (83%). However, immediate postoperative MRI detected small masses that were suspicious for residual tumor in 45 of these patients. In some patients, this finding required staged surgeries utilizing multiple approaches when the tumors extended into areas that could not be removed using a single approach. In 19 patients (17%) only partial resection, but resection adequate to proceed with proton beam radiotherapy, was achieved due to adhesion of the tumor to vital structures. After surgical resection, radiotherapy was administered as indicated to chordoma and chondroid chordoma patients. Fifty chordoma patients (75%) and 16 chondroid chordoma patients (73%) received postoperative radiotherapy. Those chordoma and chondroid chordoma patients not receiving postoperative radiotherapy did not receive it because of prior maximum radiation dosage or noncompliance. The majority of the patients receiving radiotherapy were treated with proton beam radiotherapy at outside facilities. Patients with chondrosarcoma often underwent postoperative radiotherapy (50%), but after a period of successful surgical treatments, radiotherapy was reserved due to the risks associated with it. Cytogenetic analysis was performed in 47 chordoma patients (70%), 13 chondroid chordoma patients (59%), and 9 chondrosarcoma patients (45%) using standard in situ culture techniques and G-band karyotype or spectral karyotype (SKY technique).
For statistical analysis, the Fisher exact test and Kaplan-Meier survival curves with the log-rank test P value were calculated with Medcalc software (version 7; Medcalc Software, Mariakerke, Belgium) to evaluate and compare clinical outcome and disease recurrence between groups. In addition, we compared outcome between de novo cases and recurrent cases, which was defined as a documented disease recurrence after prior treatment.
There were 67 chordoma patients, 22 chondroid chordoma patients, and 20 chondrosarcoma patients treated for de novo, residual, or recurrent disease (Table 1). The average follow-up was 48 ± 37.5 months (range, 1–191 months). Patients age <17 years accounted for 16% of the chordoma group, 23% of the chondroid chordoma group, and 0% of the chondrosarcoma group. The mean age for chordoma patients was 38.3 ± 19.3 years (range, 1–87 years), and was similar in male and female patients and comparable to that of chondroid chordoma patients (mean age of 35.9 ± 19.1 years [range, 4–68 years]). The average age of the chondrosarcoma patients was 46.4 ± 14.2 years (range, 19–77 years). Approximately 84% of patients in the chordoma group were white and 3% were African American. In the group of patients with chondroid chordoma, 77% were white and 0% were African American. Of those patients with chondrosarcoma, 95% were white and 5% were African American. There was a small female predominance in the chordoma group with a male: female ratio of 0.67:1, whereas a more equal distribution between females and males was noted in the chondroid chordoma group (male:female ratio of 1.09:1) and the chondrosarcoma group (male:female ratio of 0.8:1).
MRI Appearance and Location
On MRI examination of de novo cases, all 3 groups demonstrated isointensity or hypointensity on T1, and hyperintensity on T2. Enhancement with gadolinium contrast was present in 73% of chordomas, 100% of chondroid chordomas, and 88% of chondrosarcomas. Details of the MRI findings are presented in Table 2. Tumors were large at the time of presentation in 88% of chordoma patients, 95.0% of chondroid chordoma patients, and 85% of chondrosarcoma patients.
Table 2. MRI Characteristics in De Novo Cases of Chordoma, Chondroid Chordoma, and Chondrosarcoma
MRI indicates magnetic resonance imaging.
Hypointense or isointense
The clivus was involved in 90% of chordoma patients, 91% of chondroid chordoma patients, and 85% of chondrosarcoma patients (Table 3). The clivus was exclusively involved in 18%, 5%, and 0%, respectively, of chordoma, chondroid chordoma, and chondrosarcoma patients. The cavernous sinus and clivus were involved in 37% of chordoma patients, 73% of chondroid chordoma patients, and 65% of chondrosarcoma patients. The craniospinal junction with the clivus was involved in 22% of chordoma patients, 23% of chondroid chordoma patients, and 5% of chondrosarcoma patients. The petrous bone was involved with the clivus in 12% of chordoma patients, 32% of chondroid chordoma patients, and 80% of chondrosarcoma patients (Figs. 1 and 2).
Table 3. Location and Extension of Tumors at the Time of Treatment
Type of tumor
Chordoma (n = 67)
Chondroid chordoma (n = 22)
Chondrosarcoma (n = 20)
Involving cavernous sinus
Involving craniospinal junction
Involving petrous bone
Involving other structures
Not involving clivus
Cervical spine and craniospinal junction
Involving cavernous sinus
Involving petrous bone
Histology and Cytogenetics
Of 67 chordoma patients, only 3 cases had a histological appearance of atypical chordoma (1 of which occurred in a child). Of the 22 chondroid chordoma patients, 1 child had an atypical chondroid chordoma. Grading of chondrosarcomas was performed based on a 3-tier grading system according to previously established criteria5 tumors were classified as grade I, 9 chondrosarcomas were classified as grade I to II, and only 1 chondrosarcoma in the current study was classified as grade III (Fig. 3).
Chromosomal analysis demonstrated that 43% of chordoma cases, 38% of chondroid chordoma cases, and none of the chondrosarcomas cases had an abnormal karyotype (Table 4). It is interesting to note that there was a significant increase (P = .013) of cytogenetic abnormalities between de novo chordoma (29%) and recurrent chordoma (69%) cases. This finding was similar between de novo (0%) and recurrent chondroid chordoma (83%) cases and implies tumor progression with an accumulation of cytogenetic abnormalities.
Table 4. Patient Outcome (With an Average Follow-up of 48 ± 37.5 Months [Range, 1–191 Months])
Chordoma (n = 67)
Chondroid chordoma (n = 22)
Chondrosarcoma (n = 20)
Residual tumor after surgery
Treated with proton beam radiotherapy
Disease recurrence without residual tumor
Disease recurrence with residual tumor
Death from disease progression
Time to disease recurrence, mo (de novo cases)
24.2 ± 27.4 (range, 2–104)
9 (1 case)
Time to disease recurrence, mo (recurrent cases)
14.0 ± 10.7 (range, 2–29)
13.0 ± 14.9 (range, 3–48)
Abnormal karyotype (de novo tumors)
9 of 31 (29%)
0 of 7 (0%)
0 of 9 (0%)
Abnormal karyotype (recurrent tumors)
11 of 16 (69%)
5 of 6 (83%)
0 of 0
There were no deaths or instances of disease recurrence in any chondrosarcoma patient after treatment. The 10-year survival rate for chordoma was 59%, and was 82% for chondroid chordoma (Fig. 4). There was a statistically significant difference noted between chordoma and chondrosarcoma (P = .028), but no significant difference in survival was observed between chordoma and chondroid chordoma (P = .337). The median recurrence-free survival for chordoma was 88 months; the recurrence-free survival for chondroid chordoma did not reach its median (Fig. 5). However, there was no statistically significant difference with regard to disease recurrence observed between chordoma and chondroid chordoma (P = .906), whereas there was a statistically significant difference noted between chordoma and chondrosarcoma (P < .001).
The 10-year survival rate for de novo cases (chordoma or chondroid chordoma) was 74% and that for the recurrent cases was 48% (Fig. 6). There was a statistically significant difference in survival noted between recurrent and de novo cases (P < .001). The median recurrence-free survival time for patients with disease recurrence was 12 months; the recurrence-free survival time for de novo cases did not reach the median (Fig. 7). There was a statistically significant difference in disease recurrence noted between de novo and recurrent cases (P < .001). The average time to disease recurrence for de novo and recurrent tumors was 24 months (range, 2–104 months) and 14 months (range, 2–29 months), respectively, for chordoma patients, and was 9 months (1 case) and 13 months (range, 3–48 months) for chondroid chordomas cases (Table 4). In de novo cases, when no residual tumor was observed on postoperative MRI, the 10-year recurrence free survival rate was 74%. When residual tumor was observed, the 10-year recurrence-free survival rate was 40%. Recurrence-free survival was significantly improved when no residual tumor was observed (P = .044) (Fig. 8).
There was not a significantly higher number of deaths from disease progression noted in females compared with males in the chordoma group (P = .134), nor were there more cases of disease recurrence (Table 5). There was no statistically significant difference noted between the age groups of <17 years, 18 to 40 years, and >40 years with regard to disease recurrence or incidence of death from disease progression. Metastasis or seeding of tumors was demonstrated in 10% of chordoma patients, 14 % of chondroid chordoma patients, and 0% of chondrosarcoma patients (Fig. 9).
Table 5. Clinical Outcome Comparing Male and Female Patients With Chordoma, Chondroid Chordoma, and Chondrosarcoma
Chordoma (n = 67)
Chondroid Chordoma (n = 22)
Chondrosarcoma (n = 20)
M indicates male; F, female.
Death from disease progression
Intracranial chordomas and chondrosarcomas have erroneously been grouped together due to their similar anatomic location, radiologic features, and surgical treatment.4, 6 Chondroid chordoma has controversially been described as a variant of chordoma. The literature shows some controversy and confusion with regard to the similarities and differences between these tumors. From this controversy, 2 important questions have arisen: 1) are these distinct clinicopathologic entities? and 2) is there a prognostic difference between them?
Chordomas are rare and considered to be slow growing yet locally aggressive neoplasms of the axial skeleton, with 32% occurring in the clivus.3, 4, 7 The origin of chordomas is believed to be notochordal rests. Chordomas appear similar to fetal notochord on both light and electron microscopy2, 8, 9 and are immunohistochemically and ultrastructurally similar.3 Thus, chordoma of the skull base originates at the spheno-occipital junction. In soft tissues, they appear encapsulated; however, this encapsulated appearance is not present in bone. Microscopic examination shows the tumors to be pseudoencapsulated by fibrous strands forming thick hylanized septa or thin septa creating lobules. The lobules appear as a sheet of vacuolated physaliferous cells or a pool of mucin. The sheets of cells contain varying amounts of intracytoplasmic mucin, ranging from hardly visible to quantities causing the cells to rupture.2 Immunohistochemistry is an important tool in the diagnosis of chordoma. Chordomas are positive for cytokeratin, EMA, S-100, and vimentin (Fig. 3).2
Chondrosarcomas are rare neoplasms of cartilage that occur most frequently in the long bones and pelvis, with <10% reported to occur in the head and neck.10 They account for only 0.16% of all intracranial tumors and the majority of these occur at the base of the skull.11, 12 Chondrosarcomas are believed to arise from primitive mesenchymal cells or from embryonic rest of the cartilaginous matrix in the cranium.2, 13, 14 They can be classified into 3 subgroups: classic, mesenchymal, and myxoid, with the classic form being the most common type. The appearance of the classic form is comprised of large cells with single or multiple nuclei in a variable amount of chondroid matrix (Fig. 3).
Classic chondrosarcomas are divided into grades I, II, or III based on mitotic rates, cellularity, nuclear size of the cells, and chondroid matrix. Grades II and III tumors have less chondroid matrix and more mitosis.4, 5, 14 The mesenchymal type contains areas of relatively well-differentiated cartilage and undifferentiated round or spindle-shaped cells. The myxoid type has strings of rounded cells in a myxoid matrix.12 In chondrosarcomas, immunohistochemical stains are negative for cytokeratin and EMA, and positive for S-100 and vimentin.3 Intracranial chondrosarcomas are typically of the classic type and are low-grade grade I (51%) or grade II (11%) tumors followed by the mesenchymal subtype (30%).12, 15 All but 1 chondrosarcoma in the current series was grade I or grade II classic type.
Chondroid Chordoma Is Just a Chordoma
While studying the histology of chordomas and cartilaginous tumors of the skull base, Heffelfinger et al.2 noted a set of chordomas that were difficult to identify as such because of a significant chondroid element that had features of chondrosarcoma or chondroma. The tissue appeared similar to hyaline cartilage with neoplastic cells in lacunae. The tumors ranged from mostly chordoma with small, scattered foci of cartilaginous differentiation to mostly cartilaginous tumors with small areas of chordoma. These tumors can be potentially difficult to classify. In fact, the study by Heffelfinger et al. found that several such tumors had been previously misdiagnosed as chondrosarcomas. Heffelfinger et al. designated these as “chondroid chordomas” and reported that the chondroid type indicated a better prognosis.2
Since this initial designation, chondroid chordomas have been a point of controversy. Bottles and Beckstead concluded, based on enzyme immunohistochemical study, that chondroid chordoma is most likely a cartilage neoplasm and not a chordoma variant.16 Brooks et al. and Walker et al. determined based on immunohistochemical studies that chondroid chordoma either does not exist or is an extremely rare variant of low-grade chondrosarcoma and that the term “chondroid chordoma” should be abandoned.17, 18 In an attempt to resolve the issue of chondroid chordomas as a separate entity, Rosenberg et al.3 studied a series of tumors from their institution and found that chondroid chordoma is indeed a variant of chordoma and is distinct from chondrosarcoma. Their immunohistochemical study demonstrated that chondroid chordoma and chordoma reacts with cytokeratin and EMA, whereas chondrosarcoma is negative for cytokeratin and EMA, thus providing a definitive tool with which to distinguish chondrosarcoma from chordoma.3, 15 Rosenberg et al.3 suggested that the previous studies by Brooks et al.17 and Walker et al.18 used different criteria in determining chondroid chordoma and that in fact they should have considered these chondrosarcomas. Rosenberg et al. concluded that the term “chondroid chordoma” should be kept because of its reported better prognosis than chordoma and that further clinical and pathologic investigations are necessary to better determine the prognostic significance of chondroid chordomas.3 The results of the current study confirm the presence of the pathologic variant of chondroid chordoma and the value of immunohistochemical staining in distinguishing it from chondrosarcoma. The biologic behavior and outcome of chondroid chordoma is similar to that of typical chordoma and quite different from that of chondrosarcoma.
Location and Appearance on MRI
Chordoma and chondrosarcoma have a similar radiologic appearance. Typically, they are hypointense or isointense on T1 and hyperintense on T2, and frequently enhanced with gadolinium.12, 19, 20 This was the typical appearance noted on MRI in the current series. Despite the finding that chondrosarcomas have a similar appearance on MRI, they can periodically be distinguished from chordomas due to the occupation of a more lateral location along the petroclival fissure.4, 19
Skull base chordomas are typically located in the midline19 and appear to originate in the bone, infiltrating the path of least resistance and eventually producing a soft tissue mass. This is consistent with findings in the current series, in which the majority of cases involved the clivus or spine. The results of the current study demonstrated that 80% of chondrosarcomas involved the clivus and petrous bone, 10% involved the petrous bone and not the clivus, and 0% involved the clivus and not the petrous bone. This indicates that chondrosarcomas begin laterally and grow into the midline, whereas chordomas are tumors that originate in the midline and grow laterally.
In a population-based survey of chordoma patients, it was found that chordomas were distinctly unusual among the African-American population and those cases that did exist were frequently diagnosed before the age of 40 years.1, 7 In the current series, there were only 2 African-American patients among the patients with chordomas or chondroid chordomas. Both patients were aged <40 years.
In a series of 62 cases of intracranial chordomas, males were reported to fare significantly better than females with regard to disease-free survival and overall survival.21 The current series reproduced only a statistically significant difference in the number of deaths from progression of disease and demonstrated no significant difference with regard to disease recurrence between males and females. Forsyth et al. reported an improved survival for patients age <40 years22; our analysis showed there to be no significant difference with regard to disease recurrence or death from disease progression between the 2 populations.
Intracranial chondrosarcomas have a slow progressive growth that can be fatal due to eventual compression of the brainstem. Therefore, local control is the primary goal of therapy. Surgery with adjuvant radiotherapy has been the recent treatment of choice, with good success and a 5-year progression-free survival rate of 95%.15
The results of the current series demonstrated a clear distinction in the outcomes of patients with chordomas and chondrosarcomas. Chondrosarcoma patients clearly have a better prognosis. There were no deaths from disease or disease recurrence after treatment in chondrosarcoma patients. Chondrosarcomas of the skull base appear to be benign tumors that can be controlled effectively with surgical resection.
The benign nature of chondrosarcomas in the current series brings into question the use of radiation in treatment of these tumors. Radiation studies have shown great success in the management of chondrosarcomas of the skull base with adjunct proton beam radiotherapy after surgery.23–26 In the current series, we had 12 patients who were treated with radiotherapy (10 by proton beam) and 8 patients who were not treated with radiotherapy. Neither group developed disease recurrence or metastases or died. Radiotherapy has significant risks such as brain necrosis, blindness, demyelination, and radiation-induced tumor or malignant transformation.23, 27–29 These risks may be avoided in patients with grade I or grade II chondrosarcomas when complete surgical resection is achieved.
The chondrosarcoma cases in the current series were the classic type, grade I or II. The tumors appeared very benign with no metastases, recurrences, or chromosomal aberrations. This is in contrast to chondrosarcomas presenting elsewhere, which can be aggressive tumors with frequent recurrence, aberrations, and metastases. The dedifferentiated and mesenchymal types of chondrosarcoma can be much more aggressive when located in the skull base. It is important to distinguish between these subtypes of tumors and chromosomal analysis could be of value in indicating the aggressiveness of the chondrosarcoma.30–34
The clinical significance of the distinction of chondroid chordomas is in its proposed prognostic value. Since the study by Heffelfinger et al., this has been examined in several series. Eriksson et al. noted increased survival in a series of 51 chordoma cases,35 as did Rich et al. in 48 cases.36 In a study of 26 patients with cranial chordomas, Raffel et al. found the chondroid group to have a longer disease-free interval.37 In a series of 51 patients with intracranial chordoma, Forsyth et al. concluded that patients with chondroid chordomas do not have a significantly longer survival than those with typical chordomas.22 O'Connell et al. presented a series of 62 cases of chordoma of the skull base and found the presence of chondroid foci in chordomas did not have any clinical significance. The marked discrepancies in the literature could be due to the inclusion of chondrosarcomas into the chondroid chordoma group.21 In the current series and others,4, 6, 38 chondrosarcomas carried a much better outcome than chordomas, giving support to the idea that previous studies, including that of Heffelfinger et al.2 in which immunohistochemistry was not used, may have included chondrosarcomas with the chondroid chordoma groups, which would falsely improve outcome in the chondroid chordoma group. This makes immunohistochemistry an important tool in making the differentiation between chondroid chordoma and chondrosarcoma.
In the current series we found there to be no significant difference in disease recurrence and survival between the chondroid chordoma and chordoma groups. This study and the studies of Forsyth et al.22 and O'Connell et al.21 are to our knowledge the largest series published to date regarding intracranial chordomas and chondrosarcomas since Heffelfinger et al.2 made the distinction. The distinction of chondroid chordoma does not appear to have any clinical significance or prognostic value. The histologic distinction is clear and further study into the genesis of such a histologic variant would be instructive. However, these patients should receive the same prognosis and treatment regimen as chordoma patients.
A significantly improved outcome was observed in patients who presented with de novo tumors and in those patients in whom no residual tumor was noted on postoperative MRI. This indicates that the extent of resection performed at the time of initial presentation has a significant impact on patient outcome and radical resection should be attempted as part of the initial treatment.14, 38–40
Chordoma and chondrosarcoma have a common presentation and similar anatomic locations, and can be difficult to distinguish before histopathologic and immunohistochemical examination; however, these 2 tumors have a different origin and very different prognosis. Skull base chondrosarcomas carry a markedly more favorable outcome than chordoma or chondroid chordoma, with minimal recurrences and excellent long-term control. Chondrosarcomas have traditionally been treated with adjunct radiotherapy; however, radiotherapy may not have an adequate benefit-to-risk ratio in well-resected, low-grade chondrosarcomas.
Chondroid chordomas, despite having some pathologic characteristics that are similar to chondrosarcoma, behave clinically like chordomas. Chordomas and chondroid chordomas demonstrate a high rate of disease recurrence. In addition, they both demonstrate an aggressive clinical course and have uniformly poor outcome after disease recurrence. The optimal treatment for chordomas and chondroid chordomas involves radical surgical resection followed by high-dose radiotherapy. Chordomas and chondroid chordomas are similar with regard to presentation, anatomic location, immunohistochemistry, and prognosis. Despite this, the specification of chondroid chordoma should be kept for further investigation into the tumor's biology.