Patients with chordomas admitted to Samsung Medical Center between March 1996 and February 2015 were retrospectively enrolled. Data on the patients' age at initial diagnosis, gender, tumor size, type of tumor resection (complete or incomplete), and postoperative adjuvant treatment were obtained from medical records. Magnetic resonance images of all the patients were reviewed to identify the tumor location. Chordomas were divided into two groups: SBCs and sacral/spinal chordomas (SCs). The SBCs included tumors of the clivus, sphenoid, spheno-occipital, sella/parasellar regions, petrosal part of the temporal bone, and intradural chordomas. Meanwhile, SCs included tumors of the spine and sacrococcygeal regions. For SBCs, tumor extension beyond the clivus or skull base was defined as an extracranial extension.

Histologic examination was performed using hematoxylin and eosin (H&E)-stained slides. All slides were retrospectively reviewed by two pathologists. The following histologic parameters were evaluated: tumor matrix (chondroid or myxoid [non-chondroid]), growth pattern (lobulating or diffuse), necrosis, presence and proportion of BNCT-like component, nuclear pleomorphism, and intracytoplasmic hyaline globules. Chondroid matrix was defined as bluish homogeneous matrix reminiscent of hyaline cartilaginous matrix, which was distinguished from eosinophilic, pinkish secretory materials or myxoid matrix (Fig. 1A). Lobulating growth pattern was assessed when obvious intralesional fibrous septa were identified under a low-power magnification (Fig. 1B). BNCT-like component was defined as sheets of large uni- or multivacuolated adipocyte-like cells with eccentrically located small nuclei and lacked the intervening tumor matrix (Fig. 1C). Nuclear pleomorphism was evaluated according to the following criteria: absent, small bland looking nuclei with inconspicuous nucleoli; present, enlarged nuclei having prominent nucleoli and membrane irregularity.

To evaluate the proliferative activity of tumors, Ki-67 immunohistochemistry was performed for each tumor. Sections from formalin-fixed paraffin-embedded tissue were used for immunohistochemistry with antibodies against Ki-67 (clone MIB-1; 1:1,000; Abcam, Cambridge, UK). Briefly, 4-µm-thick paraffin sections were deparaffinized and rehydrated using xylene and alcohol solution. Immunohistochemistry was performed using the Ventana Discovery XT automated stainer (Ventana Medical System, Tucson, AZ, USA). Antigen retrieval was performed using cell conditioning 1 buffer (citrate buffer Ph 6.0, Ventana Medical System). Appropriate positive and negative controls for immunohistochemistry were included. The findings of immunohistochemical staining of all cases were assessed with eyeball interpretation method by two pathologists using light microscopes.

Results were expressed as mean ± standard deviation. Data were analyzed using SPSS for Windows (version 18.0; SPSS Inc., Chicago, IL, USA). Mann-Whitney U test and Kruskal-Wallis test were used to compare continuous variables, and Pearson's χ² tests and Fisher's exact tests were used to compare categorical variables. Wilcoxon signed-rank tests were used to compare matched variables. Logistic regression analysis was used to assess the risk factors for death and recurrence. Disease-free survival (DFS) was calculated from the date of the first surgery to the date of the first regional or systemic relapse. Overall survival (OS) was estimated from the date of the first surgery to the date of the last follow-up or death from any cause. Cox proportional hazard model was used to assess the risk factors for overall survival and disease-free survival, and patients who were followed-up over 6 months were included for analysis. Kaplan-Meier survival curves and log-rank statistics were employed to evaluate time to survival. P < 0.05 was considered significant.

This study was approved by the Institutional Review Board of Samsung Medical Center with a waiver of informed consent (No. 2015-02-071).

In total, 52 patients were included in the analysis. Of them, 32 (61.5%) were men, and 20 (38.5%) were women. The mean age was 54.4 years (range, 18–82 years). The course of disease duration ranged from 1 to 276 months (average, 57.4 months). Regarding tumor location, 32 patients (61.5%) and 20 patients, (38.5%) had SBCs and SCs, respectively. The clinicopathologic features of the patients with SBCs and SCs are shown in Table 1. There was no dedifferentiated chordoma in the present cohort. Most of the SBCs affected the clivus (27 cases, 84.4%), from which extracranial extension was found in 17 patients. There were 3 cases of intradural chordoma. Extracranial extension of the tumor affected the sphenoid bone, pterygoid bone, and nasopharynx. Meanwhile, the most frequently affected region in SCs was the sacrococcygeal region (16 cases), followed by the cervical (3 cases) and thoracic spines (1 case). The mean age at diagnosis of the patients with SBCs was younger (53.5 years) than those with SCs (57.7 years). Meanwhile, the mean size of the tumors was significantly larger in SCs (65.4 mm) than that in SBCs (40.5 mm). Death and tumor recurrence were more common in SBCs than those in SCs (28.1% and 37.5% vs. 15% and 25%). Forty-four patients (84.6%) underwent surgical resection. The total resection rate was lower in SBCs (46.4%) than that in SCs (68.8%). Among patients who underwent surgical resection, complete resection without residual tumor was achieved in only approximately 50%. Adjuvant therapy including radiotherapy and proton therapy after biopsy or surgical resection was given in approximately 50% and 65% of patients with SBC and SC, respectively. The 5-year survival rate was lower for patients with SBCs than that for patients with SCs (78.1% vs. 95.0%), but not statistically significant. Other parameters including age, gender, patient death, and tumor recurrence were not significantly different between the two groups.

Regarding histopathological features, the lobulating pattern was more significantly common in SCs (85.7%) than that in SBCs (34.4%). Chondroid matrix was observed more frequently in SBCs (75%) than in SCs (43.8%). Meanwhile, nuclear atypia and necrosis tended to be more frequently found in SCs, but the frequency is not significantly different. BNCT-like features were observed in 37.5% (12/32) of SBCs and 25.0% (5/20) of SCs. Hyaline globules were found in 18.8% (6/32) of SBCs and 25.0% (5/20) of SCs. For the mean Ki-67 labeling index (LI), no significant difference was found between SBCs (mean, 2.6%) and SCs (mean, 3.2%). The Ki-67 LI of tumors and BNCT-like areas showed no significant difference according to the tumor location. However, a comparison of the Ki-67 LI of tumors and BNCT-like areas in each patient showed that large cells of BNCT-like areas tended to show a lower Ki-67 LI than those of typical chordoma areas (0.83% ± 1.82% vs. 2.28% ± 3.70%; P = 0.063).

In univariate logistic analysis, residual tumor was the only risk factor for patients' death in SBCs (Table 2). History of adjuvant treatment and presence of BNCT-like component were independent negative risk factors for recurrence in SBCs (Table 3). No significant variables associated with death or recurrence was observed in SCs.

In SBCs, older age and residual tumors were risk factors for shorter OS, and older age was the only independent risk factor for shorter OS (Table 4). No significant risk factor was found in SCs. There was no significant risk factor for shorter DFS in both SBCs and SCs. Only chondroid matrix seemed to be the risk factor for shorter DFS in SCs, but it was not significant (Table 5).

In the overall analysis regardless of tumor location, older age was the only independent risk factor for shorter OS, and chondroid matrix was the only independent risk factor for shorter DFS. The OS was not significantly different between patients with SBCs and SCs. However, the patients with SBCs showed significantly shorter DFS than those with SCs (Fig. 2).

When all patients were analyzed together, risk factors for short OS was age, which was significant in univariate and multivariate analysis (Table 6). For risk factors of short DFS, chondroid matrix was an only significant factor by univariate and multivariate analysis in all patients (Table 7). Clinvus/intradural location of tumor was a significant risk factor of short DFS by univariate analysis but not significant by multivariate analysis (Table 7).