Age

Age has been identified inconsistently as a prognostic factor. In a large series abstracted from the Surveillance, Epidemiology, and End Result (SEER) database from 1973 to 2011, Lau and colleagues compared outcomes of 86 pediatric patients (≤19 years old) and 1272 adult patients (≥20 years). Pediatric patients had a higher incidence of cancer-specific mortality (37.2 vs. 28.6%, P < .005), yet a longer overall survival (17.2 vs. 12.6 years) than did adults. The incidence of distant disease (14.8 vs. 9.2%, P < .05) was higher, and surgical resection more significantly improved survival in the pediatric subset than in the adult (22.5 vs. 14.3 years, P < .001).

Despite longer survival in this and other series, a subset of pediatric patients, typically with atypical or dedifferentiated chordoma, has significantly worse survival. Coffin et al. associated the poorer survival in pediatric patients to higher mitotic activity, hypercellularity, and pleomorphism in pediatric tumors. Similarly, Borba and colleagues described poorer prognosis in children under 5 years of age, which also correlated with greater histologic atypia.

Among adults, rates of survival outcomes for patients less than 40-year old were better than patients ≥40-year old. The rates of 5- and 10-year OS were 75% and 63% in those <40-year old, and only 30% and 11% in ≥40-year old patients, respectively. Another evaluation from the SEER database confirmed worsening prognosis with advancing age. On multivariate analysis of 416 chordoma cases, age ≥50 years and tumor size ≥4 cm were significantly associated with poorer survival. Decade of age at diagnosis and radical resection were not significant factors in OS. The rates of OS at 5- and 10-year were 65% and 32%, respectively. The best outcomes were for patients under 50-year old with tumors <4 cm; they had rates of 5- and 10-year OS of 91% and 57%, respectively. The worst outcomes were seen in patients ≥50 years with tumors ≥4 cm; they had rates of 5-and 10-year OS of 35% and 9%, respectively.

Site of Origin

The site of chordoma origin can influence various clinical factors such as the likelihood of a good surgical outcome, the interval between radiographic recognition and appearance of symptoms and tumor size at presentation. Many case series report a difference in survival based on location. Because of the rarity of the disease and the small numbers of patients, statistical analysis is not robust, and thus data are often just descriptive. One series of 219 patients treated at two institutions reported that patients with primary tumors of the cervical and thoracic spine had shorter median durations of survival (74.7 and 76.8 months, respectively) than did patients with tumors of the lumbar spine and sacrum (126.7 and 159.3 months, respectively). In this series, only 1.8% of patients had a clivus/skull base chordoma. Given the variability of natural history and surgical options among chordomas with different locations, many studies investigating prognosis have focused on chordomas originating at a specific sites.

A comprehensive meta analysis examined 23 observational studies from 1999 to 2010, totaling 807 chordomas originating in the cranial base, with an average follow-up of 53.6 months. The rates of 5- and 10-year OS were 70% and 63%, respectively. There was a significant difference in survival based upon the extent of resection. The rate of 5-year OS was 95% after complete resection and 71% without it. Adjuvant radiation did not significantly alter duration of survival. Another retrospective study of 594 cranial chordomas extracted from the SEER database found improvement in survival over three consecutive decades. Rates of 5-year OS for the decades 1975–84, 1985–94, and 1995–2004 were 48.5%, 73.0%, and 80.7%, respectively. This suggests that treatment for chordoma, particularly surgery, has improved over time.

Ma and colleagues recently proposed and validated a progression scoring system that incorporates age, treatment history, performance status, and features on magnetic resonance imaging for skull base chordomas. Their scoring system segregated their validation set of 170 patients into three prognostic groups of high, intermediate, and low-risk disease with median durations of progression-free survivals of 45 (95% CI 33.69–56.31), 31 (95% CI 23.57–38.43), and 7 (95% CI 4.87–9.13) months, respectively. The authors suggest that their scoring system may inform surveillance strategies and direct high-risk patients to new therapies.

Zou and colleagues, in a search of over a thousand citations and 65 manuscripts for prognostic factors for spinal chordoma, found that data heterogeneity hampered its aggregation. Nevertheless, from eight studies, they found that location in the upper cervical spine, intralesional surgery, extent of tumor invasion, and revision surgery were associated with inferior survival. Another retrospective study of 166 chordomas of the mobile spine treated at multiple international centers found that the risk of local recurrence, but not duration of overall survival, depended on whether an Enneking appropriate (EA, tumor-free margins) or an Enneking inappropriate (EI, positive margins) was achieved: EA versus EI (HR 7.02; 95% CI 2.96–16.6; P < .001) and that the probability of achieving an Enneking appropriate resection depended on tumor grade, location, and extent.

A recent retrospective review of 115 patients with completely resected sacral chordoma between 2003 and 2012, noted a 5-year OS rate of 1%. Patients with tumor-free margins of resection fared significantly better than those without tumor-free margins, with rates of 5-year OS of 86% versus 67%, respectively. Multivariate analysis of long-term outcomes in 99 patients with sacral chordoma treated in Italy over a span of 30 years and followed for a median of 8.7 years found that only tumor size and surgical margin significantly affected durations of disease-free and overall survival. Radiation therapy administration did not significantly affect the incidence of local relapse, but most patients received less than 60 Gy. A trend toward worse outcome was observed for tumors arising from S1–S2 than for those arising from S3 and below.

Tumor Molecular Features

Molecular features of chordomas linked to duration of patient survival may be useful biomarkers of prognostic and therapeutic significance. Some of these involve cell-cycle dysregulation and invasive growth.

Loss of chromosome 1p36, 9p loss of heterozygosity, and an elevated Ki67 proliferative index correlate with aggressive behavior and shorter OS in skull base chordomas. Among 74 chordoma patients, poly-ADP-ribose polymerase 1 (PARP1) over expression by immunohistochemistry significantly correlated with shorter continuous disease-free survival time. In another study limited to sacral chordoma, high expression of pleckstrin homology domain leucine-rich repeat protein phosphatase (PHLPP), a negative regulator of the phosphatidylinositide 3-kinase/protein kinase B (PI3K/AKT) pathway, significantly correlated with longer progression-free survival but not overall survival.

Human telomerase reverse transcriptase (hTERT), part of the telomerase complex, may be critical for unrestricted cell growth. Elevated levels of hTERT correlate with poor prognosis in many malignancies, and with a high Ki-67 proliferative index, tumor muscle tissue invasion, tumor infiltrating lymphocytes, and shorter regional relapse–free survival by multivariate analysis of 54 spinal chordomas. As hTERT is rarely identified in normal tissues, it may be a promising tumor-specific therapeutic target in chordoma.

SWI/SNF-related, matrix associated, actin dependent regulator of chromatin, subfamily B, member 1(SMARCB1, INI1), a component of the SWI/SNF chromatic remodeling complex is absent in poorly differentiated chordoma, particularly in the pediatric population. Seven patients whose chordomas had SMARCB1loss on fluorescence in situ hybridization (FISH), sequencing, or DNA methylation studies, had dismal outcomes with a median overall survival of 9 months. This suggests possible activity of novel EZH2 inhibitors in this small subset of aggressive chordoma.

MicroRNAs (miRNAs) are noncoding RNA involved in cell growth and apoptosis. Specific abnormal miRNA expression has been linked to chordoma carcinogenesis. Retrospective analysis of 42 patients with spinal chordoma found that patients with decreased miR-1237-3p had brief recurrence-free survival and increased risk of tumor invasion of surrounding muscles. MMP-2 may be the target gene, consistent with the increased invasiveness and worse clinical outcomes for chordomas with increased MMP-2 expression.

Since miR-155 expression also significantly inversely correlates with duration of overall survival (HR 5.32; 95% C.I. 1.04–27.36; P < .045), targeting it may also have therapeutic potential in chordoma.

Multiple molecular changes in chordoma may prove to be prognostic biomarkers, but evidence this far comes only from one study of small sample size.

Presence of Metastasis

Not surprisingly, patients who developed metastatic chordoma fare poorly relative to patients with localized disease. Median times from initial diagnosis to development of metastatic disease range from 0 to 13 years. Thi influence of local tumor control on local-regional recurrence and metastasis likely explains its influence on overall survival.

The site of chordoma metastasis is also prognostically significant. The median duration of overall survival is longer in patients with lung metastasis (130 months) is longer than that of patients with metastasis to the bone or liver (46 and 72 months, respectively).