Stroke & Vascular Disorders
Neoplastic and infectious aneurysms
In this article, the author reviews current knowledge about intracranial aneurysms due to infectious and neoplastic causes. Direct mural injury or invasion
Jul. 21, 2021
At vero eos et accusamus et iusto odio dignissimos ducimus qui blanditiis praesentium voluptatum deleniti atque corrupti quos dolores et quas.
This article includes discussion of chordoma; chordomas of the sacrum; chordomas of the skull base, clivus, and intracranial cavity; and chordomas of the true vertebrae. The foregoing terms may include synonyms, similar disorders, variations in usage, and abbreviations.
In this article, the author provides an in-depth review of the pathology, biology, clinical presentation, and treatment options for chordoma. These tumors are thought to derive from notochordal remnants and are locally invasive. Chordomas usually present at the skull base, but can arise anywhere along the spinal axis and pelvis. Aggressive surgical resection is the initial approach to treatment. However, in many cases the tumor cannot be completely removed. For residual tumors, radiotherapy is the most important treatment option. Chemotherapy continues to be investigated for therapeutic potential, especially molecular targeted agents.
• Chordomas are histologically non-anaplastic tumors that are locally invasive.
• MRI typically shows a locally invasive, enhancing mass within the bones of the clivus and skull base, sacrum, or vertebral bodies.
• A small percentage of cases can progress to systemic metastases.
• Gross total resection should be attempted in all cases and is correlated with improved local control and overall survival.
• Postoperative radiotherapy should be considered in all cases; heavy ion therapy (eg, protons) may provide a slight therapeutic advantage.
A chordoma of the clivus was first noted by both Virchow and Luschka in 1856 (224; 136). Virchow described the tumor as "ecchondrosis physaliphora spheno-occipitalis" and believed it was of cartilaginous origin. He used the term "physaliphora," meaning “bubble-bearing” because there were prominent cytoplasmic vacuoles. Muller suggested in 1858 that the origin of this tumor was the primitive notochord, the "chorda dorsalis" (153). The first description of a symptomatic chordoma was made in 1864 by Klebs in a patient with a tumor of the spheno-occipital region (117). In 1894 Ribbert was the first to use the term "chordoma," and further characterized Muller's theory by producing experimental chordomas after releasing tissue of notochordal origin from the nucleus pulposus of rabbits (178; 179). The tumors produced in these experiments were histologically similar to de novo chordomas. The experiments of Ribbert were replicated by Congdon in 1952, using a similar rabbit model (44).
A "chondroid chordoma" subtype, which has lacunae of hyaline cartilage, was described in 1973 by Heffelfinger and colleagues (91).
Approximately 50% of chordomas arise in the sacrum, 35% to 40% within the skull base and clivus, and 10% to 15% throughout the vertebral column (207; 89; 57; 77; 118; 138). In general, chordomas are relatively slow growing and often have a prolonged duration of symptoms before diagnosis. The specific symptoms and neurologic findings noted at presentation will vary according to the location of the tumor. Although these tumors are often benign histologically, systemic metastases have been noted in 10% to 40% of cases (207; 89; 77; 85; 165; 146). The most frequent sites for metastases are the lungs, regional lymph nodes, liver, bone, and skin.
Chordomas of the skull base, clivus, and intracranial cavity. Chordomas comprise 6.15% of all skull base tumors and 0.1% to 0.2% of all intracranial tumors (207; 89; 77). They occur most often in the clivus, but can arise in other areas such as the sphenoid sinus, cavernous sinus, occipital condyle, and sella (65; 207; 89; 236; 71; 225; 233; 164; 77; 05; 216; 66; 149; 232). Depending on the primary site of tumor involvement and direction of growth (eg, anterior, lateral, posterior), symptoms and signs may vary considerably. The mean age of patients with skull base chordomas is in the range of 38 to 45 years (225; 233; 05). In the majority of series, the most common symptoms are either diplopia or headache. Diplopia is the initial symptom in 50% to 90% of patients. The diplopia is usually horizontal and exacerbated by attempts at lateral gaze. Headache is noted at presentation in 25% to 60% of patients. In many patients, headache and diplopia develop simultaneously. Symptoms such as facial pain, vertigo, tinnitus, dysphagia, hoarseness, alterations of vision, and gait disturbance are present in 12% to 15% of patients (225; 233; 05). Infrequent complaints include hearing loss, dizziness, unilateral weakness, facial dysesthesias, and neck pain.
On neurologic examination, the most common findings are cranial nerve palsies (182; 207; 71; 225; 233; 77; 05; 149). The fifth cranial nerve is involved most frequently, with abnormal function noted in 45% to 75% of patients. The deficit is usually unilateral, but can be bilateral in some cases. Abducens palsy can be associated with dysfunction of cranial nerves II, III, IV, VI, and VII in 15% to 25% of patients. Although uncommon, isolated palsy of cranial nerves II, III, or IV is seen in some patients. Abnormalities of the lower cranial nerves (ie, IX, X, XI, XII) are noted in 25% to 40% of patients (71; 225; 233; 05). Similar to abducens palsy, the deficits are usually unilateral but can be bilateral in some cases. Cranial nerves of the cerebellopontine angle (ie, VII, VII) are rarely affected on examination at presentation but can develop in patients with large tumors. Pyramidal tract dysfunction is present in 15% to 20% of patients and develops from tumors that compress the ventral surface of the brainstem (207; 71; 77). The findings may be unilateral or bilateral and in some cases are associated with ataxia. Furthermore, patients with brainstem compression by tumor may manifest inappropriate laughing or crying. The emotional lability is thought to occur from disturbance of ventral pontine tegmental pathways.
Chordomas of the sacrum. Chordomas represent the most common primary neoplasm of the sacrum (137; 207; 89; 21; 194; 14; 197; 75). They often reach substantial size prior to diagnosis because of the ample room for tumor growth before critical structures are disturbed. The median age of patients in the majority of series is approximately 60 years; males are affected more often than females. The most common symptom (60% to 70% of patients) consists of persistent low back pain, which is slowly progressive and often present for 12 to 18 months before diagnosis. Patients occasionally complain of more specific locations of the pain, such as the coccygeal, buttock, or anal regions. The pain may have a radicular component to it, with radiation down 1 of the legs. This presentation often leads to the erroneous diagnosis of nonspecific "sciatica," delaying discovery of the tumor by many months. Rectal dysfunction consisting of alteration of bowel habits (ie, constipation), tenesmus, or bleeding is common (approximately 40% of patients). As the tumor continues to enlarge, it usually grows ventrally and may encroach on the sacral foramina and nerve roots, causing neurologic dysfunction. Symptoms from sacral nerve root compression are variable and include perianal numbness, urinary hesitancy or retention, urinary incontinence, impotence, and rectal incontinence. The general physical is typically benign, except for the rectal examination, which often demonstrates a presacral mass (207; 21). The neurologic examination may be normal or show evidence for sacral root dysfunction (eg, perianal numbness, loss of anal sphincter tone).
Chordomas of the true vertebrae. Chordomas are uncommon tumors of the vertebral column (usually the vertebral body), representing less than 5% of all tumors in this region (57; 118; 235; 14; 197). Approximately 60% of vertebral chordomas arise in the lumbar region; 10% to 15% develop in the thoracic area, and 25% to 30% in the cervical spine. Rarely, chordomas can develop as extra-osseous, intradural spinal masses (13). Ventral tumor growth will cause bone destruction and infiltration into paraspinal soft tissues, whereas dorsal expansion may cause nerve root displacement or spinal cord compression. The mean age of patients with spinal chordomas ranges from 45 to 50 years. Patients with these tumors often have a shorter duration of symptoms before diagnosis than patients with tumors of the sacrum, due to the smaller volume of bone in proximity to sensitive neural structures. In 1 series, the mean duration of symptoms prior to diagnosis was 7 months (118). In the majority of cases (more than 90%), the initial symptom is localized pain in and around the involved vertebral body (235). There may be a radicular component to the pain from displacement or compression of nerve roots, with lancinating pain into a limb or anteriorly around the thorax. Other alterations of sensation such as dysesthesias or sensory deficits may occur. Cervical chordomas that grow ventrally and compress the esophagus may cause dysphagia (57). Occasionally, tumors can cause myelopathic weakness, gait ataxia, or sphincter dysfunction.
Several authors have attempted to correlate pathological features of chordomas with prognosis (182; 71; 164). In a series of 48 mixed chordomas, Rich and colleagues were unable to detect a correlation between cellular pleomorphism, mitotic figures, or hyperchromatic nuclei with survival (182). The only histologic variable to correlate with survival was the presence of chondroid elements. Chondroid chordomas had a more indolent course, longer duration of symptoms, and increased survival. Forsyth and colleagues evaluated 51 intracranial chordomas and were unable to detect a correlation between mitosis, chondroid elements, and survival (71). In a series of 62 skull base chordomas, O'Connell and colleagues found that tumors with greater than 10% necrosis were associated with shorter patient survival (164). The presence of chondroid elements, mitoses, pleomorphism, nucleolar prominence, and vascular invasion were not correlated with overall survival. An evaluation of molecular markers in a cohort of 16 patients with skull base chordoma suggests that 1p36 loss of heterozygosity or the presence of tumor necrosis factor receptor SF8 (TNFRSF8) expression might be associated with a better prognosis (135). A report has correlated cytogenetic analysis with risk of recurrence and impact on survival time (04). Karyotypic analysis was performed on a series of 64 patients with skull base chordomas. In 74% of the cases, the karyotypes were normal, and the recurrence rate was only 3%. Tumors with abnormal karyotypes had a 45% recurrence rate--a significant difference (p < 0.01). In general, recurrent tumors had a high rate of abnormal karyotype (75%). Tumors with abnormal karyotypes had an odds ratio for recurrence of 12.45, in comparison to those with normal karyotypes (p < 0.01). Alterations in chromosomes 3, 4, 12, 13, and 14 were associated with frequent recurrence and decreased survival time. Tumors with simultaneous aberrations of chromosomes 3 and 13 had a very high recurrence rate and median survival time of only 4 months. A study analyzed the pathological data from 43 chordomas (24 sacral, 13 spine, 5 clivus, 1 nasal septum), and it determined that the presence of metastases, a Ki-67 labeling index of greater than or equal to 10%, and an extracellular matrix poor phenotype, were all associated with poor overall survival (p < 0.05) (226). The extracellular matrix poor phenotype and the presence of metastases were also associated with a higher Ki-67 labeling index (p < 0.05). Another study evaluated the expression of EGFR, c-MET, and c-Erb-B2, and correlated the results with recurrence and overall survival (218). EGFR was noted to be expressed in the majority of cases and did not correlate with recurrence. However, c-MET was not expressed in the majority of cases, but when it was present was typically associated with recurrent tumors (p < 0.05). A more recent report analyzed the presence of 1p36 and 9p21 chromosomal alterations, as well as Ki-67 labeling index in clival chordomas, and correlated the results with progression-free survival after surgical resection and radiotherapy (243). In univariate analysis, Ki-67 labeling index, 1q25 hyperploidy, 1p36 deletions, and 9p21 deletions were all found to be predictive of progression-free survival after surgery and radiotherapy. However, on multivariate analysis, only 1p36 deletions and 9p21 deletions were demonstrated to be independently predictive.
Several reports have attempted to correlate various clinical parameters with overall survival and prognosis (71; 164). Patients less than 40 years of age appear to have improved survival and a better prognosis. Forsyth and colleagues noted a significant difference in survival (5-year survival in 75% versus 30%) for patients less than 40 years of age (p less than 0.0001) (71). Although uncommon, when chordomas occur in children and adolescents, some reports suggest they may have a more indolent course and longer patient survival (93). In a review of 73 young patients with skull base chordomas, Hoch and colleagues noted an overall survival of 81% at a median of 7.25 years follow-up. The presence of diplopia was also suggestive of a better prognosis and improved survival, especially when correlated with patient age. Female sex was associated with improved survival (median 158 months vs. 86 months; p less than 0.004) and a better prognosis in both univariate and multivariate analyses by O'Connell and colleagues (164). During a review of 39 patients with sacral and mobile spine, inadequate surgical margins, tumor necrosis, and Ki-67 labeling index greater than 5% were all found to be independent markers for adverse outcome, local recurrence, and metastases (19). Zou and colleagues have evaluated clinical prognostic factors and surgical approaches in a series of 347 patients with clival chordoma, and correlated the results with progression-free survival and overall survival (247). The 5- and 10-year progression-free survival rates were 59.2% and 47.9%, respectively. The 5- and 10-year overall survival rates were 77.3% and 63.9%, respectively. On multivariate analysis for progression-free survival and overall survival, gross total resection demonstrated significantly improved outcomes when compared with subtotal resection (HR = 0.45; p = 0.025 for progression-free survival, p = 0.008 for overall survival). Using the SEER database, Lee and coworkers reviewed a series of 1598 chordoma patients and performed univariate and multivariate analyses to examine prognostic factors affecting progression-free survival and overall survival (129). Multivariate analysis demonstrated that older age (p = 0.002), greater tumor size (p < 0.001), and distant metastasis (p < 0.001) were correlated with decreased progression-free survival and overall survival. In contrast, surgical resection (p < 0.001) was correlated with increased progression-free survival and overall survival.
The median overall survival for all types of chordoma, including all races and genders, is 6.29 years (145). The overall 5- and 10-year relative survival rates for all types of chordoma are 67.65% and 39.9%, respectively. With maximal surgery and appropriate radiation therapy, patients with clivus chordomas have a 5-year tumor control rate of 82% and a 10-year survival rate of at least 67% (115). Patients with sacrococcygeal tumors have a 5-year survival rate of 77% and a 10-year survival rate of 50% to 64% (190; 19). A review of the world’s literature regarding intracranial chordoma by Jian and colleagues cited a total of 560 cases and 5- and 10-year survival rates of 63% and 16%, respectively (111). When stratified by age with a cut-off at under or over 40 years of age, there was no difference in 5-year survival (p = 0.1). However, when 5 years of age was used as the cut-off, there was a significant difference in 5-year survival, with the older patients doing better (< 5 years: 14%, > 5 years: 66% (p = 0.01)). A similar analysis of cranial chordoma using the SEER registries calculated the 5-year survival rates over time starting in the mid-1970s (36). For a total of 594 cases, the overall median survival time with surgery plus radiotherapy was 9.2 years. Age over 50 years was associated with a significant increase in mortality rate (p < .05). The 5-year survival rates for the 1975 to 1984, 1985 to 1994, and 1995 to 2004 cohorts were 48.5%, 73.0%, and 80.7%, respectively. Improved survival was noted for the more recent cohorts (p < 0.01).
Complications of chordomas vary depending on the location and size of the tumor, as well as on the form of treatment. In general, patients with skull base tumors are most likely to develop cranial nerve palsies (usually VI, V, III, IX, X, VII), hemiparesis, and gait disturbance (198; 125; 77; 76). Less common complications include cerebrospinal fluid leak, hydrocephalus, hypopituitarism, hemorrhage, ataxia, and leptomeningeal spread with drop metastases (220). The most common complications of sacral chordomas are persistent pain (low back or radicular), urinary retention, and rectal incontinence (89; 21; 170). Less frequent complications include urinary incontinence, sexual dysfunction, lower extremity weakness, systemic metastases, with rare spread to the cerebrospinal fluid, cutaneous metastases, and pudendal neuralgia (110; 34; 80). In a review of 39 patients with chordomas of the sacrum and mobile spine, the incidence of systemic metastases was 28%, mostly to the lungs and soft tissues (19). A series of 37 patients with spinal chordomas had a 19% rate of metastasis, involving the lungs, liver and lungs, or remote spine (146). Sites of metastasis were stabilized in several patients with the use of chemotherapy (gemcitabine- or Gleevec-based). Tumors of the spinal vertebrae can also metastasize to other vertebral bodies along the spinal axis (53). The most likely complications related to chordomas of the vertebrae are persistent back pain (localized, radicular, or both), dysesthesias, lower extremity weakness, and gait disturbance (208; 118). Less frequent complications include urinary retention or incontinence, dysphagia, fecal incontinence, and ataxia. Another recognized complication is seeding of chordoma cells along the surgical resection pathway (08). In their review of 82 patients, Arnautovic and al-Mefty noted an incidence of 7.3%, with a mean time to tumor growth of 12 months. Surgical seeding could occur from skull base and spinal chordomas.
Clivus chordoma. A 29-year-old male with an unremarkable past medical history presented with diplopia and headache. These symptoms were slowly progressive over several months and were eventually accompanied by complaints of difficulty with speech and gait. MRI scan revealed a large, enhancing, clival-based lesion with extension anteriorly into the sinuses.
On some of the higher cuts, posterior extension of tumor was causing compression of the brainstem. The neurologic examination demonstrated a partial left-sided third nerve palsy, right hemiparesis, and gait instability. The tumor was partially resected using a combined subfrontal-transethmoidal skull base approach. The pathology was consistent with chondroid chordoma; no sarcomatous features were evident. In the immediate postoperative period, the patient developed a stroke affecting the left side of the brainstem. He made an excellent recovery from the surgery and stroke during rehabilitation. Concomitantly, a 6-week course of external beam radiation therapy was administered. The patient was clinically stable on neurologic examination and follow-up MRI after several years.
Vertebral chordoma. A 51-year-old woman with an unremarkable past medical history presented with pain in the upper neck region that was slowly progressive over 4 to 6 months. These symptoms were initially attributed to arthritis and degenerative joint disease. In the last few months, the pain was accompanied by a fullness in the back of her throat and dysphagia. The dysphagia was mild at first but became more severe over several weeks. MRI scan demonstrated an enhancing 3 cm mass arising ventrally from the C3 vertebral body and extending into the posterior aspect of the tongue musculature.
On neurologic examination she was nonfocal, with intact cranial nerves and motor function. The patient was brought to the operating room for surgical resection of the lesion using a transoral approach and a temporary tracheostomy. An aggressive subtotal resection was performed in combination with C2 through C4 posterior fusion and halo placement. Pathology was consistent with a typical chordoma, without chondroid or sarcomatous features. Postoperatively, the patient received a 6-week course of external beam radiation therapy while undergoing in-patient rehabilitation. She remains neurologically intact and has noted significant improvement of the cervical spine pain and complete resolution of the dysphagia.
The modern theory of the origin of chordomas proposes that the tumors derive from embryonic rests of the primitive notochord that persist within the axial skeleton (207; 89; 77). The notochord forms from ectodermal cells during the third or fourth week of development and is believed to act as an embryonic organizer (187). During the fourth to sixth weeks of development, mesenchymal cells from adjacent sclerotomes envelop the notochord as they merge to form the spinal vertebral bodies (187; 77). The notochord degenerates during this process, and by the seventh week it remains only between the vertebral bodies as the nucleus pulposus of the intervertebral discs. Pathologic studies of tissue using both light and electron microscopic techniques have demonstrated similarities between chordoma and human intervertebral disc (187; 77). It is postulated that incomplete degeneration of residual notochord may occur within the vertebral body at the junction of the adjacent sclerotomal regions. These incompletely degenerated rests can potentially undergo malignant transformation and develop into a chordoma. Investigations of the persistence and regression of the human notochord in fetuses of 4 to 18 weeks' gestation suggest great variation in this process and that the presence of aberrant notochordal tissue is not uncommon (97). Furthermore, the topographical distribution of these heterotopic notochordal rests correspond closely to the common sites of chordoma in the adult (ie, sacrococcygeal, clivus) (187; 187). Autopsy studies reveal rests of presumed notochordal tissue anterior to the clivus and around the sacrum in up to 2.0% of cases (207; 89). Finally, a shared immunophenotype is noted between notochordal and chordoma cells, with both types of cells containing S100 protein, cytokeratin, and human epithelial polymorphic mucin (187). Although the initial genesis of cellular transformation is unknown, various contributory cytogenetic, chromosomal, and molecular biological events are discussed in this review.
Chordomas are generally slow-growing, unencapsulated neoplasms that are locally invasive within bone and soft tissues (182; 207; 89; 77; 193; 138). A pseudocapsule may be noted around tumors that grow into soft tissues or the dura mater. As the tumors enlarge, they often stretch cranial nerves and displace structures such as blood vessels and the brainstem. Grossly, the tumors are usually reddish or purple in color, with a nodular appearance to the surface. Internally, the mass is frequently gelatinous and soft; regions that contain cartilage or calcium are more firm. Foci of hemorrhage may be present and can be small or extensive. The size of the lesion can be variable, with sacral tumors often becoming extremely large. In 1 series of cranial base chordomas, average tumor volume was 58 cm3 (77).
On microscopic examination, chordomas can be grouped into several different histological categories, including a typical pattern, a chondroid pattern, and tumors with features of malignant degeneration (182; 207; 89; 39; 71; 164; 77; 193; 45; 138). The typical or classic pattern of chordoma (65% to 80% of all cases) is distinguished by a lobular arrangement, with the neoplastic cells disposed in solid sheets or irregular intersecting cords.
The sheets and cords of cells are set in a stroma that contains an abundant mucinous matrix. The individual cells are large, often with vacuolated eosinophilic cytoplasm and contain variable amounts of mucin. The cell type considered diagnostic for chordomas is called physaliphorous. These cells are distinctively large and vacuolated, with eccentric nuclei.
Nuclei tend to be hyperchromatic, with prominent nucleoli, and rarely demonstrate atypia. Potentially aggressive features such as mitoses, necrosis, hypervascularity, and spindle cells (ie, sarcomatous degeneration) are typically absent or rare (182; 39; 71; 164; 193). DNA ploidy analysis of typical chordomas demonstrates aneuploidy in 15% to 40% of cases (207; 89; 77; 193). There is a trend for tumors with aneuploid DNA content to behave more aggressively and for patients with these tumors to have shorter survival (193). Data would also suggest that nuclear pleomorphism is correlated with higher proliferative rates in skull base and nonskull base chordomas (156). In addition, the presence of nuclear pleomorphism was associated with a poorer prognosis and shorter survival time (p=0.027) in patients with nonskull base tumors.
Many authors contend that the chondroid pattern (15% to 30% of all cases) is a separate histological variant of chordoma, although this is controversial (91; 182; 207; 89; 39; 71; 106; 164; 77; 193; 138). The chondroid pattern has been associated with a more favorable prognosis, as originally described by Heffelfinger and colleagues (91). However, other authors contend that chondroid chordomas are a subgroup of low-grade chondrosarcoma and are not related to chordomas (24; 29). By definition, chondroid chordomas contain regions of typical chordoma with physaliphorous cells, against a background of areas characterized by cartilaginous matrix that have stellate tumor cells occupying lacunar spaces (resembling chondrocytes) (91; 106).
As in typical chordoma, anaplastic or aggressive features such as mitoses, necrosis, hypervascularity, and spindle cells are typically absent or rare (91; 182; 39; 71; 106; 164; 193). Electron microscopic studies support the dual nature (ie, epithelial-mesenchymal) of these neoplasms by identifying cells with epithelial features and other cells consistent with chondrocytes (223).
Chordomas with malignant degeneration (less than 5% of all nonirradiated cases) typically demonstrate sarcomatous features (ie, spindle cells) (207; 89; 98; 99; 74; 217; 77; 45). These tumors will contain areas of classic chordoma admixed with regions characterized by the presence of atypical spindle cells. The spindle cell component demonstrates high cellularity, marked nuclear pleomorphism, and a high mitotic rate. Within the malignant spindle cell zones, regions of cartilaginous or osseous differentiation may be noted. In many tumors, a transitional zone may be present between the regions of typical chordoma and regions containing the malignant spindle cells. This transitional zone may contain an intermediate, stellate shaped type of cell. DNA ploidy analysis of chordomas with sarcomatous degeneration usually demonstrates aneuploid cell populations with a high proliferating fraction (99; 217). The mean proliferating fraction (%S+G2M) of a series of spindle cell chordomas was 34.1% (99). In comparison to the mean proliferating fraction of a series of typical chordomas (20.2%), the growth fraction of spindle cell chordomas was significantly larger (p < 0.01).
Work by Camacho-Arroyo and colleagues has demonstrated the presence of sex steroid receptors in chordomas (33). In a series of 6 patients, all of the samples were positive by immunohistochemical analysis for progesterone receptors, with a predominance of isoform B. In addition, all of the tumors stained for estrogen receptor alpha. Similar immunohistochemical data suggest that E-cadherin is frequently expressed in chordoma cells and may be useful as a diagnostic marker (151). An immunohistochemical analysis confirms the presence of E-cadherin, although in this report it was mainly contained within the nucleus (96). In addition, frequent expression of other adhesion molecules was also noted, including neural cell adhesion molecule (NCAM), vascular cell adhesion molecule-1 (VCAM-1), CD44, and N-cadherin. Other authors have extended these preliminary findings and confirmed that chordomas routinely express E- and N-cadherin (219). However, the expression is inversely correlated, with a predominance of N-cadherin expression in tumors with a more aggressive and infiltrative phenotype. The up-regulation of N-cadherin was associated with a 3.28-fold increased risk for tumor recurrence and a 10.98-fold increased risk for death from the tumor.
Preliminary data from studies using cytogenetic and molecular techniques are beginning to elucidate the mechanisms of transformation of chordomas (28; 32; 64; 144; 206; 83). No specific or characteristic chromosomal abnormalities have been described thus far. Although some of the cases have had normal karyotypes, others have shown hypodiploidy or near diploidy, which is in contrast to the DNA flow cytometry data (30). A cytogenetic analysis of different subtypes of chordoma suggests that conventional chordomas often display a near diploid karyotype (78). In contrast, dedifferentiated chordomas are more likely to have polyploidy, complex karyotypes that may be related to their more aggressive biological behavior and poor prognosis. Several tumors have shown structural anomalies of chromosomes 1 and 21, whereas others have alterations (usually elongation) of the telomere (28; 32; 49). In particular, deletions of chromosome 1p appear to be frequent and warrant further study (49). An evaluation of 27 chordomas, using loss of heterozygosity microsatellite analysis, further localizes the important region of chromosome 1 to a span encompassing 1p36.32-36.11 (183). The minimal loss of heterozygosity interval shared by 85% of the tumors was 1p36.13. These data were corroborated by Longoni and associates in a review of 16 patients with skull base chordomas (135). Genes in this region that may have oncogenic roles as tumor suppressors include CASP9, EPH2A, and DVL1. Other authors have found abnormalities on chromosome 7, including familial linkage and chromosomal gains (27; 239; 18). Yang and colleagues used single nucleotide polymorphism (SNP) analysis on several chordoma kindreds to demonstrate a linkage to chromosome 7q33 (239). The region on 7q encompasses 16 megabases with 128 known or putative genes, including growth factors, receptors, and signaling proteins. Brandal and colleagues applied comparative genomic hybridization to 6 sacral chordomas and noted gains of chromosome 7 in the 7p21-22 and 7q regions (27). Using immunohistochemical techniques, Matsuno and colleagues found that p53 protein was present in chordomas that had a high proliferative index (as determined by MIB-1 staining) and in recurrent tumors (144). Furthermore, a significant correlation was noted between cyclin D1 staining and MIB-1 proliferative index or tumor recurrence. Interestingly, none of the tumors evaluated for bcl-2 were immunopositive, suggesting that apoptosis may not contribute to the recurrence of chordomas. Other authors have noted a correlation between the presence of mutated p53 protein and expression of telomerase reverse transcriptase mRNA (171). In these patients, the tumors tended to exhibit rapid growth rates and rapid recurrence. Data by Bergh and colleagues have demonstrated that a Ki-67 labeling index of greater than 5% is an independent marker for an adverse outcome (19). This has been corroborated by Horbinski and colleagues in a series of 28 cases of clival chordoma (95). They noted that tumors with a Ki-67 of 5% or greater were associated with a more aggressive clinical course and shorter overall survival. In addition, a similar poor prognosis was associated with loss of heterozygosity of chromosome 9p, especially at 9p21. Eisenberg and colleagues studied 7 skull base chordomas and found that in 2 aggressive cases, there was loss of heterozygosity for the Rb tumor suppressor gene on chromosome 13, suggesting that alterations or loss of Rb may play a role in the transformation of chordomas (64). In contrast, Klingler and colleagues studied 12 patients with sacral tumors and did not find loss of heterozygosity for Rb (119). However, they did find microsatellite instability in 6 of 12 (50%) tumors, suggestive of defective DNA mismatch repair. Other researchers have performed a more in-depth analysis of the proteins involved in control of the G1-S checkpoint (155). Naka and colleagues noted p53 overexpression in approximately 30% of a series of 101 chordomas; p53 gene mutations were not present. MDM2 gene amplification and overexpression were noted in 15% and 13% of the tumors, respectively. p53 overexpression was correlated with a higher MIB-1 labeling index and associated with a poorer prognosis and shorter survival times. Kelley and colleagues have done a genetic linkage analysis of a cohort of patients with familial chordoma (116). The results suggest the presence of a locus for familial chordoma at chromosome 7q33. A similar study was performed on 4 multiplex families with familial chordomas, using high-resolution array comparative genomic hybridization (CGH) (240). A new duplication was noted in the region of 6q27, which contains the T (brachyury) gene. The T gene is important for notochordal development and is expressed in most sporadic chordomas. Data corroborate the importance of the T (brachyury) gene in the pathogenesis of chordoma (206; 205). Presneau and colleagues analyzed the chordomas from 181 patients and noted frequent abnormalities and aberrations of the chromosome and gene (175). In 12 of 181 cases (7%) there was amplification of the T locus, with an additional 2 cases that showed focal amplification of the gene. Another 70 of 181 tumors (39%) were polysomic for chromosome 6, whereas 8 of 181 primary tumors (4.5%) showed a minor allelic gain of T on FISH analysis. When the T gene was knocked down in the chordoma cell line, U-CH1, which shows polysomy of chromosome 6q27, there was a marked decrease in cell proliferation. The same group then performed a functional genomics analysis of the brachyury gene in human chordoma samples, which demonstrates that the gene directly binds 99 other genes and indirectly influences the expression of another 64 genes (160). Brachyury appears to be in control of an oncogenic transcriptional network that involves signaling pathways that include components of the cell cycle and extracellular matrix. Another study suggests that 1 particular variant of brachyury is most prevalent in chordoma (173). The authors performed a single-nucleotide polymorphism analysis of the brachyury gene in a sample of 40 patients with chordoma. The common non-synonymous single nucleotide polymorphism rs2305089 was highly associated with risk of chordoma, with an odds ratio of 6.1. Overall, the extent of brachyury expression does not seem to correlate with prognosis in chordomas (205). An immunohistochemical evaluation of expression of transforming growth factor alpha (TGFa), basic fibroblast growth factor (bFGF), vascular endothelial growth factor (VEGF), and several structural proteins was performed by Deniz and colleagues, and correlated with chordoma recurrence (54). The results suggest that high levels of expression of TGFa and bFGF are associated with higher rates of recurrence. In addition, strong expression of fibronectin was associated with more aggressive biological behavior. A series of 6 patients with advanced chordomas (sacrum 5, clivus 1) was evaluated for the expression of platelet-derived growth factor receptors (35). All 6 tumors demonstrated significant expression of platelet-derived growth factor receptors as determined by reverse-transcriptase PCR. Data from other researchers are in agreement and suggest that platelet-derived growth factor and platelet-derived growth factor receptors are frequently expressed in skull base chordomas, consistent with the presence of autocrine and paracrine loops (212; 168; 67). A similar study was performed to examine the expression of c-Met, HER2/neu, and epidermal growth factor receptor in a series of 12 chordomas (234). Strong expression of c-Met and epidermal growth factor receptor was noted in most chordomas, with variable expression of HER2/neu. In addition, a strong correlation was noted between epidermal growth factor receptor and c-Met expression. Because the gene for c-Met is located at 7q31, it may be amplified in some tumors, consistent with the findings of Brandal and colleagues, noted above. Another study analyzing c-Met noted its expression in 70% of primary skull base chordomas and 88% of recurrent tumors (158). Hepatocyte growth factor was minimally detectable in tumor tissues. c-Met expression correlated with younger patient age, the expression of low molecular weight cytokeratin (CAM5.2), and a favorable prognosis. It did not correlate with the MIB-1 labeling index. A study correlated changes in copy number of chromosome 7, EGFR expression, and c-MET expression in a series of 22 chordomas (229). Aneusomy of chromosome 7 was noted in 73% of the samples. There was a significant correlation with increased copy number of chromosome 7 and overexpression of c-MET (p < 0.001). An analysis of 173 chordomas was performed by Shalaby and colleagues to further characterize the role of EGFR in the pathogenesis of chordoma (200). Using immunohistochemical techniques, total EGFR expression was noted in 69% of samples, whereas FISH analysis revealed high-level EGFR polysomy in 38%. Direct sequencing of EGFR did not reveal any mutations. Use of an EGFR inhibitor (AG 1478) resulted in significant inhibition of the chordoma cell line, U-CH1 in vitro, and reduced phosphorylation of EGFR. Other investigators have also noted increased expression and activation of EGFR in chordomas, especially advanced cases (56). In a study by Dewaele and colleagues, EGFR was the most frequently and significantly activated of the receptor tyrosine kinases. A report by Akhavan-Sigari and colleagues reviewed the molecular phenotype from 145 skull base chordoma tumor specimens (02). They noted the following results: PDGFR-α was detected in 100% of specimens, EGFR was present in 92% of specimens, c-Met was detected in 100% of specimens, and CD-34 was present in 59% of specimens. Patients with higher expression of PDGFR-α and EGFR were found to have reduced overall survival.
Studies by Naka and colleagues has analyzed the expression of tissue invasion-related enzymes in skull base chordoma and spinal chordoma (159; 157). They noted frequent expression of matrix metalloproteinase-1 and -2 (MMP-1 and MMP-2), tissue inhibitors of matrix metalloproteinases-1 and -2 (TIMP-1 and TIMP-2), cathepsin B (CatB), and urokinase plasminogen activator (uPA) in skull base tumors (158). In spinal chordoma, the expression of c-MET was correlated with the expression of CAM5.2 and stronger expression of MMP-1 and MMP-2 (157). Immunoreactivity for these proteins was significantly higher in tumors with aggressive infiltration into surrounding bone. In addition, higher expression of matrix metalloproteinase-1 and urokinase plasminogen activator was associated with a worse prognosis. Similar work by Chen and colleagues evaluated the expression levels of MMP-9 and VEGF in sacral chordomas (38). They noted higher expression of both MMP-9 and VEGF, and higher microvascular density, within tumor tissue in comparison to surrounding normal tissues. A significant correlation was noted between expression of MMP-9 and VEGF (p = 0.002). Continuous disease-free survival time was significantly shorter in the MMP-9 positive group when compared to MMP-9 negative controls. An evaluation of sex steroid receptor expression (ie, estrogen-alpha and -beta, progesterone, androgen) in chordomas was performed by Fasig and coworkers (68). Most tumors were noted to express estrogen-beta and androgen receptors, but not estrogen-alpha or progesterone receptors. The expression of estrogen-beta and androgen receptors did not correlate with each other. Investigators have now begun to evaluate the involvement of the signal transducers and activators of transcription (Stat3) pathway in chordomas (237). In a series of 70 chordomas, every tumor had evidence of phosphorylated Stat3 (pStat3) in the nuclei of stained tissue samples on immunohistochemical analysis. The expression of pStat3 correlated with patient survival and the severity of disease. The high-staining group had a significantly shorter median survival (p = 0.039) and was more likely to recur (p = 0.0002) and have metastases (p = 0.01) than the low-staining group. Using SD-1029, an inhibitor of Stat3 activation, it was noted that the growth of 3 different chordoma cell lines was inhibited, with reduced activation of the Stat3 signaling cascade. When used in combination with chemotherapy (cisplatin, doxorubicin), the inhibition of chordoma cell lines was augmented and more effective than any of the drugs used alone. SD-1029 was also effective against chordoma cells growing in 3-dimensional cultures. Similar studies have examined the role of the Akt/mTOR signaling pathway in chordomas (87; 174; 196; 56; 51). Activation of the Akt/mTOR pathways was demonstrated by all of the groups, with increased expression of phosphorylated Akt and mTOR and frequent loss of expression of the PTEN tumor suppressor gene. In the study by Han and colleagues, the mTOR inhibitor, rapamycin, was used against chordoma cell lines and was able to reduce mTOR activation and inhibit cellular proliferation (87). Schwab and colleagues used PI-103, an inhibitor of PI3K/Akt and mTOR activation, on chordoma cell lines and noted a reduction in activation of the Akt and mTOR pathways. In addition, PI-103 was able to inhibit proliferation and induce apoptosis in chordoma cell lines. In a study by Dewaele and colleagues, there was also significant co-activation of the PI3K/Akt pathway and mTOR, as well as loss of PTEN, although EGFR was the most frequently activated molecular marker (56). The study by de Castro and colleagues noted significant expression (75%-95%) of PDGFR, c-kit, c-MET, pAkt, and mTOR, with less frequent expression of EGFR (51). The only factor with a direct correlation with shorter survival rates was expression of pAkt (p = 0.042). The 5-year survival rate for patients with pAkt-positive tumors was only 45%, whereas for those with pAkt-negative tumors it was 100%. Scheipl and colleagues analyzed a cohort of 50 chordomas for the presence of insulin-like growth factor receptor (IGF-1R) and ligand (IGF-1, IGF-2) using immunohistochemical techniques (192). They demonstrated that 76% of chordomas express IGF-1R, 92% express IGF-1, and 50% express IGF-2. In addition, overall size of the tumor (ie, tumor volume) correlated significantly with IGF-1R staining intensity in primary chordomas (p = 0.042). An analysis of a related protein, insulin-like growth factor II mRNA-binding protein 3 (IMP3) in a series of sacral chordomas was performed by Zhou and colleagues (246). They noted expression of IMP3 in 20 of 32 tumor samples (62.5%). IMP3 was associated with tumor invasion into surrounding soft tissues (p - 0.028), high levels of Ki-67 expression (p = 0.009), and tumor recurrence (p = 0.012). Patients with positive expression of IMP3 had shorter disease-free survival times than patients with tumors that did not express the protein (p = 0.016). A report from Feng and colleagues reviewed the expression of bone morphogenesis protein 4 (BMP4) and SMAD (phospho-SMAD5 and SMAD4) in a series of 40 skull base chordomas (69). High expression within this pathway was correlated with larger tumors (≥ 4 cm) and the presence of dural invasion. In addition, the 5-year overall survival rate for high expression was significantly lower than for low expression: 71.4% versus 90.2% (p = 0.010).
Using the technique of array comparative genomic hybridization, Hallor and colleagues evaluated a series of 21 chordomas for significant copy number changes (86). Most of their results were consistent with the data listed above, except for a consistent loss of the CDKN2A and CDKN2B loci in 9p21, which were homo- or heterozygously lost in 70% of the tumors. These findings were subsequently corroborated by FISH analysis, suggesting that inactivation of CDKN2A and CDKN2B are involved in chordoma development.
Liu and colleagues evaluated the expression of CDK4 in chordoma cell lines and resection specimens using immunohistochemical and microarray techniques (134). CDK4 was found to be highly expressed in chordoma cell lines. In the clinical specimens, CDK4 was found to be expressed in the majority of tumors (97.7%). There was also a loss of expression of p16 in all of the chordoma cell lines and clinical specimens. Higher expression of CDK4 was correlated with the presence of metastasis and an increased risk of tumor recurrence.
An evaluation of the expression of survivin, which is important for the inhibitory control of apoptosis in tumor cells, was performed in a series of 30 patients with sacral chordomas (37). Significant expression of survivin was noted in 70% of the cohort. Survivin expression was significantly higher in the group of tumors with recurrence in comparison to the group without recurrence (p = 0.017), and was inversely correlated with disease-free survival time (p < 0.001). A similar study by Froehlich and colleagues evaluated the expression of survivin in a series of 50 chordomas, as well as 3 chordoma cell lines (72). Using the agent YM155 for survivin knockdown experiments, they noted decreased growth behavior in chordoma cells in a dose- and time-dependent manner. Use of YM155 led to G2/M arrest, decreased proliferation, an increase in polyploidy, and the induction of apoptosis.
Zhou and colleagues have used the technique of differential proteomic profiling to evaluate a series of 37 chordomas (245). They identified 14 up-regulated and 5 down-regulated proteins in chordomas, including alpha enolase (ENO1), pyruvate kinase M2 (PDM2), and gp96. On univariate analysis, overexpression of ENO1 and PKM2 where shown to be adverse prognostic factors for disease-free survival. On multivariate analysis, adverse prognostic factors included inadequate surgical margins and multiple contiguous vertebral levels; protein expression was not significant.
Using whole-genome single-nucleotide polymorphism microarray analysis, Diaz and colleagues studied a series of patients with skull base and sacral chordomas (59). They noted absent or reduced expression of fragile histidine triad protein (FHIT) in 98% of sacral chordoma specimens and 67% of skull base chordomas. These results suggest that epigenetic regulation of FHIT contributes to the loss of the FHIT tumor suppressor gene in chordomas, and is an important aspect of chordoma pathogenesis.
Micro-RNA (mi-RNA) expression has been shown to be important in many solid tumors, so several authors have now used mi-RNA micro-array technology to analyze the expression patterns in a series of chordoma cell lines and resection specimens (63; Gulluoglu et al 2015; 206). Several mi-RNA had differential expression in comparison to controls, including mi-RNA-1 and mi-RNA-206. Both types of mi-RNA were noted to have significantly decreased expression in tumor tissue, with overexpression of target gene products, such as c-MET and HDAC4. After transfection of mi-RNA-1 into tumor cells, the expression of c-MET and HDAC4 was reduced. Another study compared the mi-RNA profile from a series of chordoma tumors with normal nucleus pulposus samples (17). The analysis demonstrated that 30 mi-RNAs were upregulated in tumor cells, whereas 23 mi-RNAs were downregulated. In most of the tumors, hsa-miR-140-3p and hsa-miR-148a were upregulated relative to nucleus pulposus cells. Similarly, hsa-miR-31 and hsa-miR-222 were usually downregulated in most of the tumor specimens. Functional analyses suggest that hsa-miR-31 has an apoptotic effect on chordoma cells and downregulates the expression of c-MET and radixin. A similar study has shown that the expression of miR-608 and miR-34a are downregulated in chordoma (244). These 2 micro-RNAs seem to function as tumor suppressors, target EGFR/Bcl-xL and MET, respectively. Furthermore, EGFR and MET activation is shown to promote chordoma cell proliferation and invasion. Pharmacological inhibition of EGFR and MET induces arrest of cell proliferation and invasion in chordomas. A similar degree of inhibition occurs when miR-608 and miR-34a expression and activity are restored. Overall, miR-34a inversely correlates with MET expression, whereas miR-608 inversely correlates with EGFR expression in chordoma cells. A study by Osaka and colleagues analyzed the expression of miR-155 in a series of chordoma specimens (169). miR-155 was highly expressed and biologically active in the specimens, and it significantly correlated with disease stage, presence of metastases, and poor outcome (p = 0.045; hazard ratio, 5.32). Inhibition of miR-155 expression was noted to suppress proliferation, migratory activity, and invasive capacity in chordoma cells.
Marucci and colleagues evaluated the MGMT promoter methylation status of nonrecurrent and recurrent clival chordomas (141). MGMT is a tumor treatment resistance enzyme that removes adducts to DNA that can occur during radiotherapy and/or chemotherapy. When the promoter is methylated, expression is reduced, making the cells more susceptible to treatment. In all of their tested nonrecurrent chordomas, the MGMT promoter was always unmethylated. In contrast, in a significant portion of the recurrent chordomas, the MGMT promoter was methylated (p = 0.0317). These results prompted the authors to speculate if recurrent clival chordomas might be responsive to temozolomide chemotherapy as adjuvant therapy.
Chordomas are rare neoplasms, representing only 0.1% to 0.2% of all intracranial tumors, 6.15% of all primitive skull base tumors, and 1% to 4% of primary malignant bone tumors (207; 89; 77; 145). The overall incidence rate in the United States is 0.8 per 100,000 (145). Chordomas can arise anywhere within the midline axial skeleton where the notochord existed (ie, clivus, sellar and parasellar region, nasopharynx, foramen magnum, vertebrae, and sacrococcygeal region) but have a predilection for the sacrum and clivus. In adults, approximately 50% of chordomas arise in the sacrum, 35% to 40% within the base of skull and clivus, and 10% to 15% throughout the true vertebrae (207; 89; 57; 77; 118). When chordomas affect the vertebral column, more than half will occur in the lumbar region, 25% to 30% in the cervical vertebrae, and 10% to 15% in the thoracic spine (57). In children, chordomas most often involve the skull base (236; 42). On rare occasions, chordomas can arise in extra-osseous or off-the-midline sites such as the transverse process of a vertebra, skin, paranasal sinuses, sella turcica, hypothalamus, or foramen magnum (65; 207; 89; 114; 43; 77; 13).
Chordomas can occur at any age, but are most common between the fourth and sixth decades of life, with a median age of 58.5 years (207; 89; 77; 145). Although these tumors can arise in children, less than 5% of all cases develop before 20 years of age (236; 42; 93). There is a male predominance in some series, especially for tumors of the sacrum, with a ratio ranging from 2:1 to 3:1 (207; 89; 77; 145). In other series, especially chordomas of the skull base, the male to female frequency is equal (77). Although rare, chordomas can be familial (49; 116; 22; 231). The familial cases can often have an earlier onset than the typical patients with sporadic chordomas (231).
Pathologically, several other tumor types must be considered, including ependymoma, schwannoma, neurofibroma, metastasis (eg, clear cell type), chondrosarcoma, and fibrous histiocytoma. Immunohistochemical analysis can be helpful in this differential diagnostic workup (187; 166; 45; 138). The immunohistochemical profile of typical chordomas illustrates the dual epithelial-mesenchymal nature of these tumors and consists of frequent positivity to cytokeratin, epithelial membrane antigen, and HBME-1, and less consistent staining for S100 protein and vimentin (148; 99; 187; 106; 77; 166; 138). Variable staining has also been noted with alpha1-antichymotrypsin, tissue polypeptide antigen, and tau proteins (31; 148; 100). Chondroid chordomas with a small cartilaginous component may have variable staining of S100, with preserved positivity to cytokeratin and epithelial membrane antigen (106). When the cartilaginous component is more robust (20% to 50%), the staining within the chondroid regions for cytokeratin and epithelial membrane antigen may become variable, with persistent positivity for S100 (106). Chondrosarcomas stain consistently negative for cytokeratin and epithelial membrane antigen because there is no epithelial component to these tumors. Chordomas with sarcomatous degeneration have an alteration of the immunohistochemical profile in the malignant regions containing spindle cells (98; 99; 217). The staining for vimentin becomes more prominent, whereas staining for cytokeratin and epithelial membrane antigen is markedly decreased. In some tumors, staining for S100 may also be reduced (217).
On CT and MRI, the differential diagnosis of a mass in the region of the clivus would include other tumors of the skull base such as chondrosarcoma, meningioma, metastases, pituitary adenoma, epidermoid, chondroma, plasmacytoma, craniopharyngioma, schwannoma, histiocytosis X, fibrous dysplasia, cholesterol granuloma, and paraganglioma (150; 123; 62; 216; 66; 138). In general, it is difficult to make a definitive diagnosis of clival chordoma based solely on CT or MRI characteristics (150). The differential diagnosis of tumors of the sacrum or vertebrae includes other destructive or sclerotic lesions such as metastases, Paget disease, plasmacytoma, chronic osteomyelitis, chondrosarcoma, and nerve sheath tumors (184; de Bruïne and Kroon 1988; 90; 152).
Patients with a history and neurologic examination suspicious for a chordoma of the skull base, vertebral column, or sacrum require a radiologic evaluation with either computed tomography or magnetic resonance imaging (184; 126; 50; 167; 209; 150; 123; 130; 77; 102; 235; 66; 138; 25). CT and MRI are equivalent in their ability to delineate the presence of a tumor. Both modalities clearly demonstrate the mass within bone, bone erosion or destruction, and extension into soft tissues (184; 126; 50; 167; 209; 77; 45). Rarely, MRI may have trouble detecting small tumors confined within the margins of the clivus (209). On noncontrast CT, the tumor usually appears as a soft tissue mass, isodense or hyperdense with neural tissues, causing destruction of adjacent bone.
Bone windowed CT scans demonstrate the precise amount of bone destruction caused by the tumor, with sharp margins. Calcification is noted in 40% to 70% of chordomas (especially clival) with CT imaging. Small regions of sequestered bone can also be noted in approximately 15% to 20% of cases. MRI is inferior to CT in the ability to delineate the exact margins of bone destruction or the presence of calcification (126; 167). With the administration of contrast, chordomas always demonstrate contrast enhancement. The amount of enhancement may vary, but is often dense and homogeneous. Sagittal and coronal reconstruction of CT images is sometimes helpful to better delineate the extent of skull base and sacral tumors. However, the ability of CT to evaluate tumors in the sagittal and coronal planes is inferior to MRI.
In general, MRI with sagittal, coronal, and axial sections clearly define the margins of chordomas of the skull base, vertebral column, and sacrum (184; 126; 50; 167; 209; 150; 123; 130; 77; 102; 62; 90; 235; 241; 45; 66).
On T1-weighted images, 75% of tumors appear isointense, whereas 25% appear hypointense, compared to surrounding neural tissues (209; 77). With administration of gadolinium, chordomas usually enhance. As with CT, the degree of enhancement is variable; in most tumors, the pattern is heterogeneous. On T2-weighted images, chordomas are always hyperintense to all surrounding structures. The pattern of hyperintensity is homogeneous is 20% of cases and heterogeneous in the remaining 80% (mild in 30%, marked in 50%) (126; 209; 150). Tumor calcification, exact margins of eroded bone, and presence of sequestered bone are noted in some tumors, but these are not demonstrated as well as with CT. However, the multiplanar capability of MRI allows for better visualization of the extent of tumor margins and infiltration into soft tissues than is possible on CT. Sagittal MRI clearly shows the anterior-posterior margins of tumor involvement. For chordomas of the skull base, MRI is helpful for determining the anterior extension of tumor into the sinuses or nasopharynx and posterior extension towards the brainstem (126; 150).
The posterior fossa is affected by tumor (eg, brainstem compression, cranial nerve displacement) in approximately 80% of skull base tumors (150). Sagittal MRI is essential for tumors of the sacrum to determine the extent of the lesion anteriorly into the rectum and other soft tissues (184; 241). Coronal MRI images are useful for assessing lateral extension of skull base tumors towards the cavernous sinuses. The cavernous sinuses are infiltrated by tumor in approximately 65% of skull base chordomas (150). Furthermore, with sacral chordomas, coronal MRI can determine involvement of sacral nerve roots within the neural foramina. MRI is far superior to CT in demonstrating the relationship of tumor to cranial nerves and vascular structures (126; 167; 209; 150; 152; 66). The carotid and basilar arteries are delineated clearly on T2-weighted images because of the contrast between the flow void inside the vessels and surrounding high-signal tumor. Meyers and colleagues noted displacement of the carotid or basilar arteries in 57% of skull base chordomas (150). Furthermore, in 36% of their cohort, vascular encasement was present. Encasement of vessels by chordomas has also been reported by other authors (126; 167; 209; 235; 152). Universally, the lumen of encased vessels is not compromised by tumor, and they continue to have normal blood flow. MRI angiography can be very helpful to delineate tumor vasculature and the relationship of the mass to encased vessels (66). Most authors do not report any MRI signal characteristics that can be used to differentiate between typical chordomas and those with chondroid regions (150). However, Sze and colleagues noted that chondroid chordomas were less intense on T2-weighted images than were conventional chordomas (209). Assessment of mean quantitative T1 and T2 relaxation values (msec) has shown that chondroid tumors often have shorter times than conventional chordomas.
In addition to neuroimaging, patients with suspected chordomas of the clivus region should undergo pituitary function endocrine testing and neuro-ophthalmologic evaluation. For patients with suspected chordomas of the sacrococcygeal region, urodynamic testing and electromyography may be helpful.
The treatment of many chordomas is limited by the invasive and infiltrative nature of these tumors. The tumor is often too extensive at diagnosis for a complete, curative resection (194; 77; 05). Even when the lesion is small and radical surgery is attempted, local recurrence rates remain high (ie, 50% to 100%). Therefore, the therapeutic approach for chordomas is primarily to maintain local control and minimize regional damage to neural structures. Despite the emphasis on local disease, systemic metastases can occur and are noted in 10% to 30% of patients (85; 165). The most common sites for metastases are the lungs, regional lymph nodes, liver, bone, and skin. Infrequent sites include cardiac muscle, brain, adrenal glands, pancreas, pituitary gland, and eyelids (85; 165). In the majority of patients, recurrence at the local site is most likely to affect morbidity and survival.
Surgical resection. Most authors agree that surgical resection is an important aspect of the initial treatment of patients with chordoma (207; 208; 89; 21; 71; 125; 77; 76; 05; 170; 19; 14; 171; 149; 232). The most aggressive resection possible should be attempted after initial diagnosis, depending on the location (eg, clivus, vertebral body, sacrum) and extent of the tumor. It appears that the aggressiveness of resection has a critical impact on local control rates and may correlate with overall survival. In a review of 51 patients with intracranial chordomas, Forsyth and colleagues found that the extent of resection affected survival (71). In a univariate analysis, the extent of resection was significantly (p = 0.02) associated with survival. For patients receiving only biopsy, the 5- and 10-year survival rates were 36% and 0%, respectively. In the cohort of patients undergoing subtotal resections, the 5- and 10-year survival rates were 55% and 45%, respectively. This effect of resection on survival was most apparent in younger patients.
Chordomas of the skull base and clivus can be grouped according to their size and extension into contiguous areas (05). Type I tumors are small and restricted to 1 compartment of the skull base (eg, clivus or sphenoid sinus). Type II tumors are larger and extend to 2 or more contiguous areas of the skull base. Type III lesions are extensive and involve several contiguous compartments of the skull base (eg, clivus, sphenoid sinus, and middle fossa). Most series type I chordomas are rare and usually amenable to radical resection using a single skull base approach (77; 05). Type II tumors are most common (50% to 65%) and in many cases can also be radically resected using a single skull base procedure. The type III chordomas develop in 10% to 20% of patients and require 2 or more surgical procedures to attempt radical removal.
Numerous surgical approaches and techniques are available for resection of skull base and clivus chordomas (09; 46; 198; 125; 77; 76; 81; 140; 147; 05; 109; 189). The approach will depend on the location of the tumor and the degree of extension from the primary site. Most often, tumors are centered within the lower, middle, or upper clivus and extend into the cavernous sinus or petrous apex. The 4 most common approaches allow for an extensive resection of tumor either extradurally or intradurally. The subtemporal, transcavernous, transpetrous apex approach is used most often (30% to 35%) and provides access to the clivus, cavernous sinus, sella turcica, and petrous apex (77; 147; 05). The extended frontal approach is used in 25% to 30% of patients and is advantageous for tumors with extension into the orbits, ethmoid sinus, and anterior skull base (125). The subtemporal-infratemporal approach is utilized in approximately 20% of patients and offers excellent exposure of the middle fossa, clivus, and lateral skull base. For chordomas of the lower clivus, temporal bone, and occiput, the extreme lateral transcondylar and transjugular approach is used (approximately 15% of patients). Uncommon surgical approaches include the transoral, transmaxillary, transcervical-transclival, anterior cervical, and transsphenoidal procedures (09; 46; 198; 77; 76; 81; 140; 147; 109; 216).
Studies using the most advanced skull base approaches for removal of chordomas report various results. Radical or total resection of tumor is achieved in 43.5% to 55% of patients, near total or subtotal resection is noted in 40% to 47% of patients, and partial resection is attained in 8% to 10% of patients (125; 77; 76; 05; 189; 232). In a series of patients with chordomas and chondrosarcomas involving the skull base and cavernous sinus, after a median follow-up of 24 months, Lanzino and colleagues noted 3 recurrences in 14 patients with subtotal or partial removal of tumor (125). No recurrences were observed in the group of patients that had undergone radical resections. In a study of skull base chordomas by Gay and colleagues, there was a statistically significant difference (p less than 0.05) between the risk of recurrence in patients with radical or near-total resections and patients with subtotal or partial resections (76). The overall recurrence-free survival estimates were 80% at 3 years and 76% at 5 years. In contrast, the survival estimates for patients that had recurrence of disease were 52% at 2 years and 26% at 3 years. Previous surgery or radiation therapy was associated with an increased risk of recurrence and surgical complications. In the series reported by Samii and colleagues, the transethmoidal approach was used most often (36.3%), with gross total removal of tumor in 49.4% and subtotal removal in 50.6% of cases (189). The 5- and 10-year survival rates were 65% and 39%, respectively. Using a historically controlled study design, Di Maio and colleagues reviewed their experience in surgical resection for cranial base chordomas, comparing results using older techniques from 1988 to 1999 to the results from 2000 to 2011 using more advanced skull base approaches (60). The mean 5-year overall survival and recurrence-free survival for the entire cohort (N = 95) was 74% and 56%, respectively. Complete resection rates were similar between groups (68% vs. 74%). However, the modern era had reduced complication rates overall, as well as for cranial nerves and vascular and systemic issues. Although there was no difference in the 5-year recurrence-free survival rate, there was a higher overall survival rate in the more modern cohort (93% vs. 64%, p = 0.012).
Some surgeons are now attempting more "minimally invasive" approaches, using endoscopic techniques for removal of clival chordomas (52; 186). In the series of 12 patients reported by Dehdashti and colleagues, endoscopy was used during an expanded endonasal approach, with gross total removal of tumor in 7 patients (58%) and subtotal removal (more than 80%) in another 5 patients (42%). No tumor recurrence was noted with a median follow-up of 16 months. The technique was well tolerated. Similar results have been reported in a series of patients with clival chordomas in a study by Saito and colleagues (186). Another "minimally invasive" approach is the expanded transsphenoidal approach, which has been used, in combination with neuro-navigation, by Al-Mefty and co-workers in a series of 38 clivus chordomas (06). With this technique, a gross-total resection was achieved in 79% of the cases.
Sacral chordomas are often extremely large at diagnosis; however, most authors advise radical resection whenever possible (207; 89; 21; 190; 170; 228; 241; 242; 19; 197; 75). Similar to the experience with skull base tumors, recurrence-free survival is improved after radical or near total resection. For tumors of the lower sacrum and coccygeal region, many authors recommend a posterior approach (89; 170; 228). Other investigators argue that a combined anterior-posterior approach is preferable (21). Tumors of the upper sacrum are resected most efficiently with a staged, combined anterior-posterior approach (207; 89; 21; 190; 170). Some authors also suggest using transcatheter arterial embolization of the main arteries feeding the tumor, prior to attempts at radical resection (238). They report that pre-operative embolization can lead to less intra-operative blood loss, a better view of the surgical field, and facilitation of maximal removal of the sacral chordoma. Regardless of the approach used to resect the tumor, it is important to attempt preservation of the upper sacral nerve roots and the pudendal nerve. If the bilateral S2 nerve roots are sectioned during surgery, urogenital and rectal function will be lost or impaired. If both S nerve roots are preserved, 50% of patients will retain at least partial bladder and bowel control (190; 170). To maintain normal bowel continence, preservation of at least 1 set of ipsilateral S1, S2, and S3 nerve roots is recommended. The local recurrence rates are approximately 25% to 30% for tumors removed en bloc by radical resection (207; 89; 21; 242). If the tumor is removed by subtotal or partial resection, local recurrence rates increase to approximately 60% to 65%.
It is also recommended that chordomas of the true vertebrae be radically resected whenever feasible (208; 89; 57; 90; 19; 23; 197; 41). For tumors of the cervical vertebrae, most authors recommend an anterior approach to perform a corporectomy, followed by bone grafting, if necessary (57; 15). For carefully selected patients, a 2-stage, en bloc resection of the tumor may also be considered (176; 41). Although an en bloc resection of cervical tumors is the preferred approach, some authors don't feel it is possible in most cases (15). Thoracic tumors are best approached by thoracotomy or a staged procedure that combines a laminectomy and thoracotomy (89). Lumbar chordomas will usually require an anterior approach; on occasion, a posterolateral approach may be necessary (89; 90; 23).
Radiation therapy. Although radical resection is considered in each patient with a chordoma of the skull base, vertebrae, or sacrum, it is often impossible due to the invasive nature of these tumors. Therefore, radiation therapy to eradicate residual or recurrent disease is an important therapeutic consideration in many patients (207; 89; 71; 194; 77; 242; 14; 149). Unfortunately, chordomas have proved to be relatively radioresistant tumors. The clinical results in most radiation therapy trials of chordomas have demonstrated only modest improvements in local control, recurrence-free survival, and overall survival (20; 11; 12; 115; 10; 210; 214).
Early reports in the radiation oncology literature using photon-based megavoltage therapy suggested a dose-response relationship for chordoma (210; 214). It was recommended that patients receive at least 6000 to 7000 cGy to the tumor bed for optimal response. However, other studies have been unable to document a consistent dose-response relationship for chordoma using conventional photon techniques (191; 48; 77; 210). In the reports by Cummings and colleagues and by Saxton, doses of 2500 to 7000 cGy were used for patients with chordomas of various sites after surgical resection (191; 48). Palliation of symptoms and improvement of relapse-free survival was as likely to occur with doses of 4000 to 5500 cGy as with higher doses. In an extensive review of reported dose-response data for photon techniques in treatment of cranial chordoma, Tai and colleagues concluded that no dose-response relationship was evident (210). Administration of doses in the range of 4500 to 5500 cGy was as effective as higher doses. In addition, the authors state that surgical resection in combination with irradiation significantly prolongs survival when compared to either modality used alone. Similar results are reported by York and colleagues, who studied 27 patients with sacral chordomas and noted a significant difference in disease-free interval (24 months vs. 8 months; p less than 0.02) for patients receiving irradiation after subtotal resection (242). In a study of 21 patients with chordomas of various sites, Keisch and colleagues concluded that irradiation prolonged the time to first relapse for tumors of the lower spine and sacrum but not for tumors of the skull base (115). The overall 5- and 10-year actuarial survival rates were 74% and 46%, respectively. Forsyth and colleagues evaluated the results of 51 patients with intracranial chordomas and determined that conventional irradiation did not affect the overall survival of the cohort, but did prolong the disease-free survival, especially in younger patients (71). The 5- and 10-year disease-free survival in irradiated patients was 39% and 31%, respectively. A similar improvement of progression-free survival, without a change in overall survival, has been reported by Thieblemont and colleagues in 26 patients with chordomas of various sites (214).
Irradiation of chordomas with charged particles (ie, protons, helium, neon, carbon) has shown promise as a more efficacious therapeutic option (20; 11; 12; 194; 77; 210; 101; 213; 163; 195; 104; 149; 161; 105). There are several radiobiological advantages of charged particles over photons. The high linear energy transfer of charged particles allows for a more defined and superior dose distribution (ie, steeper fall-off in dose). Higher doses can be prescribed to the tumor volume with minimal risk of augmented toxicity to surrounding normal structures.
Austin-Seymour and colleagues have used fractionated proton irradiation for skull base chordomas and chondrosarcomas, administering a mean total dose of 69 GyE (11; 12; 210). The 5- and 10-year local control rates were 82% and 58%, respectively, whereas the 5- and 10-year disease-free survival rates were 76% and 53%, respectively. The median time to local failure was 53 months. In their opinion, these results represent a significant improvement over the results of conventional radiation techniques.
Similar results have been reported by Hug and colleagues in a review of 33 patients with skull base chordomas (101). Using a mean target dose of 70.7 Cobalt Equivalent, they achieved a local control rate of 76% and an actuarial 5-year survival rate of 79%.
A meta-analysis of the literature has reviewed 210 articles and analyzed a total of 416 patients who had received proton therapy for skull base chordoma (07). These data were tabulated and then compared to the results from other irradiation techniques. Although no randomized phase III studies were available, the remaining studies did suggest that the use of protons allows for the use of higher doses (up to and above 70 CGE) and more precise tumor targeting. Local control rates at 3 and 5 years were in the range of 67.4% to 87.5% and 46% to 74%, respectively. The estimated overall survival rates at 5 years were 66.7% and 80.5%, and at 10 years were 54%. These results were superior to those obtained with conventional irradiation approaches, with acceptable toxicity. A more recent meta-analysis of proton beam radiotherapy for postsurgical treatment of skull base chordomas reviewed the data from 76 articles (143). The analysis provided Level IIb/III evidence that proton beam therapy demonstrated improved long-term local control and survival in comparison to other radiotherapy modalities. In a study using helium and neon particles, Berson and colleagues treated 25 patients with chordomas of the skull base and cervical spine and reported a 5-year local control rate of 55% (20; 210). In a review of 14 patients with sacral chordomas treated with charged helium and neon particles, Schoenthaler and colleagues report a 5-year local control rate of 55% (194). The 5- and 10-year survival rates were 85% and 22%, respectively, with an overall median survival of 77 months. Several papers from Noel and colleagues describe their experience in treating 100 patients with skull base and spinal chordomas using a combination of photon (median dose 45 Gy) and proton (median dose 22 GCE) conformal irradiation (163; 162). The 2- and 4-year local tumor control rates were 86.3% and 53.8%, respectively. Overall survival rates at 2 and 5 years were 94.3% and 80.5%, respectively. Photon and proton combination irradiation was tolerated fairly well, with acceptable toxicity. Similar results have been noted in a series of 17 patients with primary and recurrent sacral chordomas by Park and colleagues (172). Local control rates and progression-free survival were significantly worse in the patients with recurrent disease. A report of the experience from the University of Florida Proton Therapy Institute also suggests excellent results (55). From 2007 to 2011, 33 patients with skull base chordomas received proton therapy; 79% were male, with 27% having had a gross total resection. The tumor bed received a total dose between 77.4 and 79.4 CGE. At a median follow-up of 21 months, the local control and overall survival rates at 2 years were 86% and 92%, respectively. Grade 2 toxicity in the form of unilateral hearing loss was noted in 18% of the cohort (partially corrected with a hearing aid). In a report focusing on patients with spine chordomas (N = 126), Rotondo and colleagues used high-dose proton-based radiotherapy, with a mean dose of 72.4 Gy (185). At median follow-up of 41 months, the 5-year overall survival rate, local control rate, locoregional control rate, and distant control rate for the entire cohort was 81%, 62%, 60%, and 77%, respectively. Local control was significantly improved with en bloc resection versus intralesional resection.
Results from studies with carbon ion radiotherapy have been reported by several groups and are similar to those with protons. In a series of 24 patients with skull base chordomas treated with carbon ions, the 2-year local tumor control rate was 90% (195). In a similar study of 32 patients with skull base chordomas, carbon ion radiotherapy was utilized after aggressive resection (211). The 3-year recurrence-free survival rate was 70%, as opposed to 57.1% for patients receiving more conventional radiotherapy (p= 0.001). For patients with unresectable sacral chordomas, carbon ion radiotherapy has also been effective (104; 105; 103). The most recent summary of their series of 188 patients demonstrated that most patients received a dose of 67.2 to 70.4 Gray equivalents in 16 fractions, with 5-year local control, overall survival, and disease-free survival rates of 77.2%, 81.1%, and 50.3%, respectively (103). A similar study used carbon ion radiotherapy on a series of 23 patients with sacral chordoma who had not undergone surgical resection, with a median follow-up of 46 months (199). Initially after treatment, T2-weighted MR imaging revealed mild tumor progression in 13 of the 23 cases, with regression in 6 cases. After long-term follow-up, imaging showed a reduction in tumor volume in 20 of 23 cases. A large study from a German group used a Raster scan technique to evaluate 155 patients with skull base chordoma after surgery and irradiation with carbon ions from 1998 to 2008 (221). The median age of the cohort was 48 years old (range 15 to 85 years). The median total dose was 60 gray, with a median boost planning target volume of 70 mL. The 3-year, 5-year, and 10-year local control rates were 82%, 72%, and 54%, respectively. The 3-year, 5-year, and 10-year overall survival rates were 95%, 85%, and 75%, respectively. No late toxicity from carbon ion therapy was noted in the cohort. Another report from the same German group applied carbon ion beam therapy to a series of 56 patients with sacral chordomas, either alone or in combination with photon IMRT (222). The median total dose was 66 Gy (range 60 to 74 Gy). The 2- and 3-year local control rates (median follow-up time of 25 months) were 76% and 53%, respectively, with an overall survival rate of 100%.
Other methods of irradiation of chordoma include brachytherapy with radioactive seeds (eg, iodine-125), radiosurgery, intensity-modulated radiotherapy, conformal techniques, intraoperative radiotherapy for sacral tumors, and radiation sensitizers (124; 11; 120; 77; 154; 121; 215; 108; 177; 112). It is difficult to evaluate the efficacy of these modalities because of the small number of patients that have been treated and the limited follow-up intervals reported (11; 77).
In a series of 25 patients with cranial base chordoma, Krishnan and colleagues used radiosurgery (median dose 15 Gy) as an adjunct to surgical resection (122). The 2- and 5-year overall actuarial tumor control rates were 89% and 32%, respectively. The 5-year freedom from progression rate was 67%.
In a report of 7 patients with skull base and cervical chordoma, Gwak and colleagues describe their initial experience using a hypofractionated radiosurgical approach after surgery, with doses of 21 to 43.6 Gy in 3 to 5 fractions (84). All patients except for 1 remained stable through 27 months of follow-up.
Razoxane is a piperazine derivative that functions as a radiation sensitizer and has been used in a series of patients with chordoma and chondrosarcoma (177). Preliminary results suggest that this form or irradiation is well tolerated and may extend survival.
A review of experience using Gamma Knife radiosurgery after tumor resection of skull base chordomas suggests efficacy in patients with low residual tumor volumes (88). With marginal doses of at least 15 Gy, the actuarial 5- and 10-year survival rates were 80% and 53%, respectively. The 5- and 10-year actuarial local tumor control rates were 76% and 67%, respectively. The local control rates were significantly improved when residual tumor volume was 20 mL or less (p = 0.018). Another review of Gamma Knife treatment of residual skull base chordomas noted survival of 90.9% and 75.8% after 3 and 5 years, respectively (133). The actuarial tumor control rate was 64.2% and 21.4% after 3 and 5 years, respectively. An update demonstrates a similar experience using Gamma Knife in the setting of recurrent disease in the clivus after an initial aggressive resection (107). A report by the North American Gamma Knife Consortium analyzed data from 6 centers, for a total of 71 patients with skull base chordomas who had undergone treatment (113). The median patient age was 45 years, with a median dose of 15.0 Gy (range, 9 to 25 Gy) and median target volume of 7.1 cm3 (range, 0.9 to 109 cm3). After a median follow-up of 5 years, there were 23 patient deaths. The 5-year actuarial overall survival was 80%: 93% for the cohort that had not received prior irradiation (N = 50) and 43% for the cohort that had received prior radiotherapy. The overall 5-year treated tumor control rate was 66%.
CyberKnife radiosurgery has also been applied to chordomas of the skull base, spine, and sacrum in a series of 18 patients (92). The mean tumor volume was 128.0 mL, with a median dose of 3 Gy over 5 treatment sessions. Two patients had partial responses, 9 had stable disease, and 7 progressive disease (median 10 months after treatment). The local control rate at 65 months follow-up was 59.1%, with an overall survival of 74.3%. Although there may be a role for brachytherapy, radiosurgery, intensity-modulated radiotherapy, conformal techniques, and radiation sensitizers in palliation of residual and recurrent disease in carefully selected patients, further studies are needed.
A new form of treatment involves high-intensity focused ultrasonic ablation of tissue. This technique has been applied to 4 patients with advanced sacral chordomas (79). Patients were treated under general anesthesia or sedation and had tumor regression noted in 3 cases on follow-up imaging. Tumor necrosis was also noted in several patients.
Chemotherapy. The role of chemotherapy in the treatment of chordomas remains limited (208; 89; 70; 77; 214; 242). The main indication for chemotherapy has been in patients with recurrent or widespread disease not amenable to further surgery or radiation therapy. In most cases, the regimens have been designed to resemble protocols used for soft-tissue sarcomas. Unfortunately, few patients have responded to this approach. Chemotherapeutic agents that have been used (as single agents or in combination) without success include methotrexate, vincristine, cisplatin, doxorubicin, etoposide, actinomycin D, and cyclophosphamide (208; 89; 70; 214).
Fleming and colleagues report 2 patients with malignant sacral chordomas and lung metastases in whom chemotherapy produced objective responses (70). One patient responded to a multiagent regimen consisting of etoposide, cisplatin, vincristine, dacarbazine, cyclophosphamide, and doxorubicin administered intravenously over 3 days every 4 to 5 weeks. The second patient responded to this multiagent regimen for 5 cycles and then progressed. A second regimen of single-agent continuous infusion ifosfamide was initiated, which produced dramatic shrinkage of the lung lesion.
A novel approach using a conventional chemotherapy drug has been reported by Guiu and colleagues in a case report of a 46-year-old male with recurrent cervical chordoma and spinal cord compression (82). They used a direct infusion of intratumoral carboplatin in combination with epinephrine (3 to 5 mL) over 5 minutes and through 3 separate catheters. Treatment was performed monthly, intermittently, during an 18-month period. There was a 42% reduction in tumor volume, with regression of spinal cord compression.
The report of Casali and colleagues would suggest that molecular approaches to chemotherapy of chordomas should be investigated further (35; 61). A series of 6 tumors were noted to express platelet-derived growth factor receptors, prompting an attempt at therapeutic intervention with imatinib mesylate, a small molecule tyrosine kinase inhibitor of platelet-derived growth factor receptors. Overall, the cohort responded with stabilization of disease, several of which were long-term. In addition, 4 of 5 symptomatic patients had improvement of symptoms. A follow-up study by an Italian-Swiss group treated 56 patients with advanced chordoma using imatinib in a phase II clinical trial (202). There was only 1 objective response (partial response, overall response rate of 2%), whereas 70% of the cohort demonstrated stable disease. There were also minor responses (< 20% tumor reduction) noted in 9 patients. The median progression-free survival for the entire cohort was 9 months. In a series of 10 patients with progressive chordoma, imatinib (400 mg/day) was combined with rapamycin (sirolimus 2 mg/day), an mTOR inhibitor (203). There were 7 patients with objective tumor shrinkage on follow-up MRI scan, as well as 7 patients who demonstrated a metabolic response on PET imaging. Responses and stabilization of disease were durable for more than 6 months in 89% of the cohort. In a related paper, cell lines were established from a recurrent clival chordoma and used to establish tumor xenografts in severe combined immunodeficiency (SCID) mice (181). Molecular analysis of the tumor cell line revealed persistent expression of brachyury, activation of the mTOR pathway, and mutation of the K-RAS gene. Treatment of the cell lines and xenografted tumors with rapamycin resulted in significant growth inhibition. When rapamycin was then used for the patient with the recurrent clival chordoma, there was a 6-fold reduction of tumor growth that was durable for 10 months.
A patient has been reported with a sacral chordoma that was shown to express epidermal growth factor receptor, in whom a molecular chemotherapy approach was also successful (94). The patient had local progression and enlargement of pulmonary metastases, prompting placement on a combination of cetuximab and gefitinib. The sacral tumor and pulmonary metastases responded with partial responses that have been maintained for greater than 9 months.
Another patient treated with cetuximab and gefitinib has been reported. The patient had a progressive chordoma of the cervical spine with spinal cord compression (131). After combination treatment for 4 months, follow-up MRI revealed shrinkage of the tumor and improvement of the spinal cord compression. In addition, the patient had some objective neurologic improvement. Singhal and colleagues report a patient with a sacral chordoma that had progressed through multiple surgeries, irradiation, imatinib, and CYT997, a vascular disrupting agent and finally responded to erlotinib, an EGFR tyrosine kinase inhibitor (201). The patient took 150 mg/day, and after 3 months had objective tumor shrinkage on follow-up MR imaging. Molecular analysis of the tumor did not reveal any EGFR mutations or gene amplification. According to the report after 11 months, the patient remained stable and his MRIs continued to show a 30% reduction in tumor size. Another patient with an advanced chordoma responsive to erlotinib has been reported by Launay and colleagues (128). The patient, a 76-year-old male with an EGFR overexpressing tumor, initially progressed on imatinib and then responded to erlotinib for 12 months. An Italian phase 2 study used lapatinib, a tyrosine kinase inhibitor that blocks HER2/neu and EGFR, in a series of 18 EGFR-positive progressive chordomas (204). All patients received 1500 mg/day of lapatinib until progression or toxicity. There were 6 partial response (33%) and 7 stable disease (38.9%) patients, with a median progression-free survival of 8 months using response evaluation criteria in solid tumor (RECIST) criteria. A similar patient with a spinal chordoma was on a clinical trial of erlotinib and linsitinib (IGF-1R/insulin receptor inhibitor), and had a partial response after 18 months of treatment (03). The patient was able to maintain the tumor shrinkage for another 43 months--thereby achieving a durable response for 5 years. Afatinib is another EGFR inhibitor that was tested against a panel of chordoma cell lines by Magnachi and colleagues (139). It was the only anti-EGFR drug to have inhibitory activity against every cell line in the panel. When the molecular mechanisms were analyzed in more detail, the antiproliferative IC50s correlated with the ability of the drug to promote degradation of EGFR and Brachyury. In addition, high EGFR phosphorylation correlated with higher sensitivity to afatinib. This data supports the upcoming European phase 2 clinical trial of afatinib in advanced chordoma patients.
Several authors have noted advances in the molecular biology of chordoma and the dysregulation of several key receptor tyrosine kinase signaling pathways, including EGFR, PDGFR, and c-Met, along with downstream elements such as Akt and mTOR (16; 58; 61; 75). They are consistent in their recommendation that these pathways represent excellent therapeutic targets and that appropriate treatment should involve the rational use of targeted molecular drugs against these targets. In addition, the availability of targeted drugs for these pathways should allow for the design of high-quality clinical trials. Using chordoma cell lines with overactivity of the cell cycle (via loss of CDKN2A and p16, and activation of the CDK4/6 and Rb pathways), von Witzleben and colleagues investigated the efficacy of palbociclib, a CDK4/6-specific inhibitor (227). Treatment with palbociclib was able to significantly inhibit tumor cell growth in vitro. Another pathway that appears to be active in chordoma cells is the Signal Transducer and Activator (STAT) pathway that is involved in the regulation of cellular proliferation and apoptosis. Wang and colleagues used FLLL32, a STAT3 small molecule inhibitor, on sacral and clival chordoma cell cultures (230). FLLL32 induced significant cytotoxicity in both types of chordoma cell lines, eliminating most of the viable cells. The cytotoxicity correlated with the degree of downregulation of phosphorylated STAT3 and involved an increase in apoptosis and a reduction in cellular proliferation through inhibition of mitosis.
As noted above and consistent with the report by von Witzleben and colleagues, Liu and associates analyzed the expression of CDK4 in cell lines and clinical specimens, and found it to be highly expressed (134). Based on the high levels of expression, they treated chordoma cell lines with the CDK4 inhibitor, palbociclib, which demonstrated significant inhibition of cell growth and proliferation via arresting the cell cycle in the G1 phase.
The vascular endothelial growth factor (VEGF) pathway is also being explored as an avenue of treatment. Lipplaa and co-workers treated 5 patients with unresectable or metastatic chordoma with sunitinib and pazobanib--VEGF receptor inhibitors (132). Two of 4 patients on pazopanib had stable disease for 14 and 15 months, respectively, whereas the patient treated with sunitinib had a partial response that lasted for 27 months. Ribeiro and colleagues reported the case of a patient with a multifocal recurrence of a lumbar spine chordoma that had further progression after re-resection and a course of radiotherapy (180). He was then placed on pazopanib (800 mg/day), and had an objective response, with a 23.1% reduction in size of the recurrent lesions, which was durable.
In a series of 15 patients with advanced chordoma, Chugh and colleagues used 9-nitro-camptothecin (9-NC; 1.5 mg/m2 per day for 5 days), an oral inhibitor of DNA topoisomerase I (40). There was 1 objective response, with an overall median time to progression of 9.9 weeks and a 6-month progression-free survival rate of 33%. Limiting toxicity included anemia, leukopenia, fatigue, nausea, and diarrhea.
The immune checkpoint pathways and associated immune checkpoint inhibitor drugs (eg, ipilimumab, nivolumab, pembrolizumab) are becoming very important for systemic therapy of many solid tumors (26; 01). Work by Mathios and colleagues has shown that chordomas express PD-1, and that expression of PD-1 is further induced by proinflammatory cytokines in the tumor environment (142). Ambient expression of PD-L1 is not very high in chordoma tissue but is present on tumor-infiltrating macrophages and lymphocytes. Based on this early data, Fujii and colleagues used the anti-PD-L1 IgG1 monoclonal antibody, avelumab, on a series of chordoma cell lines (73). Avelumab is capable of inducing antibody-dependent cell-mediated cytotoxicity (ADCC) and was able to induce ADCC of chordoma cells when incubated with brachyury-specific CD8+T cells.
The only risk associated with chordomas and pregnancy would be radiation to the sacral region in a pregnant patient or the possible loss of fertility related to surgery or radiation or both for tumors in the region of the pituitary. Pregnancy does not affect the clinical behavior of chordomas of the clivus, sacrum, or true vertebrae.
If the transoral or transmaxillary surgical approaches are used, the airway needs to be protected, and a tracheostomy may be necessary. In patients with skull base chordomas large enough to elevate intracranial pressure, agents that produce excessive sedation and ventilatory depression should be avoided because these could exacerbate intracranial pressure (127; 47). Hypotonic fluids should also be avoided whenever possible. During the induction and maintenance of anesthesia, agents that minimize hypertension, cerebral vasodilation and blood flow, cerebral metabolic rate, chest wall rigidity, and hypercapnia should be chosen.
Herbert B Newton MD
Dr. Newton, Director of the Neuro-Oncology Center at Advent Health Cancer Institute Orlando, has no relevant financial relationships to disclose.See Profile
Rimas V Lukas MD
Dr. Lukas of Northwestern University Feinberg School of Medicine received honorariums from Novocure for speaking engagements, honorariums from Novocure for advisory board membership, and research support from BMS.See Profile
Nearly 3,000 illustrations, including video clips of neurologic disorders.
Every article is reviewed by our esteemed Editorial Board for accuracy and currency.
Full spectrum of neurology in 1,200 comprehensive articles.
Stroke & Vascular Disorders
In this article, the author reviews current knowledge about intracranial aneurysms due to infectious and neoplastic causes. Direct mural injury or invasion
Jul. 21, 2021
This article reviews the diverse systemic manifestations of von Hippel-Lindau disease and discusses the historical context of the disease as well as
Jul. 09, 2021
Jul. 06, 2021
Jun. 22, 2021
Paraneoplastic limbic encephalitis is a subset of a larger group of autoimmune encephalitides characterized by the predominant involvement of the limbic system. Patients present with varied combinations of short-term memory loss, seizures, and psychiatric disturbance. Small cell lung carcinoma, thymoma, testicular germ cell tumors, and Hodgkin disease are the tumors most often associated with paraneoplastic limbic encephalitis.
Jun. 01, 2021
Hemangioblastoma is a relatively uncommon tumor that most often arises in the cerebellum, brainstem, or spinal cord. Hemangioblastomas usually occur
May. 25, 2021
Headache & Pain
Apr. 18, 2021
Dysembryoplastic neuroepithelial tumors are rare, indolent, low-grade tumors found in children and young adults. Most commonly affecting the temporal
Mar. 22, 2021