Historical Background

Virchow first described small nodules along the clivus in 1846 and named them “ecchondrosis physaliphora” in 1857 because he hypothesized them to be of cartilaginous origin.¹ The following year, Luschka, Henker, and Hasse described similar nodules. Concurrently, Muller postulated that ecchondrosis physaliphora was of notochordal rather than cartilaginous origin. In 1894, Ribbert coined the term “chordoma” and presented a case that supported Muller’s hypothesis. A year later, he experimentally induced chordomas in rabbits by perforating the anterior intervertebral ligament.^1,2 In 1909, Harvey Cushing reported successful removal of a chordoma (initially described as a teratoma, case XVII) in a 35-year-old man (although the patient died during reoperation some 6 months later).^3,4 Today, chordomas present the same challenges to neurosurgeons that they did 100 years ago; despite the remarkable advances made in operative and imaging techniques, the individual surgeon must often choose between the chance for cure and reduction of morbidity to achieve the greatest quality of life.

Natural History

Chordomas are rare, slow-growing tumors that are believed to arise from rests of vestigial notochordal cells. They represent only 1% to 4% of primary bone tumors,⁵ with an age-adjusted incidence rate of 0.08 per 100,000 patient-years. There is a twofold greater incidence in males and a fivefold greater incidence in the white population.⁶ The peak incidence occurs in the eighth decade of life, and development of chordomas is quite rare before the age of 40. One recent, multi-institutional review suggested that anatomic location is evenly divided among the spheno-occipital region (32%), mobile spine (32.8%), and sacrum (29.2%).⁶ Older, often-cited reports, based on single-institution data, estimated that sacral chordomas constituted 50% of cases whereas 35% and 15% were cranial and spinal chordomas, respectively.^5,7 Irrespective of the statistical distribution along the craniospinal axis, the anatomic diversity of skull base chordomas reflects the complex, serpentine course of the notochord within the clivus (Fig. 140-1). The clinical findings and symptomatology depend on the anatomic location of the tumor. Patients with cranial chordomas often have headaches, neck pain, cranial neuropathies, and even endocrine abnormalities, whereas those with spinal chordomas have back pain, pathologic fractures, myelopathy, and radiculopathy. Sacral chordomas tend to be quite large at the time of diagnosis because of the insidious onset of symptoms such as pain and bowel or bladder dysfunction.⁸

FIGURE 140-1 Development and differentiation of the notochord. A, Early development of the notochord orients the embryo along the cephalocaudal axis. The hypoblast thickens at what will become the cranial end and forms the prochordal plate. Subsequently, migration of mesodermal cells into the primitive pit forms the notochordal process. By means of induction, the notochord causes the overlying ectoderm to differentiate into the neural plate and thereby also establishes the dorsoventral axis. The notochord is bifurcated at its terminal ends. This bifurcation may reflect a vestige of the migratory process and may also play a role in the ultimate distribution of chordomas. B, Around the 20th day of fetal life, somites begin to appear. Subsequently, the vertebral column develops as a result of further migration and differentiation of mesenchymal cells through membranous, chondrification, and ossification phases into ventromedial sclerotomes. After resegmentation, the sclerotomes then fuse and form the vertebral bodies. The notochord begins to involute. C, As the spinal column develops, the notochord continues to involute and eventually forms the nucleus pulposus of the intervertebral disks along the midline of the spinal column. The physaliphorous cells of the notochord contain a hydrated protein matrix that allows the cells to swell and exert pressure on a thick surrounding basement membrane. Thus, the notochord provides structural support to the developing embryo and later acts as a force buffer in the adult spine. D-F, Notochordal remnants. These images demonstrate the distribution of notochordal remnants. D, Clival ecchondroses are illustrated in schematic form. These small rests of cells appear to be a normal anatomic variant resulting from abnormal migration and failure of notochordal cells to involute. They may, however, constitute the origin of clival chordomas. E, Both normal nucleus pulposus and vertebral body remnants are illustrated. F, Schematic distribution of sacral notochordal remnants. It is unknown whether chordomas arise from fetal notochordal rests or benign notochordal tumor, a recently recognized entity that is found within the clivus (11.5% of autopsies) and vertebral bodies (14% of autopsies).

Overall, 5- and 10-year survival rates are 67.6% and 39.9%, respectively.⁶ Local recurrence is the most important predictor of mortality in patients with sacral chordomas. Distant metastases to the lung, liver, and bone can occur with chordomas, but they generally take place quite late in the disease process and, as a result, probably contribute little to overall mortality.

The notochord is an evolutionarily conserved structure (hence, the phylum Chordata) that arises early in embryonic development from the dorsal organizer region. It plays a critical role in many aspects of development, such as determining the dorsal-ventral axis of the neural tube (through secretion of the sonic hedgehog homologue [SHH] gene product), left-right asymmetries, and arterial-versus-venous fates of early blood vessels. The notochord is made up of highly vacuolated cells (the physaliphorous cells) that contain a hydrated protein matrix and are surrounded by a thick basement membrane. The protein matrix absorbs water, which allows the cells to swell and exert pressure on the basement membrane. In this fashion, the notochord provides structural support to the developing embryo. In animals with a bony spine, the notochord involutes as the adjacent somites form the vertebral bodies. Some notochordal cells persist and form the nucleus pulposus of the intervertebral disks, whereas others survive as “rests” within the clivus, sacrum, and vertebrae.^9,10 The relationship between these cellular rests and the development of later chordomas is complex. Along the dorsal clivus, for example, approximately 2% of the population, based on autopsies (Ribbert’s theory, see Bailey and Bagdasar¹), have benign growths termed ecchordosis physaliphora (i.e., the ecchondrosis of Virchow). Rarely, they can grow and become symptomatic giant ecchordoses or intradural benign chordomas.^11,12 Similarly, benign notochordal tumor is a recently recognized entity found within the clivus (11.5% of autopsies) and vertebral bodies (14% of autopsies) that can expand and become a classic chordoma.^13,14 These tumors are immunohistochemically distinct from fetal notochord rests.¹³ However, whether all classic chordomas arise from benign tumor precursors is unknown.

Radiography

Chordomas and chondrosarcomas are indistinguishable from one another on radiographic examination, except for the fact that chondrosarcomas tend to arise from a paramedian location and chordomas are nearly always found along or near the midline.^15–18 One caveat in this regard is that chordomas can distort their midline origin by becoming quite large before they are detected and growing asymmetrically in one direction or another. This asymmetric growth may extend into the nasopharynx, orbits, or sella or along the cranial nerves, with dorsal deflection of the brainstem. Although chondrosarcomas typically arise from the skull base, they can also arise from the meninges. On non–contrast-enhanced computed tomography (CT), both chordomas and chondrosarcomas are slightly hyperdense masses with variable amounts of calcification (Fig. 140-2). Bone windows often demonstrate significant bony erosion. For chordomas, the erosion is predominantly along the clivus and petrous apex. Tumors of both types show variable enhancement on contrast-enhanced CT. Chordomas and chondrosarcomas are both isointense to hypointense on T1-weighted magnetic resonance imaging (MRI), again with variable enhancement. On T2-weighted MRI these tumors are hyperintense, but they may have some heterogeneity in signal intensity because of calcifications and bony sequestrations (Fig. 140-3). Benign ecchordosis physaliphora is radiographically similar but does not show enhancement.^19,20

FIGURE 140-2 Non–contrast-enhanced axial CT image of a chondrosarcoma within the left infratemporal fossa. Chordomas and chondrosarcomas are typically slightly hyperdense masses with a variable amount of calcification. Bone windows often demonstrate significant bony erosion. For chordomas, the erosion is predominantly along the clivus and petrous apex.

FIGURE 140-3 Large clival chordoma demonstrated on T1-weighted (left) and T2-weighted (right) axial magnetic resonance imaging (MRI). Chordomas and chondrosarcomas both appear isointense to hypointense on T1-weighted MRI and display variable enhancement with the administration of intravenous contrast material. These tumors appear hyperintense on T2-weighted images, although they may have some heterogeneity in signal intensity because of calcifications and bony sequestrations.

Pathology

Macroscopically, chordomas and chondrosarcomas are nodular, exophytic masses with a smooth whitish or occasionally reddish appearance.^1,21 They often display a fibrous outer layer that is contiguous with the fibrous septations found within the tumor. This pattern of growth gives the tumor a lobulated appearance in cross section. These tumors lack a thick capsule. Internally, the tumors are soft, gray, and translucent but generally lack significant areas of necrosis. In addition to their local bony destruction, these tumors will extend out into adjacent cancellous bone, thus making oncologic excision difficult if not impossible. Microscopically, chordomas can be divided into three main histochemical types: conventional, chondroid, and dedifferentiated. Conventional chordomas are composed of sheets or cords of cells with eosinophilic cytoplasm embedded in a mucinous matrix. The appearance of these cells can range from having a homogeneous cytoplasm completely lacking in vacuoles to a cytoplasm markedly engorged with vacuoles, the so-called physaliphorous or bubble-bearing cells (Fig. 140-4). Differentiation into cartilaginous or, occasionally, osseous-appearing cells may occur focally or throughout the tumor and, if extensive, distinguishes the chondroid chordoma subtype from conventional chordomas. Rarely, chordomas undergo malignant transformation, which gives rise to dedifferentiated chordomas. In this subtype there is a predominant sarcomatous appearance with malignant spindle cells, increased atypia, and increased mitotic index. Immunohistochemically, chordomas are reactive for epithelial membrane antigen, cytokeratins, α-fetoprotein, and rarely, S-100.

FIGURE 140-4 Histopathology of chordoma as shown on a low-power photomicrograph of a conventional chordoma. Chordomas are composed of sheets or cords of cells with eosinophilic cytoplasm embedded in a mucinous matrix. As seen here, these cells may display a homogeneous cytoplasm completely lacking in vacuoles or be markedly vacuolated, the latter type being the so-called physaliphorous or bubble-bearing cells.

Chondrosarcomas can also be divided into three subtypes: classic or conventional, mesenchymal, and dedifferentiated. Conventional chondrosarcomas are composed of large, often multinucleated cells within a cartilaginous matrix (Fig. 140-5). Their pathology has three histologic grades. The degree of atypia, number of mitoses, nuclear-to-cytoplasmic ratio, and amount of cartilaginous matrix correlate with the aggressiveness of the tumor, with grade III being the most aggressive. Mesenchymal chondrosarcomas have islands of cartilage mixed with primitive mesenchymal cells, and undifferentiated tumors have a marked sarcomatous appearance. Both mesenchymal and undifferentiated chondrosarcomas have a more aggressive course. Immunohistochemically, chondrosarcomas lack the epithelial antigens expressed by chordomas despite a histologic appearance similar to that of chondroid chondromas.

FIGURE 140-5 Low-power photomicrograph of a conventional chondrosarcoma. Conventional chondrosarcomas are composed of large, often multinucleated cells within a cartilaginous matrix. The degree of atypia, number of mitoses, nuclear-to-cytoplasmic ratio, and amount of cartilaginous matrix correlate with the aggressiveness of the tumor, with grade III being the most aggressive.

Chondrosarcomas express mesenchymal antigens such as HBME-1, tau, and vimentin. Their origin differs from that of chordomas (mesenchymal versus epithelial), and they appear to have a more favorable prognosis. HMBE-1 and tau help distinguish these tumors from secondary carcinoma.

Cytogenetics

Numerous studies of the cytogenetics of chordomas have been published.^22,23 The most common chromosomal abnormalities identified are chromosome 1 monosomy and gain of chromosome 7. Additional chromosomal abnormalities (either gain or loss) described in the literature involve 1q, 2p, 3p, 5p, 9p, 10, 12q, 13q, 17, and 20q. One recent study suggested that an abnormality in the 6p32 region may have a specific association with the genesis of chordoma (rather than progression or recurrence). Although most chordomas are of sporadic origin, case reports of potentially familial chordoma lines have been published. These studies show no consistent cytogenetic change consistent with the heterogeneity seen in sporadic cases. To date, no consistent cytogenetic profile predicts tumor biology or response to therapy.

The variability of chromosomal abnormalities in individual examples of chondrosarcoma prompted one recent reviewer to characterize this tumor’s cytogenetic and molecular variation as “protean” and another to call its genetic instability “striking.” A recent stepwise model of chondrosarcoma development, progression, and recurrence/metastasis proposes chondroblast loss of portions of 5q, 9p, 10q, 11p, 13q, and 19q as components of development into chondrosarcoma. Further loss of portions of 17p and 9p then plays a role in tumor progression. Finally, in this model, 13q loss of heterozygosity facilitates metastasis or recurrence, or both. Although only a simplified model of what is undoubtedly complex tumor biology, this recitation of involved chromosomes well reflects the diversity of genetic abnormalities identified in chondrosarcoma pathology.

Clearly, the genetic instability of these tumors provides an attractive, if elusive target for chemotherapeutic interventions. Unfortunately, however, successful identification and use of appropriate agents have yet to be demonstrated.

Treatment

As a result of the chemoresistant and relatively radioresistant nature of these tumors, the mainstay of treatment since the time of Cushing has been surgical excision. Because local recurrence rather than distant metastasis is responsible for the greatest amount of morbidity and mortality in patients with these lesions, it is a reasonable therapeutic goal to completely excise the tumor. Indeed, it has been observed in the case of sacral chordomas that en bloc resection without violation of the tumor margins or true oncologic resection (i.e., in which complete extirpation of both gross and microscopic disease is achieved) is associated with increased survival and a reduced rate of local recurrence. With the advent of modern skull base techniques, the same principles were applied to cranial chordomas. However, it was soon realized that true oncologic resection was very difficult if not impossible and associated with a high morbidity rate. Enthusiasm for aggressive resection has now been tempered by the recognition that safe surgical decompression followed by adjuvant radiation therapy (RT) can lead to improved long-term survival and better quality of life. Currently, surgeons must weigh three basic treatment options: surgery alone, surgery followed by RT, or biopsy followed by RT, and decide among them based on the individual characteristics of the patient and tumor.

Surgical Approach and Technique

The appropriate surgical approach to clival chordomas and chondrosarcomas is based primarily on the location and size of the tumor (Fig. 140-6). If sufficiently large, two approaches may be used to remove the bulk of the tumor safely. Under all circumstances, careful preoperative planning is essential. It is important to identify which anatomic compartment or compartments the tumor occupies,²⁵ the presence or absence of natural planes, and the relationship of the tumor to vital surrounding structures such as the cranial nerves, cavernous sinus, and brainstem. Based on these characteristics, the surgeon can then choose the appropriate approach or approaches that will safely provide the largest view with the least morbidity.

FIGURE 140-6 Skull base approaches to clival tumors. The amount of bone removed (green) and the working area obtained (orange) for a variety of standard skull base approaches are illustrated. A, An example of a large clival chordoma growing asymmetrically along the skull base is provided for comparison with each illustrated approach. Of note, large tumors can involve multiple compartments and require a combination of standard skull base approaches for adequate resection. B, Far lateral transcondylar transtubercular approach to the lower petroclival region. The working area illustrated represents the quantity of exposure obtained after C1 hemilaminectomy, partial resection of the condyle (to the hypoglossal canal), and removal of the jugular tuberculum. An additional mastoidectomy, resection of the lateral mass of C1, and complete resection of the condyle will increase exposure by just 30%. C, Transcrusal transpetrosal approach to the mid petroclival region. This approach provides up to four times the exposure of a retrolabyrinthine transpetrosal approach, along with hearing preservation. In contrast, more extensive transpetrosal approaches, such as the transotic or transcochlear approaches, provide little additional exposure but have much higher morbidity.²⁴ D, Middle fossa approach to the mid and upper petroclival region. The working area illustrated depicts the area obtained after drilling out Kawase’s triangle. Removal of the bone of Glasscock’s triangle or the zygoma adds no additional exposure, whereas removing the cochlea will double it along the petrous bone. E, Frontotemporal approach to the parasellar region. This standard approach can be tailored to reach a variety of anterior or anterolateral regions of the skull base. It provides only limited access to the upper clivus. F, Transoral with (light orange) or without (dark orange) a Le Fort I osteotomy approach to the clivus. A standard transoral approach allows only limited access to the lower clivus. The addition of a palatal split will increase exposure to the lower clivus somewhat, whereas the addition of a Le Fort I osteotomy will increase clival exposure from 8% to 85%.

A variety of different skull base approaches have been used for the treatment of chordomas. For tumors located along the mid and upper portion of the clivus or extending into the ethmoidal and sphenoidal sinuses, or both, anterior approaches such as the transsphenoidal (with or without the addition of endoscopy), extended subfrontal, or transpharyngeal transpalatal have all been used. If the tumor extends into the lateral cavernous or paraclinoid region, an anterolateral approach such as the pterional (with or without an orbitozygomatic osteotomy) is used, with a more posterior or lateral extension necessitating a middle fossa or transpetrosal approach. For tumors of the lower clivus and foramen magnum, approaches such as the far lateral (transjugular, transcondylar) or transoral may be more appropriate.

Variations on and combinations of these approaches may be necessary, depending on individual variation in the location and size of the tumor, as well as associated comorbid conditions. For example, in a patient with intact hearing and a tumor extending posteriorly from the midclivus, a transcrusal transpetrosal approach may be chosen over the transcochlear version. The transcrusal approach has a lower risk of injury to the facial and auditory nerves and provides an almost identical operative field.^24,26 The exact approach chosen should provide the largest surgical corridor with the least amount of morbidity to minimize brain and cranial nerve retraction.

Once the appropriate approach has been performed, it is imperative to first debulk the tumor to avoid retraction injury to adjacent structures and provide adequate exposure of the tumor margins. This is generally relatively easy because chordomas and chondrosarcomas have a soft gelatinous core that can readily be aspirated. However, calcifications and sequestrations may be present and necessitate the use of rongeurs. Once the bulk of the tumor has been removed, an attempt can be made to remove the residual fibrous outer layer, again with care taken not to injure adjacent neural structures. Cancellous bone, whenever possible, should be resected until the bone begins to bleed in “normal” fashion because chordomas tend to extend long fingers into cancellous bone. It should be noted that violation of the tumor bed can cause seeding of the tumor.²⁷ However, total resection determined at the time of surgery does not predict the future presence or absence of local recurrence; in one recent case series,²⁸ intraoperative gross total resection was associated with the presence of residual tumor on postoperative MRI in approximately 50% of cases (Fig. 140-7).

FIGURE 140-7 Large clival chordoma. Sagittal T1-weighted magnetic resonance images of the large clival chordoma illustrated in Figure 140-3 before (left) and after (right) resection. The preoperative scan is a contrast-enhanced image and demonstrates a modest heterogeneous pattern of enhancement. This large tumor involved several different anatomic compartments and required a combination of skull base approaches. The postoperative image demonstrates nearly total resection with decompression of the brainstem.

Surgical Outcomes

Determining accurate morbidity and mortality statistics is difficult; chordomas and chondrosarcomas are rare tumors, which limits the number of patients available for study. Furthermore, there is considerable variability in specific treatment paradigms, length of follow-up, and the method of determining gross total versus partial resection among reported case series. Another potential complicating factor is that some authors lump together patients with chordomas and chondrosarcomas. Some studies suggest that chondrosarcomas are much less aggressive and have a higher rate of recurrence-free survival.²⁸ Nevertheless, one consistent observation across most studies (reviewed by Muro and colleagues⁸) is that primary tumors, as a group, fare better after resection than do recurrent or previously treated tumors, regardless of whether intraoperative gross total resection was achieved. This suggests that the individual biology of the tumor heavily influences outcome. Reported mortality rates at 5 years range from 33% to 75%.^29–33 The lowest survival rates³⁰ included patients who underwent biopsy only before RT. Surgical morbidity occurred in 13% to 90% of patients and included new or persistent neurological deficits, cerebrospinal fluid leaks, meningitis, and death.

Conclusion

At this time, diagnosis and treatment of both chordomas and chondrosarcomas remain areas of considerable challenge. Radiographic studies, including both CT and MRI, cannot reliably differentiate between the two; they display similar tissue signal intensities and, although chordomas arise from or near the midline and chondrosarcomas tend to be paramedian in distribution, there can be substantial overlap in their locations by the time of initial evaluation. Surgical resection remains the mainstay of treatment of both chordoma and chondrosarcoma. Although the evidence in studies of the effectiveness of RT in the treatment of these tumors remains clouded by the tendency to group tumors of the two pathologies together and by the small numbers of cases studied, this remains a treatment option. Conventional chemotherapy is of limited utility; molecular targets offer the best hope for improved effectiveness.