35.3.1 Chordomas

Chordomas are locally aggressive bone neoplasms that arise from embryonic remnants of the notochord and show a dual epithelial–mesenchymal differentiation.13 The classical chordoma (Fig. 35.1) presents with physaliferous cells, which appear as clusters of large cells separated by fibrous septa into lobules and surrounded by basophilic extracellular matrix rich in mucin and glycogen.14 Well-differentiated chordomas usually have solid sheets of adipocyte-like tumor cells, a myxoid matrix between tumor cells, and nuclei with atypia. Poorly differentiated chordomas shows proliferation of spindle-shaped tumor cells with pleomorphic nuclei, associated with occasional vacuolated tumor cells suggestive of notochordal differentiation.15 Chondroid and dedifferentiated chordomas are rare subtypes that also affect the skull base. Chondroid chordomas have areas that resemble chondrosarcoma in addition to classic chordoma morphology. Dedifferentiated chordoma consists of two components: classic chordoma and a high-grade sarcomatous component that may resemble other high-grade sarcomas (e.g., osteosarcoma, fibrosarcomas). Prognosis of classical and chondroid subtypes is similar, but dedifferentiated chordomas have a much worse prognosis, dictated by the high-grade sarcoma component. Useful immunohistochemical markers for identification of chordomas include brachyury, epithelial markers, S100, and vimentin. Those are negative in chondrosarcomas and thus are useful for differentiation between those lesions.

The molecular pathways and genetic alterations present in chordomas are not completely understood, but some studies done in the last 10 years have added important contributions that may impact the future treatment of chordomas.14 ^,​ 16 The Pl3K/Akt/mTOR pathway is present in most chordomas and may play a crucial role in the pathogenesis of the tumor. The tuberous sclerosis complex (TSC) genes TSC 1 and 2 have also been associated with the development of these tumors.15 In the last 5 years, transcriptome analysis studies demonstrated the upregulation of specific genes in skull base chordomas, including T (brachyury transcription factor), LMX1A, ZIC4, LHX4, and HOXA1.14 ^,​ 16

35.3.2 Chondrosarcomas

Chondrosarcomas are malignant tumors that produce cartilage matrix. In the skull base, they arise from regions of endochondral ossification, such as the petroclival region. These tumors are further categorized into four different histologic subtypes: conventional, mesenchymal, dedifferentiated, and clear cell. Most of these tumors (approximately 85%) are classified as conventional chondrosarcomas, but mesenchymal and dedifferentiated subtypes also affect the skull base (Fig. 35.2).17 The clear cell subtype has been reported only at other anatomical sites, not within the skull base.

Conventional chondrosarcomas are composed of multiple interconnecting lobules of varying size and chondroid or myxoid consistency, with a variable degree of cellularity, myxoid change, and calcification. The chondrocytes may have small or large hyperchromatic nuclei, with sporadic binucleation. These tumors are classified into three grades according to their nuclear size, hyperchromasia, cellularity, and mitotic activity.18 Grade I chondrosarcomas represent most of these tumors. They are poorly cellular lesions, with chondrocytes that have a preponderance of small, densely straining nuclei, and retain a lacunar pattern. The intercellular background is usually chondroid, but there may be some myxoid components. Necrosis, nuclear atypia, and mitotic activity are not characteristics of this type of chondrosarcoma.

Grade II lesions present with areas of increased cellularity and enlarged, paler staining nuclei. A myxoid background is usually noted in areas having more cellularity. A low mitotic activity may be noted (fewer than 2 mitoses per 10 high-power fields [HPF]).

Grade III tumors characteristically display 2 or more mitoses per 10 HPF in the most cellular areas. There is usually a myxoid background associated with spindle or pleomorphic cells, and the lacunar pattern is predominantly lost. Foci of necrosis are usually seen.17 ^,​ 18

The mesenchymal subtype represents 2 to 13% of all chondrosarcomas. It is characterized by a bimorphic pattern with cellular zones of undifferentiated small or spindle cells and islands of hyaline cartilage. A hemangioperycitomatous vascular pattern and osteoclastic giant cells may be present. The cartilaginous area is positive for S100 protein, whereas the area with undifferentiated cells is consistently positive for CD99.17 Dedifferentiated chondrosarcomas contain a well-differentiated component resembling a conventional subtype, as well as areas compatible with high-grade sarcoma, such as fibrosarcoma, osteosarcoma, or histiocytoma. The malignant area determines the prognosis and expected response to currently available chemotherapy protocols. Due to the presence of sarcomatous components, the prognosis of this subtype is significantly inferior to that presented by conventional chondrosarcomas. The clear cell subtype is extremely rare and represents about 1% of all chondrosarcomas. These low-grade malignant lesions consist of clear cells arranged in an indistinct lobular pattern, with round, large, centrally located nuclei having clear cytoplasm and distinct cytoplasmic membranes. They tend to have a better prognosis than the dedifferentiated and mesenchymal subtypes.

The molecular nature of chondrosarcomas has been less studied than that of chordoma. The hedgehog signaling pathway has been associated with chondrosarcomas. The induction of the parathyroid hormone-related protein (PTHLH) pathway (of the Indian hedgehog [IHH]/PTHLH pathway) and reactivation of bcl2 have been implicated in pathogenesis and progression of conventional chondrosarcomas, and bcl2 has been suggested as a reliable marker for the distinction between low-grade chondrosarcomas and enchondromas.17 ^,​ 19

35.5.1 Surgery

Surgical objectives are similar for both tumors: obtain tissue for diagnosis and achieve maximum safe resection and decompression of neurovascular structures for preservation and/or improvement of neurological function and quality of life. However, it is important to clearly delineate the expectations for an oncologic resection of a bony malignancy. Borrowing from concepts originating from spinal oncology, the Weinstein–Boriani–Biagini (WBB) classification for primary spinal column tumors delineates the importance of addressing compartmental (vertebral body) and extracompartmental disease (i.e., tumor extension beyond the vertebral body) in an en bloc fashion. Although an en bloc resection is not feasible in the skull base, these concepts can be applied all the same. The bony skull base should be viewed as the originating compartment and any disease extension transdurally or into subcranial compartments (e.g., infratemporal fossa, longus capitis) viewed as extracompartmental extension. In general, an oncologic resection for any bony sarcoma should address compartmental and extracompartmental disease extensions.

The more aggressive behavior of chordomas demands an aggressive approach in order to maximize the chances of gross total resection and improve overall survival (OS) and progression-free survival (PFS). The first surgery, when planes for tumor dissection are preserved, is usually considered a unique opportunity to achieve maximal resection and long-term control of chordomas.21 Surgical resection should thus be performed in centers that are expert in the management of chordomas, and a full discussion with the patient regarding the goals of surgery and potential complications should be undertaken prior to the surgical procedure. Different surgical approaches may be applied according to the goals of surgery (decompression vs. gross total resection) and the tumor’s size, location, and relation with the dura mater, cranial nerves, and internal carotid artery (ICA). As a result, adequate case selection is mandatory. Skull base chordomas usually are midline tumors that affect the clivus and in most cases do not invade the intradural space. Accordingly, these lesions often are medial to the ICA and cranial nerves, which favors resection through a midline, endoscopic endonasal approach (Fig. 35.3).

An additional benefit of the endoscopic approach includes wide, less invasive exposure for drilling of potentially affected bone in the region. Though useful for a significant portion of clivus chordomas, this approach may be insufficient if the lesion is located lateral to the petrous and/or paraclival ICA or the cranial nerves or if large intradural components are present. In such cases, transcranial approaches should be considered, including posterior petrosectomy and far lateral/transcondylar or retrosigmoid approaches, for tumors having components in the posterior fossa, and an orbitozygomatic approach and its variations for lesions having lateral extensions into the cavernous sinus, middle fossa, and infratemporal fossa. Combination of endoscopic endonasal and transcranial approaches may be required in certain cases, especially in some chordomas, to achieve gross total resection.

Surgical results for chordomas have been mostly reported through retrospective single center observational studies. As an attempt to better evaluate the modern results of surgery, at least three meta-analysis have been recently published in the last 10 years.22 ^,​ 23 ^,​ 24 As reported, gross total resection rates of endoscopic and transcranial approaches are 61 and 48%, respectively.25 However, it is important to note that the extent of resection results is extremely variable in the literature, with rates of complete resection ranging from 0 to 73.7%.23 Transcranial approaches seem to have been associated with higher rates of postoperative cranial nerve dysfunction and meningitis than endoscopic surgery; postoperative cerebrospinal fluid leak rates, however, were similar in open and endoscopic approaches and ranged from 5 to 10%.23 ^,​ 25 It is important to emphasize the selection bias when discussing these results. Patients included in the open approaches cohort had a higher incidence of intradural invasion and larger tumors, whereas patients who underwent endoscopic endonasal surgery had smaller lesions with higher incidence of cavernous sinus invasion. The impact of extent of resection in the prognosis of chordomas has been demonstrated by multiple studies. In 2011 meta-analysis, Di Maio et al demonstrated that 5-year PFS was 87% in patients who had complete resection compared with 50% in patients with incomplete resection (p < 0.0001). The 5-year OS of patients who underwent complete and partial resection was 95% and 71%, (p = 0.001), respectively. Additionally, patients who underwent subtotal resection were 3.83 times more likely to experience a recurrence and 5.85 times more likely to die at 5 years than patients who underwent complete resection.23

Surgical management of recurrent chordomas is challenging and should consider the extent of recurrent disease and the modalities of treatment that were previously employed. The impact of surgery for management of recurrent chordomas was recently assessed by Raza et al.21 The retrospective analysis of 29 patients with 55 recurrences treated at the MD Anderson Cancer Center demonstrated that reoperation of patients who had not received radiation therapy (RTX) was associated with improved median freedom from progression (FFP). However, if radiation had been used, the same benefit was not observed. For patients who have received only surgery, the presence of large recurrences may adversely impact the benefits to be gained through RTX. Reoperation may lead to further reduction of the tumor volume and decompression of the brainstem and cranial nerves, improving the results of radiation therapy.26 ^,​ 27

Understandings of the role of surgery in treating chondrosarcoma have been influenced by several reports published in the last 15 years. Surgery remains the first line of treatment, but goals of surgery depend on grade and subtype of chondrosarcoma. Grade I chondrosarcomas are slow-growing lesions that have low rates of recurrence, which do not seem to require gross total resection for improvement of survival rates.2 In those cases, a less invasive approach may be useful, because the main objectives of the surgical procedure are removal of tumor for pathological analysis and decompression of cranial nerves and/or brainstem. In a recent study by the MD Anderson Cancer Center group, it was observed that gross total resection positively impacted PFS in conventional chondrosarcomas (111.8 vs. 42.9 months, P = 0.201), but statistical significance was not achieved.2 Interestingly, surgery alone was effective for conventional grade I chondrosarcomas, even if postoperative residual tumor was present and no recurrence was noted in those cases during a median follow-up of 67 months (range 13–248 months). Gross total resection seems to play a more important role in grade II and III chondrosarcomas, as suggested by 5-year PFS rates of 0% for subtotal resection and 67% for gross total resection. As discussed for chordomas, the surgical approach should be selected according to the location of the lesions and goals of surgery (Fig. 35.4).

Chondrosarcomas are usually extradural parasellar lesions, centered in the petroclival region, with variable extensions into the cavernous sinus and/or cerebellopontine angle and jugular foramen regions. Depending on their growth pattern, chondrosarcomas may push the ICA and cranial nerves in the cavernous sinus laterally and create a relatively safe midline corridor that can be used for midline endoscopic approaches. Lesions located lateral to the cranial nerves in the posterior and middle fossa, as well as those having extensions into the temporal bone, mandible, and infratemporal fossa, should be considered for transcranial approaches and/or combined approaches.

Surgical treatment of mesenchymal and dedifferentiated subtypes is more challenging. The invasiveness of those lesions and the goals of surgery, which should attempt complete resection, may require more aggressive approaches. In this scenario, gross total resection is associated with significantly better PFS rates (58.2 vs. 1.0 month, P < 0.05).2 A multidisciplinary approach is paramount in such cases (Fig. 35.5). Infiltrative lesions previously diagnosed as mesenchymal or dedifferentiated chondrosarcomas subtypes may benefit from neoadjuvant chemotherapy for tumor reduction and improvement of extent of resection.

35.5.2 Radiation

Radiation treatment plays a major role in the management of chordomas. Adjuvant radiotherapy is recommended after surgical resection of skull base chordomas and is the main treatment modality for tumors deemed inoperable.28 It is challenging and requires advanced planning, because it demands precise delivery of high doses of radiation to a relatively large field while avoiding damage to surrounding neural structures. Although most centers favor a strategy that maximizes radiation delivery to the tumor volume while respecting the tolerances of surrounding normal tissue, another school of thought favors sufficient irradiation, even at the cost of exceeding the dose threshold of surrounding normal tissue.29 Conventional fractionated radiotherapy and proton therapy are the most common techniques, but stereotactic radiosurgery (SRS) and carbon ions have also been studied.30 ^,​ 31 Hadrons have been gaining space in the treatment of chordomas and chondrosarcomas in the last 10 years. They can improve the radiobiological effect of radiation and minimize injury to surrounding neural tissue. These high-dose protons or charged particles include carbon ions, helium, and neon.32 ^,​ 33

Proton beam therapy is especially useful for treatment of skull base chordomas and chondrosarcomas. The initial results achieved in the 1980s and 1990s demonstrated the ability of this modality to deliver high doses of radiation (close to 70 Gy) while limiting the impact of radiation on the normal surrounding neurovascular structures, with 5-year control rates of 59 to 82%.34 ^,​ 35 ^,​ 36 Proton beam radiotherapy has multiple favorable features for the management of chordomas.28 Among the advantages of proton beam is the sudden dose decline beyond the target, which is related with the characteristic Bragg’s peak effect of proton beam radiation. Another benefit is that protons or charged particles allow delivery of higher doses of radiation to the target volume, reducing the collateral radiation injury and improving radiobiological effect.32 As a result, proton beam radiation therapy is able to deliver high radiation dosage more precisely than classic conventional fractionated therapy and may also be superior for tumor control and preservation of neurological function.30 ^,​ 31 ^,​ 37

For planning of RTX, the primary clinical target volume (CTV1) should include all volumes at risk for microscopic disease, including areas of preoperative tumor extension, and a second volume (CTV2) receiving a higher boost-dose of radiation should encompass any residual microscopic disease in the tumor bed after surgery. A third clinical target may be added to the treatment plan if gross residual lesions are present. Tumor seeding along the surgical corridor may occur in 5% of patients, and some groups suggest including the entire surgical corridor in the treatment plan.37 Chordomas are radioresistant tumors and demand doses of at least 74 Gy, using conventional fractionation. This dose may lead to damage to brainstem, cranial nerves, and optic pathways around the lesion, so those structures should be contoured, and dosage constraint is recommended.

Hug et al reported the results of 58 patients (33 chordomas and 25 chondrosarcomas) treated with proton beam in a single center. The authors observed a 5-year local control rate of 59% for chordomas and 75% for chondrosarcomas.38 Preradiation tumor volume > 25 mL and brainstem involvement were factors related to treatment failure. Grade III and IV toxicities were diagnosed in 4 (7%) of 58 patients and were symptomatic in 3 (5%). The failure of treatment was likely related to insufficient delivery of radiation to parts of the tumor, such as those in contact with the brainstem and cranial nerves. Austin et al analyzed 26 patients who had local tumor recurrence after proton-based radiation therapy (proton RT) at Massachusetts General Hospital (MGH)/HCL and concluded, based on CT and MR review, that treatment in 75% of patients failed in regions receiving less than the prescribed dose because of normal tissue constraints.39 The impact of tumor volume on the outcomes following RTX has also been demonstrated by Igati et al and McDonald et al.26 ^,​ 40 In those studies, tumor volumes > 30 mL and > 20 mL, respectively, were associated with worse local control rates. Additionally, in the study by McDonald et al, it was observed that each additional 1 mL of tumor volume was linked with an increased risk of progression and brainstem compression, as well as that a radiation dose less than 74.5 Gy delivered to 1 mL of gross tumor volume (GTV) was predictive of treatment failure.26 A meta-analysis by Matloob et al demonstrated that surgical resection followed by proton beam radiation was associated with 5-year disease control in 46 to 78% and 5-year survival rates of 66 to 87%.32

The most prevalent method for the delivery of proton radiotherapy is the passive scattering technique.41 In 1980, Kanai proposed scanning a narrow pencil-beam in three dimensions through the target volume.42 The flexibility of the spot scanning approach has been extended to deliver intensity-modulated proton therapy (IMPT), a direct equivalent to intensity-modulated radiotherapy (IMRT) with photons. Spot scanning provides greater control over the proximal aspects of the beam while improving conformity of the high-dose regions. A recent study from MD Anderson reported preliminary results with spot scanning proton therapy for chordomas and chondrosarcomas. The results demonstrated that compared with passive scanning, spot scanning plans provided improved high-dose conformity, sparing delivery to dose-limiting structures.43 Ares et al reported a similar result using this technique, with 5-year local control rates of 81% for chordomas and 94% for chondrosarcomas and high-grade toxicity in 4 patients (6.25%).41 No patient experienced brainstem toxicity.

Negative aspects related to proton beam therapy include the availability of fewer experienced facilities; dose distributions’ being influenced to a greater degree by differences in density, so that air cavities and surgical hardware must be considered during the treatment planning process to a greater degree than is necessary for IMRT; and significantly higher cost.31

It is important to note that the evolution of photon-based linear accelerator therapy in the last 10 years has challenged the superiority of proton-based therapy for chordomas. The development of modern multileaf collimators (MLCs) allows for IMRT while onboard image-guidance (IG) systems permit near-real-time tracking during delivery, and robotic technology has been incorporated to ensure millimeter precision in dose delivery.44 This has improved results, and several centers have adopted IG-IMRT for skull base chordoma and chondrosarcoma, delivering doses equivalent to those of proton therapy.44 ^,​ 45 ^,​ 46 A recent study by Sahgal et al evaluated 24 patients who had skull base chordomas and who underwent surgery followed by IG-IMRT at the Princess Margaret Hospital/University of Toronto. The authors achieved a median total delivery dose of 76 Gy and obtained 5-year overall survival and local control rates of 85.6% and 65.3%, comparable to results reported after proton beam therapy. Emerging therapies, such as IMPT and in-room cone-beam CT, will likely impact the results of treatment of skull base chordomas.44

Like chordomas, chondrosarcomas require delivery of high radiation doses, > 60 Gy, usually delivered via IMRT/SRS or proton beam therapy. Similar limitations and side effects are present in the treatment of those tumors. The role of adjuvant radiotherapy for chondrosarcomas is not as clear as for chordomas.2 ^,​ 47 Some centers recommend radiotherapy for all chondrosarcomas after surgery, whereas others select this treatment for tumors with aggressive features (conventional grade III or mesenchymal/dedifferentiated chondrosarcomas).2 ^,​ 44 A recent study done by the MD Anderson group demonstrated that adjuvant radiotherapy significantly impacted PFS in conventional grade II and III chondrosarcomas (182 vs. 79 mo, P < 0.05) and had a positive trend for mesenchymal/dedifferentiated CSAs (43.5 vs. 22.0 mo).2 Those results are similar to previously reported data that demonstrate 5-year overall survival and local control rates of 87.8% and 88.1%; they suggest that grade I skull base chondrosarcomas may have adjuvant radiation postponed but that other subtypes should receive postoperative radiotherapy (Fig. 35.5).

Finally, it is important to note that the ideal technique for radiation delivery in both tumors is a matter of controversy. SRS is typically not useful as first-line adjuvant therapy, because chordomas require high dosing to cover the entire tumor bed. It is, however, a useful tool for the treatment of selected focal recurrences featuring small tumor volumes within or adjacent to previous radiation fields. As demonstrated in Raza et al in 2017, SRS may have an impact on the FFP of patients who have recurrent chordomas.21 The largest SRS series to date, by the North American Gamma Knife Consortium, reports that optimal tumor control outcomes were achieved for tumor volumes less than 7 mL using a median marginal dose greater than 15 Gy.46 Although previous radiation has not been demonstrated to negatively impact control after SRS, previous radiation fields can limit the marginal delivered dose. Adverse radiation events (ARE), as expected, may follow. In that same study, AREs occurred in 30% of patients having undergone previous radiation therapy and consisted of grade II/III events affecting the cranial nerves and pituitary gland.46