Radiotherapy for Mobile Spine and Sacral Chordoma: A Critical Review and Practical Guide from the Spine Tumor Academy

Simple Summary Chordomas are rare tumors of the embryologic spinal cord remnant. They are locally aggressive and typically managed with surgery in combination with radiation therapy. However, there is great variability in practice patterns including different radiation treatment types and approaches, and limited high-level data to drive decision making. The purpose of this manuscript was to summarize the current literature specific to radiotherapy in the management of spine and sacral chordoma and to provide a practical guide on behalf of the Spine Tumor Academy, an international group of spinal oncology experts. Abstract Chordomas are rare tumors of the embryologic spinal cord remnant. They are locally aggressive and typically managed with surgery and either adjuvant or neoadjuvant radiation therapy. However, there is great variability in practice patterns including radiation type and fractionation regimen, and limited high-level data to drive decision making. The purpose of this manuscript was to summarize the current literature specific to radiotherapy in the management of spine and sacral chordoma and to provide practice recommendations on behalf of the Spine Tumor Academy. A systematic review of the literature was performed using the Preferred Reporting Items for Systematic reviews and Meta-Analyses (PRISMA) approach. Medline and Embase databases were utilized. The primary outcome measure was the rate of local control. A detailed review and interpretation of eligible studies is provided in the manuscript tables and text. Recommendations were defined as follows: (1) consensus: approved by >75% of experts; (2) predominant: approved by >50% of experts; (3) controversial: not approved by a majority of experts. Expert consensus supports dose escalation as critical in optimizing local control following radiation therapy for chordoma. In addition, comprehensive target volumes including sites of potential microscopic involvement improve local control compared with focal targets. Level I and high-quality multi-institutional data comparing treatment modalities, sequencing of radiation and surgery, and dose/fractionation schedules are needed to optimize patient outcomes in this locally aggressive malignancy.


Introduction
Chordomas are rare tumors of the embryologic notochord remnant. They may occur anywhere within the axial skeleton, but are most common in the base of skull or sacrum. However, chordomas do occur in the mobile spine as well. Although pathologically benign in appearance and generally slow growing with a median overall survival of approximately a decade [1], these tumors are considered malignant as they have metastatic potential. Specifically, 5-40% of patients develop distant metastases during their disease course [2]. Nonetheless, the primary cause of morbidity and mortality in chordoma is local recurrence.
Given the locally aggressive nature of chordoma, the standard-of-care management consists of aggressive surgical resection in combination with either neoadjuvant or adjuvant radiation therapy as deemed clinically appropriate. The role of radiation therapy is controversial and there are no level 1 data to guide decision making. As such, the optimal radiation technique and sequencing remains unclear and may consist of proton, photon, or heavy ion therapy using either conventional fractionation or hypofractionated stereotactic radiosurgery. The purpose of this collaboration was to summarize the current literature specific to radiotherapy in the management of spine and sacral chordoma and to provide practice recommendations for treatment on behalf of the Spine Tumor Academy. A brief summary of imaging and surgical approaches is also included for the benefit of the oncology audience.

Materials and Methods
A systematic review of the literature was performed using the Preferred Reporting Items for Systematic reviews and Meta-Analyses (PRISMA) approach.

Search Strategy
Medline and Embase databases were utilized to search for manuscripts reporting outcomes following surgery and radiation therapy for spine and sacral chordoma with a search end date of 29 October 2021. Search words included "spine OR spinal OR sacrum OR sacral" AND the following: "chordoma and radiation", "chordoma and stereotactic", "chordoma and SRS", "chordoma and SABR", "chordoma and SBRT", "chordoma and radiosurgery", "chordoma and carbon", "chordoma and IMRT", and "chordoma and external beam". Prospective studies and retrospective series that included at least 10 patients with spinal/sacral chordoma with results specific to the spinal/sacral chordoma subtype reported separately were included. Studies that included skull base chordoma were also included provided that results for the spinal/sacral subgroup were reported separately. Only studies published in English or with an English translation available were considered eligible. Clinicaltrials.gov was also utilized to identify ongoing trials evaluating radiation therapy approaches in spine/sacral chordoma. Abstracts without a published manuscript were excluded, as were dosimetric analyses without clinical outcome data, systematic reviews, meta-analyses, pre-clinical studies, and those in which clinical outcomes were not reported. In addition, manuscripts that reported outcomes for multiple histologies in combination with chordoma, studies including patients who did not undergo radiation or in which details of radiation dose and technique were not available, and manuscripts that reported outcomes for chordomas of skull base in combination with chordoma of the spine/sacrum were excluded from this review.

Outcome Measures
Data collected during the systematic review included local control, tumor location, surgery including extent of surgery and timing relative to RT, radiation technique and modality, prescription dose/fractionation, prior overlapping RT including type and number of patients, and overall survival. Toxicity including but not limited to wound healing complications, spinal cord myelopathy, and nerve plexopathy were included.
The spine tumor academy is an international multi-disciplinary academic collaboration of spinal oncology experts across fields including neurological surgery, orthopedic surgery, radiation oncology, medical oncology and neuro-radiology. A preliminary draft of the manuscript was reviewed at the December 2021 Spine Tumor Academy meeting which was attended by 55 people from six countries including Germany, Canada, the United States, Austria, the Netherlands, and Italy. The manuscript then underwent serial revisions and peer review by members of the Spine Tumor Academy. Ultimately, 15 experts were offered authorship given their leadership roles and extensive contributions to the manuscript. Levels of agreement regarding the recommendations outlined in the guidelines were defined as follows: (1) consensus: selected by at least 75% of respondents; (2) predominant: selected by at least 50% of respondents; and (3) controversial: no single response selected by a majority of respondents. Descriptive statistics were used to review the results.

Results
The details of the PRISMA search are shown in Figure 1. Primary database screening identified a total of 1215 candidate citations (714 from Embase and 501 from Medline). After removal of 439 duplicates, 173 conference abstracts, 80 review articles, 14 commentary, 12 letters, 6 editorials, 5 conference reviews, 4 short surveys, and 1 erratum, 481 candidate citations remained. Of those 481, 45 met the inclusion criteria, including those reporting clinical outcomes of ≥10 patients with chordoma of the spine/sacrum treated with radiation.

Proton Beam Therapy
Proton beam therapy (PBT) is a charged particle-based treatment that has been shown to address the need for dose escalation to the target for improved tumor control with the ability to spare critical organs at risk (OAR). This is achieved by the intrinsic physical properties of proton therapy where there is a penetrating dose deposition along the beam path as the particle slows down until it stops at the end of the range at which it deposits most of its dose, described by the characteristic Bragg peak, with no exit dose. Primary database screening identified a total of 1215 candidate citations (714 from Embase and 501 from Medline). After removal of 439 duplicates, 173 conference abstracts, 80 review articles, 14 commentary, 12 letters, 6 editorials, 5 conference reviews, 4 short surveys, and 1 erratum, 481 candidate citations remained. Of those 481, 45 met the inclusion criteria, including those reporting clinical outcomes of ≥10 patients with chordoma of the spine/sacrum treated with radiation.

Proton Beam Therapy
Proton beam therapy (PBT) is a charged particle-based treatment that has been shown to address the need for dose escalation to the target for improved tumor control with the ability to spare critical organs at risk (OAR). This is achieved by the intrinsic physical properties of proton therapy where there is a penetrating dose deposition along the beam path as the particle slows down until it stops at the end of the range at which it deposits most of its dose, described by the characteristic Bragg peak, with no exit dose. Compared with photon-based radiotherapy with an exponential decay function, this allows for reduced dose to OARs distal to the desired target and decreased integral dose (low-dose bath) of radiation that may translate into reduced acute and late RT treatment morbidity and secondary malignancy risk.
Evolution in the technology for delivery of proton therapy from passive-scattered (e.g., double-scatter) to pencil beam scanning (PBS) has allowed for increased high-dose conformality, particularly for the proximal component of the target, such as with concave target volumes (e.g., chordomas involving the vertebral bodies that require sparing of the adjacent spinal cord) and decreased skin dose. PBS consists of a thin pencil-beam "spot" that has a given depth defined by the beam energy. This "spot" is actively scanned with magnets on a voxel-basis on a given layer. Then with modulation of the beam energy, dose painting of the next layer commences until the target coverage is complete.
Beam angle selection is of paramount importance for PBT to maximize target coverage robustness and minimize range uncertainty. Key considerations in regard to beam selection for chordoma proton therapy plans include the following: (1) limiting distance from entrance to the target; (2) minimizing the entry beam path traversing structures with air and/or bowel gas with uncertain positions on a daily basis; (3) limiting beam number to reduce integral dose; (4) maximizing beam angle separation for maximal skin sparing (e.g., this may require prone positioning of the patient to avoid rails on the table for lower T, L spine, and sacrum plans); (5) avoiding multiple beams' end of ranges occurring in the same structure, particularly neural structures, given concern for increased relative biological effectiveness; and (6) if high Z surgical stabilization hardware is present (e.g., titanium), minimizing traversing through hardware with consideration of non-coplanar beams and/or mixed photon/proton treatment plans to maximize confidence in dosedelivery to the target and improve confidence in critical OAR dosimetry. PBT treatment plans for chordomas below the spinal cord often consist of two posterior oblique beams separated at an optimal angle for maximal skin sparing and robustness, while plans at the level of the spinal cord in the mobile spine (typically L1-2 and above) may require up to 4-6 different angles depending on the location of the tumor in relationship to the spinal cord, plexus, and other critical OARs.
PBT doses are expressed as GyRBE (relative biological equivalent) with a conversion factor of 1.1 used to account for its higher relative biological properties. Most studies evaluating PBT have investigated dose escalation to total doses ≥70 GyRBE in conventional fractionation (1.8-2 Gy per fraction), daily, five times per week. In general, comprehensive target volume coverage particularly for the at-risk microscopic clinical target volume (CTV) has been employed, which contrasts with reported more focal target volumes typically used with other heavy particle therapy and SBRT. For spinal cord delineation, the Massachusetts General Hospital (MGH) [3,4] and Paul Scherrer Institute (PSI) [5] method is the most well described, where the treatment planning CT is fused to a T2 MRI or CT myelogram, if surgical hardware is present, to delineate the spinal cord into two structures: (1) spinal cord core (cSC) which is a 2-3 mm region-of-interest in the geometric center of the spinal cord; and (2) the spinal cord surface (sSC).
Common themes emerged regarding adverse prognostic factors of local control for patients treated with PBT. Treatment in the upfront setting for primary chordomas resulted in more optimal outcomes compared with treatment in the recurrent setting [4,7,10,15,19,24], reiterating the importance of upfront multi-disciplinary evaluation for timely and appropriate multimodal care. Given nearly all PBT series included only patients treated with doseescalated radiation therapy (≥60-70 Gy or higher), a clear a dose-response relationship was not identified except for in one study of sacral chordomas showing improved local control with doses ≥ 70 Gy (HR 0.52, p = 0.17) particularly amongst patients with an R1 margin (HR 0.40, p = 0.051) or those treated with PBT compared with photon therapy (HR 0.56, p = 0.23) [11].
High Z surgical stabilization hardware (e.g., titanium) raises concern for technical limitations and dosimetric uncertainty that may contribute to dose "shadowing" (underdosage) of the target distal to the beam path with particle therapy, where the experience from PSI reports a significant decrement in the 5-year local control of 73.4% and 50% for patients without and with surgical stabilization, respectively (p = 0.02) [20,22]. Potential solutions to mitigate this effect include upfront evaluation with the surgeon to determine the extent of surgery indicated and/or necessity of hardware, position of hardware, the consideration of novel carbon-reinforced polyetheretherketone (PEEK) stabilization alternatives that result in reduced CT artifacts and less impact on proton dosimetry because of lower Z composition [26,27], evaluating the feasibility of a mixed photon/proton plan, and/or delivering a meaningful component of the microscopic dose (e.g.,  in the preoperative setting prior to a postoperative boost to reduce the need to cover the entire surgical resection bed. Importantly, MGH has shown that using a preoperative followed by an individualized post-operative boost approach compared with adjuvant PBT alone results in improved 5-year local control of 85% vs. 56%, respectively (p = 0.019), with no local failures for patients who underwent en bloc resection [19].
There is a clinical need for consensus guidelines regarding target and critical OAR delineation and dose constraints for chordoma patients treated with PBT, as some series with more focal target volumes suggest inferior local control compared with more comprehensive volumes [20][21][22]28] as well as more frequent patterns of failure in proximity to dose-limiting OARs, such as the spinal cord [20].
While PBT allows for decreased dose to OARs distal to the target, critical structures immediately adjacent or within the target volume are still at risk for significant treatmentrelated morbidity because of the high doses required for tumor control. Across all PBT series, there were only two incidences of grade 3 or greater spinal cord myelopathy, where one patient developed renewed tetraplegia 17 months after initially presenting with temporary tetraparesis that improved with surgical decompression (Dmax to sSC and cSC were 57.8 GyRBE and 54.1 Gy RBE, respectively) [5,22], and the second patient developed transient paralysis 2 years after treatment when undergoing chemotherapy conditioning for an autologous stem cell transplant for myelodysplastic syndrome [16]. In the subacute setting, there is a reported approximate 5% rate of Lhermitte's syndrome, which is a temporary demyelination phenomenon that resolves spontaneously [5,24]. In the PSI series, when adhering to dose constraints of D2% of the sSC receiving 64 GyRBE (reduced to 60 GyRBE if the target volume was longer than 3 vertebrae) and the cSC receiving 54 GyRBE, only 4% (n = 3/71) developed grade 2 or greater neurologic toxicity, whereas 40% (n = 2/5) whose dose constraints were exceeded developed toxicity [5]. Nerve plexus neuropathies have been reported in approximately 3-5%, which may manifest as pain, numbness, tingling, weakness, foot drop, erectile dysfunction, and bladder or bowel dysfunction, where doses are typically in the range of 77.4-85 GyRBE when they have been reported [4].
Other toxicities after PBT include a significant impact on the rate of wound healing toxicity with reported values of 21.6% (predominantly in patients with sacral tumors) treated with preoperative PBT compared with 12% for those treated with postoperative PBT alone [19], highlighting the absence of "skin sparing" with proton therapy and the importance of close collaboration with surgical colleagues, including plastic surgery, for consideration of flap-based closures to maximize wound healing. Other reported adverse events include a low (0-5%) rate of insufficiency fracture, esophageal stricture, subcutaneous fistula, femoral insufficiency requiring hip replacement, ureteral stenosis, laryngeal necrosis, rectal ulcer and bleeding, menopause, and bowel fistula or perforation requiring a colostomy. For sacral chordomas within 1 cm of the small bowel and/or rectum, one may consider upfront surgical spacer placement, which allows for the necessary distance for particle beam dose fall-off [23]. Reported rates of secondary malignancy are 0-5%.

Carbon Ion and Other Heavy Particle Therapy
There is an ongoing discussion about the efficacy of carbon ion radiotherapy in chordomas. In relation to protons, carbon ions offer comparable physical properties, with a low energy (and thus dose) deposition in the entry channel of the beam and precise dose deposition in the Bragg Peak, followed by a steep dose fall-off in normal tissue behind the target [29]. This, as in protons, leads to a reduction of integral dose in patients. In contrast to protons and photons, carbon ions are associated with a higher relative biological effectiveness (RBE); several preclinical studies have demonstrated this increased efficacy in various tumor entities, including pancreatic cancer, gliomas, and also sarcomas [30][31][32][33][34][35][36][37][38]. Moreover, there is a strong rationale that carbon ions can overcome radiation resistance caused by hypoxia [38]. Since chordomas are radiation-resistant tumors requiring high local doses, there is a strong rationale for carbon ions in this tumor entity, not only in terms of dose escalation based on the superior dose distribution of particles, but also based on the biological properties.
However, to date, no large series are available for chordomas of the mobile spine. Regarding chordomas, most data are available from skull-base chordomas, where particle therapy probably has the strongest rational especially because of the intricate anatomy. Most large series based on skull-base chordomas report local control rates that are relatively high compared with older photon series. For example, Koto [39]. A recent Heidelberg series by Uhl and colleagues including 155 skullbase chordomas treated with carbon ions published a local control rate of 72% and 54% at 5 and 10 years [40]. For chordomas located along the mobile spine, the data are scarce; however, smaller series have demonstrated high efficacy and low rates of side effects in a number of tumor entities and locations. The data are often mixed with chondrosarcomas of the spine, or analyzed together with sacral chordomas which are generally a different entity because of the surgical and also radiation oncology requirements related to the distinct differences in anatomy.
In terms of toxicity, rates of sacral fractures following carbon ion therapy for sacral chordoma were high, impacting approximately half of patients [41]. However, the authors did note that only about a third of fractures were clinically symptomatic, requiring regular medical care and pain therapy. In addition, rates of wound healing complications following carbon ion and heavy particle therapy were high. For example, a study of patients treated with helium and neon therapy demonstrated a 35% rate of chronic wound complications [42,43]. Table 2 summarizes the eligible series of carbon ion radiotherapy including chordomas of the mobile spine. The readers are also directed to a comprehensive review of spinal and sacral chordomas treated with carbon ions written by Pennington et al. [44].  Abbreviations: mo = months; STR = subtotal resection; GTR = gross total resection; NR = not reported; RT = radiation therapy; yr = year; SBRT = stereotactic body radiation therapy; G1 = grade 1; G2 = grade 2; CI = confidence interval; GI = gastrointestinal; G3 = grade 3; PE = pulmonary embolism. Data are listed for the specific group when available or the overall cohort if group-specific data are not available.

Stereotactic Body Radiation Therapy (SBRT)
Advances in radiation technology including micro-multileaf collimators, cone beam CT scans, robotic systems, and real-time image guidance have allowed for progressively more precise delivery of photon therapy utilizing steep dose gradients and the emergence of SBRT. SBRT is increasingly available at many community and academic centers throughout the world, and thus is more readily available than charged particle therapies such as proton and carbon ion therapy, which have been discussed in earlier sections. Hypofractionated stereotactic regimens allow the delivery of ablative doses of radiation therapy by limiting the dose to adjacent normal tissues. Compared with conventionally fractionated radiation therapy, SBRT activates unique cell-killing pathways including apoptosis and takes advantage of radiobiologic principles including a decrease in sublethal damage repair and repopulation of tumor cells between fractions. These regimens also help to destroy microvasculature and overcome the traditional radioresistance of hypoxic cells which may be found in the center of large tumors such as chordoma.
Our systematic review identified nine retrospective manuscripts including a total of 197 patients and no prospective clinical trials that met inclusion criteria for evaluating outcomes for primary and recurrent mobile spine and/or sacral chordomas treated with SBRT in the preoperative/postoperative, adjuvant, and/or definitive setting. The data are shown in Table 3. With a median follow-up of 34 months (range 1.7-216 months) across all studies, the median overall crude local recurrence free survival was 71%, ranging from 45% to 95%. For the series that reported local control for the treatment naïve patients separately, the median overall local recurrence free survival was 92% (range 86-95%). Local failure was the primary pattern of failure with distant metastatic disease developing in a median of 17.5% (range 0-30%). The median crude overall survival among reported studies was 72% (range 59.3% to 92%).  Abbreviations: mo = months; STR = subtotal resection; GTR = gross total resection; NR = not reported; RT = radiation therapy; yr = year; SBRT = stereotactic body radiation therapy; G1 = grade 1; G2 = grade 2; CI = confidence interval; GI = gastrointestinal; G3 = grade 3; PE = pulmonary embolism. Data are listed for the specific group when available or the overall cohort if group-specific data are not available.
Higher prescription doses were reported to be associated with superior local control. Specifically, in a series reporting outcomes of primarily fractionated SBRT, no local recurrences occurred in patients receiving a BED2 < 140 Gy [61]. Similarly, in two series delivering predominantly single fraction SBRT, local control was reported to be approximately 95% in patients receiving 24 Gy [55,60]. Furthermore, superior local control was reported in treatment-naïve patients undergoing definitive management than in the salvage setting. In this light, Chen and colleagues report no local recurrences in their subset of 17 patients receiving neoadjuvant high-dose hypofractionated SBRT followed by surgical resection with curative intent [61].
Manuscripts variably reported the normal tissue constraints that were utilized in treatment planning. Of those that reported spinal cord constraints, 14 Gy in a single fraction was used for the true spinal cord in a single fraction and 25.3 Gy was used for the spinal cord plus 2 mm or thecal sac in five fractions.
Overall, the toxicities associated with SBRT were low and generally correlated with anticipated sequelae based on the treated spinal levels. Skin toxicity was rare and although there was no direct comparison of modalities across studies, seemingly lower than that reported in studies of heavy particle therapy. Of greater concern, a single study [57] did report at 13% risk of grade 2 spinal cord myelopathy following SBRT, although it did resolve with corticosteroid administration.
A significant concern with adjuvant and neoadjuvant radiation therapy is the risk of wound healing complications; however, the rates reported were low in a median of 8.9% of patients (range: 3.3-1%) in studies that reported this toxicity [58,61,62]. This rate is comparable to patients undergoing surgery alone and lower than observed in patients treated with alternative approaches such as proton therapy. We speculate that this may be because of lower skin doses with SBRT given the rapid radiation dose fall-off over millimeters.
Although aggressive surgical resection remains the cornerstone of care, emerging data from a single institution suggest reasonable local control with SBRT alone. Specifically, Yamada and colleagues [55] reported 95% local control following 24 Gy in a single fraction of SBRT in a cohort of patients in which nearly one-third did not undergo surgery and all surgical patients had gross residual disease post-operatively. However, it is critical to note that the median follow-up in this series was only 38.8 months overall and 16.5 months in the subset of treatment-naïve patients. It is possible that these control rates may decrease over time and long-term follow-up is necessary.

Radiation Alone
Although aggressive surgical resection in considered the standard of care in the management of chordoma, there are times when radiation alone may be considered. The primary advantage of definitive radiotherapy is a reduction in morbidity and recovery from surgery or as a management option in medically inoperable patients. To date, there are no prospective or retrospective studies comparing radiation therapy alone to surgery followed by radiation therapy or comparing radiation treatment modalities in patients undergoing radiation alone. Our systematic review identified nine retrospective manuscripts and a single prospective phase 1-2 clinical trial including a total of 641 patients treated with radiation alone that met the inclusion criteria for evaluating outcomes for primary and recurrent mobile spine and/or sacral chordomas treated with radiation therapy alone. These data are summarized in Table 4. With a median follow-up of 52 months (range 37-80 months) across all studies, the median local control (at 3-5 year depending on the study) was 80%, ranging from 62% to 94%. Four studies utilized exclusively carbon ion therapy, two exclusively proton therapy, and three utilized combinations of charged particle ± photon therapies. For these fractionated regimens, four had a median prescription dose of 70.4 Gy RBE while three others had higher median prescription doses ranging from 74-80 Gy RBE in fractions ranging from 2.2-4.6 Gy RBE. A single study reported outcomes following SBRT alone to a median dose of 24 Gy in a single fraction and revealed a 2-year local control of 100%. The median 5-year overall survival for the studies was reported as 84% (range 74% to 88%). Toxicities were limited with the most common sequelae including sacral insufficiency fractures as well as both acute and late skin complications.  Abbreviations: mo = months; STR = subtotal resection; GTR = gross total resection; NR = not reported; RT = radiation therapy; yr = year; SBRT = stereotactic body radiation therapy; G1 = grade 1; G2 = grade 2; CI = confidence interval; GI = gastrointestinal; G3 = grade 3; PE = pulmonary embolism. Data are listed for the specific group when available or the overall cohort if group-specific data are not available.
Taken in aggregate, radiation alone remains a reasonable option in a subset of patients who are medically inoperable or elect to forgo the potential risks associated with an aggressive surgical procedure. It is important to note that all studies utilized relatively dose-escalated prescription doses in an effort to overcome the known radioresistance of chordoma. Although there are currently not sufficient data to compare outcomes following protons, carbon ion, photon, and SBRT, the local control across studies was excellent, although a longer-term follow-up will be essential.

Timing of RT
Local recurrence or progression following surgical resection occurs frequently because of the inability to achieve wide margin excision in patients with spinal and sacral chordomas. Postoperative radiotherapy using the approaches described above including photon-based intensity modulated radiotherapy, proton therapy, and carbon ion therapy has been utilized to improve local control. Based on sarcoma literature, delivery of radiation therapy in the adjuvant setting may minimize the risk of wound healing complications. In addition, it allows providers to determine the need for RT based on the extent of resection and to work with surgeons to identify the regions of close or positive margins that may be at the highest risk of recurrence. However, target delineation is more challenging in the adjuvant setting given the difficulty in discerning post-operative change from residual/recurrent disease. In addition, it is possible that tumor cells may contaminate the surgical field at areas more remote from the original gross disease. As a result, the radiation target is typically larger in the adjuvant setting than in the neoadjuvant setting, resulting in the delivery of higher doses of radiation to adjacent normal tissues.
By contrast, neoadjuvant radiotherapy simplifies target delineation as the characteristic T2 hyperintense regions of gross disease may be identified with greater confidence than in the post-operative setting. As a result, the margins may be tighter, minimizing radiation dose to the adjacent normal structure. Discussions with surgeons are essential to identify the regions at highest risk of a positive margin post-operatively so that the target and prescription dose may be modified accordingly. The rationale of neoadjuvant radiation therapy is to effectively sterilize any cells that may spill from the capsule at the time of surgery and thereby reduce the risk of microscopic residual leading to local recurrence. In addition, it may be beneficial to facilitate negative margins but with less surgical morbidity associated with sacrifice of critical neural structures. The greatest concern in this setting is the potentially increased risk of wound healing complications. To minimize this risk, care must be taken to minimize radiation dose to the skin, especially since a clear superficial margin is generally not a challenge at the time of surgical resection.
Ultimately, the decision to offer radiation therapy in the adjuvant or neoadjuvant settings is often driven by institutional bias, as there have been no studies directly comparing the two approaches. Some institutions utilize a compromise approach and deliver some dose in the neoadjuvant setting with and additional boost post-operatively.
Several retrospective studies have attempted to compare outcomes based on the timing of radiation therapy. For example, in a study from MGH [19], patients were treated with either adjuvant radiation therapy using photons or a combination of pre-operative plus postoperative combined photon and proton therapy. Patients who had preoperative plus postoperative radiotherapy showed a trend toward superior local control. However, an alternative retrospective study by the Sacral Tumor Society [11] suggests increased risk of wound complications using this approach.
Other studies have evaluated SBRT in the adjuvant and neoadjuvant setting. For example, the Johns Hopkins University [61] series demonstrated negative margins in all patients undergoing en bloc resection following neoadjuvant SBRT with no local recurrences during the study period. It is important to note that approximately one-third of patients developed post-operative wound healing complications, although the authors noted that this rate is comparable to the rate in patients undergoing surgery alone without radiation therapy. Memorial Sloan-Kettering Cancer Center [60] similarly reported excellent outcomes in 11 sacral chordoma patients receiving preoperative SBRT, with a 3-year local recurrence-free survival of 90%. However, they did not report outcomes specific to the adjuvant radiotherapy group and complications were not reported separately based on the timing of radiation. As such, given the very limited literature, the optimal timing of radiation therapy relative to surgery remains unclear. Table 5 summarizes studies showing outcomes for both preoperative and postoperative RT for mobile spine/sacral chordomas.  Abbreviations: mo = months; STR = subtotal resection; GTR = gross total resection; NR = not reported; RT = radiation therapy; yr = year; SBRT = stereotactic body radiation therapy; G1 = grade 1; G2 = grade 2; CI = confidence interval; GI = gastrointestinal; G3 = grade 3; PE = pulmonary embolism. Data are listed for the specific group when available or the overall cohort if group specific data are not available.

Summary of Ongoing Clinical Trials
Given the relatively high recurrence rates in management of chordoma, clinical trials are of the utmost importance in improving outcomes and optimizing management. Table 6 summarizes the 16 ongoing and completed but not published clinical trials involving radiotherapy for spinal and sacral chordoma that were listed on clinicaltrials.gov on the search completion date of 29 October 2021. The most common subject is an evaluation of efficacy and/or toxicity of proton therapy either alone or in combination with surgery. There are two studies incorporating PET imaging to identify hypoxic cells in target delineation for proton therapy. Although surgery remains the gold standard, SACRO is a randomized controlled trial that is currently accruing in Italy which is randomizing patients with sacral chordoma to definitive RT versus surgery. The results of this exploration will be critical given the high morbidity of en bloc sacrectomy, which may be avoided with definitive RT alone. There are three studies comparing outcomes of carbon ion therapy with proton therapy. Finally, three additional studies are exploring the addition of novel systemic therapies including nilotinib, nivolumab and brachyurea to radiotherapy. Taken together, this important compendium of studies will help advance the field in our understanding of the optimal radiation technique and help explore mechanisms to improve outcomes in this rare and aggressive malignancy. Patients who agree to be randomized will receive surgery vs. definitive RT (carbon ion radiotherapy, proton therapy, mixed photons-proton therapy). Those who do not agree to randomization will choose their modality.

Summary of Radiotherapy Recommendations
Overall consensus recommendations from the Spine Tumor Academy are shown in Table 7. Table 7. Spine Tumor Academy recommendation summary.

Level of Strength of Evidence Recommendation
The best chance of cure for mobile spine and sacral chordoma is in the upfront setting. As such, multi-disciplinary expert involvement at time of initial diagnosis is essential to optimizing patient outcomes

III Consensus
Target delineation should be performed on CT scans with at minimum a co-registered T2 weighted MRI. For patients treated in the adjuvant setting the pre-operative T2 weighted MRI should similarly be co-registered. In the adjuvant setting, a comprehensive discussion between the spine surgeon and radiation oncologist should occur to review intraoperative surgical findings and highlight regions believed to be at high risk of recurrence, which may not be obvious based on imaging alone. In the neoadjuvant setting, the discussion should include a review of the surgical plans and intentions to sacrifice or preserve specific nerves in the operating room so that the dosimetric parameters may be adjusted accordingly. Table 7. Cont.

Spine Tumor Academy Recommendations
Comprehensive target volumes that include regions of potential microscopic spread have superior local control to focal targets. For SBRT, target delineation according to the consensus contouring guidelines for solid tumor spinal metastases should be considered [63,64]. For proton and heavy ion therapy, comprehensive target delineation is based upon the Massachusetts General Hospital (MGH) Phase 2 data consisting of creation of a low-risk "microscopic" clinical target volume (CTV1) treated to a dose of 19.8-50.4 GyRBE (preoperatively) or 50.4 GyRBE). This is followed by a sequential boost to the high risk CTV2 to 70.2 GyRBE as defined by the original GTV (anatomically constrained) plus 5 mm. A further boost to gross residual disease without margin is performed after maximum safe resection and/or to the definitive GTV to 73. 19]. PTV is institution specific based upon robustness and range uncertainty analysis.

III Consensus
Although high-level data comparing outcomes comparing dose/fractionation regimens and treatment modalities are unavailable, dose escalation is critical in optimizing local control. Reasonable dose/fractionation schedules by treatment modality include the following:

III Consensus
When utilizing proton and heavy ion therapy, efforts must be made to limit the dose to the skin to less than 66 GyRBE in order to minimize the risk of long-term wound healing complications [3]. II Consensus

Limitations
Only 45 of 481 candidate citations met the inclusion criteria and were deemed eligible for inclusion in this systematic review. In addition, only two of the included studies were prospective in nature. Therefore, the preponderance of data driving these guidelines are taken from small retrospective studies that variably reported specific outcomes. As such, they suffer from challenges characteristic of single-institution and retrospective series including patients lost to follow-up, reporting bias, and selection bias. In addition, many of these studies have short follow-up periods of only a few years, which is particularly challenging given the protracted disease course of patients with chordoma. Specifically, it is unclear whether optimistic local control outcomes at short intervals will translate into similarly strong outcomes in the ensuing decade(s) that a chordoma patient would be predicted to live. Ultimately caution must be utilized when interpreting promising early results of outcomes from series with limited follow-up.
Furthermore, the included studies generally did not compare dose fractionation regimens, treatment modality of treatment, or timing for radiation therapy. Although we are able to draw conclusions that certain modalities may trend to higher (or lower) rates of certain toxicities and surmise that high doses of radiation are essential for local control of mobile spine and sacral chordoma, level I data comparing different approaches are unavailable. Therefore, although we present practice recommendations that have been heavily reviewed and discussed amongst multi-disciplinary experts in the international Spine Tumor Academy based on the best available data, larger-scale and multi-institutional studies (such as those under development by the Spine Tumor Academy) will be essential in optimizing patient outcomes in this locally aggressive malignancy. It should be noted that the literature search for this manuscript included primary research citations rather than previously published review articles. Nonetheless, a few additional reviews on this topic have been published in the last decade and are referenced here for the interested reader to access if desired [44,[65][66][67].

Conclusions
To conclude, multi-disciplinary expert involvement at the time of initial diagnosis of Kmobile spine and sacral chordoma is critical to optimizing patient outcomes. Although high-level data comparing outcomes, dose/fractionation regimens, and treatment modalities are unavailable, dose escalation is critical in optimizing local control. Target delineation should be performed using a CT scan with at minimum a co-registered T2 weighted MRI and should include a careful discussion between the spine surgeon and radiation oncologist. Comprehensive target volumes including sites of potential microscopic involvement improve local control compared with focal targets. Reasonable dose/fractionation schedules by treatment modality include 75.6-77.4 Gy RBE in 1.8-2 Gy RBE fractions using proton ± photon therapy; 24 Gy in a single fraction or 40-50 Gy in five fractions of SBRT; and at least 70.4 Gy in 2.2-4.4 Gy RBE fractions using carbon ion therapy. In addition, efforts must be made to limit skin dose when using proton therapy and heavy particles to minimize the risk of chronic wound healing complications. Level I and high-quality multiinstitutional data comparing treatment modalities, sequencing of radiation and surgery, and dose/fractionation schedules are needed to optimize patient outcomes in this locally aggressive malignancy.