Clinical manifestations

Presentation and course

Subacute (transient) myelopathy. The most common form of radiation-induced spinal cord injury is subacute (transient) myelopathy, which can occur following craniospinal axis irradiation for primary CNS tumors, or after "incidental" radiation of the cord for treatment of lymphoma or extraneural tumors of the head, neck, or thorax. The incidence of subacute radiation myelopathy varies, ranging from 4% of patients in a large series of patients irradiated for head and neck cancer (57) to upwards of 15% among patients receiving mantle field radiotherapy for Hodgkin lymphoma (103). The syndrome is more frequent among patients receiving a total spinal cord dose of greater than 5000 cGy, larger daily radiation dose fractions, or a larger volume spinal cord irradiated (20; 57).

Subacute myelopathy occurs after a latent period of 1 to 30 months following the completion of radiation, with the peak onset at 4 to 6 months (103; 32; 57). The syndrome consists of paresthesias or "electric shock" sensations radiating down the spine (Lhermitte phenomenon), which frequently extend down the limbs. These symptoms are often precipitated or worsened by neck flexion or physical exertion. Patients rarely report other symptoms and almost never show any objective signs of extremity weakness or sphincter dysfunction. The syndrome is thought to be due to temporary demyelination of the posterior columns of the spinal cord and is not thought to be a precursor of chronic progressive myelopathy. MR scans are unrevealing, although increased spinal cord metabolic activity with positive PET imaging has been reported (19). The syndrome typically resolves gradually over 1 to 9 months.

Chronic progressive (delayed) myelopathy. Delayed injury to the spinal cord typically presents as a chronic progressive myelopathy. Although rare, chronic progressive myelopathy is an ominous late complication of radiation. Unlike transient myelopathy, it is usually irreversible, and there is no treatment with established benefit. Chronic progressive radiation myelopathy usually presents with numbness or dysesthesias in the legs, followed by weakness and sphincter dysfunction, with an upper level of cord dysfunction ascending to lie within the irradiated area (66). Pain is usually not a prominent complaint. A Brown-Sequard hemicord pattern is fairly common early in the course (40).

The incidence of chronic progressive myelopathy appears to be bimodal, most commonly occurring after latent intervals of 12 to 14 months and 24 to 28 months, with a rough inverse correlation between radiation dose and the latent interval (88). Exceptional patients have latent intervals as short as 3 months or as long as 10 years after radiation (40; 66; 32). To date, most of the reported cases of chronic progressive myelopathy following stereotactic radiosurgery occurred within 12 months after treatment (84). In most patients, the neurologic deficits progress over several weeks to months and often plateau but rarely improve spontaneously.

Lower motor neuron syndrome. A rare syndrome of selective damage to lower motor neurons occurs following spinal radiation for medulloblastoma (80), lymphoma (25), or germ cell tumors (33; 37; 21). Patients develop roughly symmetric bilateral leg weakness beginning 4 to 14 months after completion of radiation. Examination shows muscle atrophy, fasciculations, normal or decreased muscle stretch reflexes, flexor plantar responses, and no sensory or sphincter involvement. Weakness is confined to the legs, even in patients who received irradiation to the entire spine (80), with little or no sensory loss or change in bladder and bowel function (21). The syndrome usually progresses slowly over several months, and then stabilizes, but does not improve. Motor and sensory nerve conduction velocities are normal, and electromyographic evidence of diffuse denervation, including lumbar paraspinal muscles, has been demonstrated (37). In some published cases, spine MR scans are normal (15), but others have reported gadolinium enhancement of the cauda equina (05). Lumbosacral anterior horn cells were initially believed to be the primary site of damage in these patients, but it is now thought to be more consistent with a radiculopathy rather than a neuronopathy (05).

Subacute radiation myelopathy, chronic progressive radiation myelopathy, and lower motor neuron syndrome are summarized in Table 1.

Table 1. Types of Radiation Myelopathy

Prognosis and complications

For patients with subacute myelopathy, symptoms tend to gradually resolve without medical intervention over 1 to 9 months. For patients with chronic progressive radiation myelopathy, the neurologic deficits progress over weeks to months in a steady or (less commonly) stepwise fashion, leading to paraplegia or quadriplegia in more than half of patients. A few patients show slow and partial spontaneous recovery (49), but it is rare for ambulation to be regained once it is lost. Patients with severe neurologic deficits frequently have shortened survival due to pneumonia, pulmonary embolus, or other medical complications (91).

In the handful of reported cases of radiation myelopathy following spinal radiosurgery, the latent interval to onset of neurologic symptoms was shorter, and the likelihood of subsequent neurologic improvement was higher, than is generally seen in myelopathy after standard fractionated external beam radiation (29). It is possible that the pathophysiology of spinal cord injury may differ according to the type of radiation administered. For lower motor neuron syndrome, bilateral lower extremity weakness progresses slowly over several months and then stabilizes, with most patients remaining ambulatory.

Clinical vignette

This is a fictitious clinical vignette of a patient developing chronic progressive radiation myelopathy.

A 70-year-old man with prostate adenocarcinoma developed diffuse back pain without other neurologic symptoms and was found to have multiple vertebral metastases without epidural tumor extension. He received radiotherapy from the T1 through T8 vertebral bodies to a total dose of 3500 cGy given in daily fractions of 250 cGy. His pain improved, and he felt well until 7 months after completion of radiotherapy when he developed painless, progressive, bilateral leg weakness and numbness. He denied change in bowel or bladder function. Examination 6 weeks after onset of symptoms showed the following: 3+/5 to 4/5 weakness in both legs, worse on the left; hyperreflexia of the left leg and a left Babinski sign; a sensory level to pinprick on the right trunk at T10 (without sacral sparing); impaired temperature sensation in the same distribution on the right; and decreased vibration and position sense on the left leg and foot. MR scan showed abnormal signal on T2-weighted images in the spinal cord from C7 through T8, without cord widening. There was patchy linear enhancement in the spinal cord from T4 through T7. The patient was given dexamethasone 16 mg per day and noted subjective improvement in gait stability, without any other change in symptoms or signs.

Biological basis

Etiology and pathogenesis

As no therapeutically effective radiotherapy regimen is known that carries zero risk of spinal cord injury, careful treatment planning and administration of radiotherapy with particular regard for spinal cord tolerance parameters is necessary to reduce the risk of radiation myelopathy. The risk of developing chronic progressive radiation myelopathy is largely, but not entirely, dependent on the total dose and size of the dose fraction of radiation. There are several formulas for calculating a "dose equivalent" that allow comparison of the radiobiological effect and risk of toxicity among different fractionation schedules. These formulas contain a correction factor to account for the fact that the spinal cord is more sensitive to the size of the daily dose fraction than are many other tissues. The most commonly used model for describing the response of the spinal cord and other "late reacting" tissues to once-daily radiation is the linear quadratic formula (23; 104; 50; 71). Radiosensitivity of tissue in this model is described by the alpha-beta ratio, with lower values corresponding to lower radiosensitivity. The practical clinical usefulness of such formulas has been questioned (99) as these formulas do not accurately predict the risk of radiation myelopathy following hyperfractionated (more than once daily) radiation schedules (104; 42; 71; 87), nor do they accurately predict spinal cord tolerance to stereotactic radiosurgery (46; 13). Furthermore, radiation myelopathy has occurred after "safe" radiation doses in combination with high-dose or intensive chemotherapy regimens, such as those used for hematopoietic stem cell transplantation (36; 10; 92; 17; 93; 27), as well as following immunotherapy administration after radiation for non-small cell lung cancer (08; 48). The risk of delayed radiation myelopathy as a function of radiation dose needs to be considered separately for patients who: (1) are previously untreated and receive conventional fractionated external-beam radiation to the entire cross-section of the spinal cord, (2) receive a second course of fractionated external-beam radiation to the entire cross-section of the spinal cord, or (3) receive stereotactic radiosurgery with high-dose exposure to a partial cross-section of the spinal cord (46).

Conventionally fractionated radiotherapy in previously untreated patients. For patients receiving a single course of fractionated radiation, the dose-risk relationship differs between the cervical and thoracic spinal cord due to differences in the radiosensitivity. Reirradiation doses are typically expressed in equivalent dose in 2 Gy fractions (EQD2), which accounts for the total dose, the dose per fraction, and the radiosensitivity of the tissue expressed using alpha-beta ratios. When radiation is given in once-daily fractions and dosing regimens are analyzed according to the equivalent dose at 200 Gy daily fractions, the risk of delayed cervical myelopathy is approximately 0.03% after 45 Gy, 0.2% at 50 Gy, 5% at 60 Gy, and 50% at 69 Gy (53; 52; 18; 41; 01; 87; 46). The thoracic spinal cord is generally less sensitive to radiation injury than the cervical cord; the risk of chronic progressive myelopathy increases to 30% to 40% at doses greater than 65 Gy but without a steep dose-response curve at higher doses as is the case for cervical myelopathy (01; 87). Unexplained differences in individual sensitivity to radiation result in the occurrence of chronic progressive radiation myelopathy in an extremely small percentage of patients who receive a "safe" radiation regimen (107; 105). Additionally, in the cervical spinal cord, the risk of chronic progressive radiation myelopathy increases with increasing lengths of spinal cord irradiated (01). For the thoracic spinal cord, published analyses do not indicate a definite relationship between irradiated rostral-caudal spinal cord volume and the risk of chronic progressive myelopathy (52; 42; 71; 99; 01; 87; 46).

Conventionally fractionated radiotherapy in previously treated patients. Some patients with primary head and neck tumors, lung cancer, or vertebral metastases receive second courses of radiation, exposing the spinal cord to total doses up to 9000 cGy or more (102; 85; 77; 35). These patients have an increased but not easily quantified risk of chronic progressive radiation myelopathy. Evaluating the risk of radiation myelopathy requires consideration of the radiation dose regimen, spinal cord volume, irradiated region for each course of radiation, and the time interval between radiation courses. Experimental animal studies and clinical data suggest that the spinal cord recovers from "occult" radiation injury over approximately 2 to 3 years, so palliative reirradiation may be given without a terribly high risk of radiation myelopathy during a patient’s lifetime (60; 61; 70; 46). Updated results from a bi-institutional study with an expanded cohort suggests that patients treated to higher cumulative doses or at higher risk of myelopathy based on prior risk scores remain free of myelopathy; however, the median follow-up was only 12 months (16).

Alternative models for estimating the retreatment radiation tolerance of the spinal cord at different points in time following an initial dose of radiation have been investigated (43; 101). These models utilize primate data (02) to fit parameters as well as incorporate other clinical and preclinical observations, such as the observation that in rodent models the recovery of the spinal cord after initial radiation dose begins around 70 days later (97; 100). The authors provide detailed sample calculations for how to utilize their models clinically, including an example in which a patient who initially received 40 Gy in 2 Gy fractions to the spinal cord could receive reirradiation 2 years later with approximately 36 Gy in 3 Gy fractions and have the same risk of developing radiation myelopathy (0.12%) with retreatment as with the initial treatment.

Stereotactic body radiotherapy (SBRT). Single-dose or fractionated stereotactic radiosurgery is increasingly used to treat metastatic vertebral tumors (with or without epidural extension) (82) or primary tumors such as meningioma, schwannoma, or hemangioblastoma (13). Many patients with spinal metastases treated with stereotactic radiosurgery have previously received standard fractionated radiation to the same area. The radiobiology of this high-dose focused radiation differs from that of conventional fractionated external beam radiotherapy, both in terms of the tumor effect and on the risk of injury to the adjacent spinal cord. The steep dose fall-off within and near the spinal radiosurgery target volume means that there is heterogeneous radiation exposure to the spinal cord and a possibility that a small volume of spinal cord receives a high radiation dose. Determining the risk of radiation myelopathy after spinal radiosurgery, therefore, needs to take this "partial volume effect" into account (78; 82; 13). The reported risk of radiation myelopathy after stereotactic radiosurgery for spinal metastases is generally less than 1% (29; 46; 84), though it is somewhat higher in some reports (26). Variability in re-irradiation doses at the time of SBRT can complicate interpretation of radiation myelopathy rates. Ito and colleagues reported myelopathy rates in 123 patients (133 lesions), with 40% of lesions having a Bilsky grade of 2 to 3 (38). Patients underwent re-irradiation with a uniform dose of 24 Gy in two fractions using PRV cord constraint of less than 12.2 Gy or less than 11 Gy if patients’ prior course of radiotherapy was more than 50 Gy EQD2. At a median follow-up of 12 months (range 1 to 57 months) four patients developed myelopathy (3%) ranging from 5 to 37 months after SBRT. There are reports of chronic progressive radiation myelopathy occurring in patients who receive stereotactic radiosurgery after prior conventional fractionated radiation (83). Chronic progressive radiation myelopathy occurred in 1 of 17 patients with spinal cord hemangioblastoma treated with single-dose or hypofractionated stereotactic radiosurgery (13). Overall, the rates of myelopathy in both the de novo and re-irradiation setting with SBRT are low, and risk of myelopathy must be weighed against complications from tumor progression. Adherence to dose constraints and technical aspects of SBRT is critical (65).

There are emerging data to estimate the risk of delayed radiation myelopathy based on the spinal cord dose and volume parameters for single-fraction and fractionated stereotactic radiosurgery (84; 34). There is possibly an influence of prior fractionated radiation therapy, or concurrent chemotherapy, or both on the development of radiation myelopathy in these patients. Although this is not clearly known, reported factors associated with a lower risk of radiation myelopathy for reirradiation patients include minimum time interval to reirradiation of 5 months or more and cumulative thecal sac equivalent dose in 2 Gy fractions with an alpha/beta of 2 Dmax ≤ 70 Gy (81). For spine SBRT in patients without a history of prior spinal radiotherapy, consensus guidelines suggest limiting single fraction maximum point dose (Dmax) to 1000 to 1400 Gy (with 1000 cGy being delivered to less than 10% of the spinal cord), or 5 fraction Dmax of 2500-3000 cGy (73). Per the Hypofractionation Treatment Effects in the Clinic (HyTEC) report, for de novo SBRT delivered in 1 to 5 fractions the following spinal cord Dmax are estimated to be associated with a 1% to 5% risk of radiation myelopathy: 12.4 to 14.0 Gy in 1 fraction, 17.0 Gy in 2 fractions, 20.3 Gy in 3 fractions, 23.0 Gy in 4 fractions, and 25.3 Gy in 5 fractions (81). Corresponding estimated risk calculations for developing radiation myelopathy based on the Stanford spine SBRT series have been developed to assist predicting the risk of developing radiation myelopathy (34). For reirradiation SBRT in 1 to 5 fractions, lower risk of myelopathy is reported for cumulative thecal sac dose of less than 70 Gy in EQD2; SBRT thecal sac Dmax less than 25 Gy, thecal sac SBRT EQD2 Dmax to cumulative EQD2 Dmax ratio of less than 0.5, and a minimum time interval to reirradiation of longer than 5 months (81). With improvements in systemic therapies and increases in survival, there is concern that rates of radiation myelopathy may increase. In a prospectively maintained database, Zeng and colleagues evaluated the risk of late toxicity for patients receiving spine SBRT whose survival was more than 3 years (106). Among those who lived more than 3 years, only 10% had prior radiotherapy. The majority (96%) of patients received 20 to 28 Gy in two fractions. The 5-year plexopathy rate was 5.1%. These data are reassuring and highlight the importance of proper technique when delivering ablative radiotherapy. Though beyond the scope of this article, it should be noted that MRI or CT myelogram are critical for accurate delineation of the spinal cord and thecal sac, which is absolutely necessary for delivery of SBRT along with image guidance during treatment.

Pathologic changes in radiation myelopathy. Subacute radiation myelopathy is generally believed to be caused by reversible demyelination of the posterior columns. This is supported by animal studies, but because of the self-limited nature of subacute radiation myelopathy, no published human pathology exists. From evidence of autopsied patients and from animal studies, chronic progressive radiation myelopathy appears to be the result of a combination of damage to small blood vessels and direct injury to glia, myelin, and inflammatory reactions (89; 62; 63; 39). The histopathology of chronic progressive radiation myelopathy is somewhat variable, both between patients and in different areas of the spinal cord in the same patient (40; 06; 66). White matter is characteristically more severely affected than gray matter, with a predilection for the posterior and superficial lateral columns. Coalescing foci of demyelination and axonal degeneration in these areas are accompanied by Wallerian degeneration above and below the necrotic zones. In the most severe cases, areas of total coagulative necrosis occur. In cases where gray matter is involved, the neurons may show central chromatolysis, but are often remarkably preserved. A spectrum of severity occurs with changes in small blood vessels, including fibrinoid necrosis of the vessel walls, hyaline thickening and obliteration of lumens, telangiectasias, extravasation of hyaline material, and occasional perivascular lymphocytic cuffing. In some autopsied cases, the severity of vascular changes appears to be mild compared to the degree of demyelination and parenchymal necrosis. The nerve roots and anterior or posterior spinal arteries are not involved, even adjacent to areas of extensive cord necrosis. Some evidence exists stipulating that direct damage to glia and myelin predominates in patients with relatively short latent intervals, whereas patients with longer latent intervals show more prominent vascular changes, although a considerable variability and overlap among patients does occur (74; 90; 49). The pathologic changes associated with lower motor neuron syndrome suggest a radiation-induced vasculopathy of the proximal spinal roots, with preservation of lower motor neuronal cell bodies and spinal cord architecture.

Epidemiology

Most patients who develop chronic progressive radiation myelopathy received "incidental" radiation of the spinal cord as part of treatment for extraneural primary carcinomas of the lung, esophagus, head, and neck, or for lymphoid tumors. Less common, are patients in whom the spinal cord itself is targeted either in treatment of a glioma or as part of craniospinal axis radiation of medulloblastoma. The reported incidence figures for severe chronic progressive radiation myelopathy range from 1% to 12% (32; 53; 52; 41) but may be misleading because many patients do not survive long enough to be at risk for developing this complication. As systemic treatments continue to improve and patients live longer with their cancer, the incidence of spinal metastases (28), palliative radiotherapeutic treatment of spinal metastases, and the development of radiation myelopathy will likely all increase.

Prevention

Meticulous treatment planning and administration of radiotherapy within the broad parameters of spinal cord tolerance reduces the risk of radiation myelopathy (46; 73), but no therapeutically effective radiotherapy regimen is known that carries zero risk of spinal cord injury. There are no approved drugs to decrease the risk of radiation myelopathy; however, animal models suggest the use of angiotensin-converting enzyme inhibitors (ACEi) may reduce the risk of radiation myelopathy (79).

Differential diagnosis

The differential diagnosis of radiation myelopathy includes recurrent or metastatic tumor, new primary spinal neoplasm occurrence (76; 59), symptomatic spinal cord cavernous malformations from prior craniospinal axis irradiation (51; 58), neurotoxicity of chemotherapy, and remote paraneoplastic effects of cancer on the spinal cord. The distinguishing features between the presentations of these disorders are briefly described below. In addition, non-treatment-related and non-cancer-related autoimmune, infectious, or toxic-metabolic etiologies for spinal cord injury can be considered. The history should help guide the likelihood of these etiologies.

Recurrent, metastatic, or new primary spinal neoplasm occurrence. Epidural spinal metastases produce early and prominent pain in upwards of 90% of patients and should easily be diagnosed by MR imaging. A more difficult differential diagnosis is radiation myelopathy versus intramedullary spinal cord metastasis, although most patients with intramedullary metastases have a primary lung carcinoma (44; 14; 95). Patients present with a combination of pain, weakness, sensory loss, and urinary incontinence, with up to 50% of patients presenting with Brown-Sequard syndrome. Any level of the cord may be affected. Among patients irradiated for primary spinal cord gliomas, no clinical or neuroimaging features reliably distinguish chronic progressive radiation myelopathy from recurrent tumor (105; 67).

Neurotoxicity of chemotherapy. Acute myelopathy is a rare, but well-recognized complication of intrathecal administration of methotrexate or cytarabine in patients with leukemia, lymphoma, or a variety of solid tumors. Myelopathy may occur even with palliative doses of radiotherapy in the setting of intrathecal chemotherapy (24). The latency is extremely variable, from within a few hours of the first intrathecal chemotherapy injection to several months after completion of a course of several injections. No clear relationship to the cumulative dose of intrathecal chemotherapy exists (75). The syndrome can occur with or without concomitant systemic chemotherapy or cord irradiation (72). Patients typically develop a rapidly ascending flaccid paraparesis with a sensory level and bowel or bladder involvement, occasionally with early back and leg pain. MR scans in a few reported patients show focal cord swelling, decreased signal intensity on T1-weighted images, or patchy cord enhancement.

Paraneoplastic disorders. There are several ways in which paraneoplastic disorders can affect the spinal cord. Patients with paraneoplastic encephalomyelitis associated with small cell lung carcinoma, thymoma, or other neoplasms may present with a predominant subacute myelopathy syndrome (22). Some of these patients have myelopathy with rigidity, myeloradiculopathy, or myelopathy plus optic neuritis (11; 68; 45). Spine MRI scans may show localized or more longitudinally extensive T2-weighted signal abnormalities (22). Associated antineuronal autoantibodies include anti-Hu, anti-CV2, and anti-amphiphysin.

A syndrome of muscle rigidity and spasms that clinically resembles “stiff-person syndrome” is associated with a variety of neoplasms, including small cell lung carcinoma and breast carcinoma. Rigidity is probably caused by multifocal encephalomyelitis affecting the spinal cord, or brainstem, or both. Patients develop progressive aching and rigidity of the axial and proximal limb musculature, usually asymmetric at onset. There are superimposed painful and sometimes violent spasms, either occurring spontaneously or triggered by voluntary movement, passive movement, or sensory stimuli. There is clinical overlap with a syndrome of "encephalomyelitis with rigidity and myoclonus." Some patients have antibodies against the synaptic vesicle-associated protein amphiphysin (68), or against glycine receptors (54).

Paraneoplastic necrotizing myelopathy is a rare complication of a variety of carcinomas, lymphomas, and leukemias, without a clear preponderance of any specific tumor type (64). Several cases have occurred after apparent cure of the associated tumor. Patients present with subacute painless bilateral weakness, sensory symptoms, and sphincter dysfunction. Physical examination may initially demonstrate a Brown-Sequard syndrome, but more commonly reveals a specific level of transverse spinal cord dysfunction. Nearly all patients suffer rapid deterioration of function and a progressively ascending level of flaccid paralysis and numbness.

Some patients with small cell lung carcinoma and multifocal paraneoplastic encephalomyelitis develop focal or diffuse lower motor neuron weakness and atrophy. In these patients, autopsy demonstrates patchy loss of anterior horn cells and variable inflammatory infiltrates within the spinal cord (12). As with other manifestations of paraneoplastic encephalomyelitis, the motor neuron involvement in these patients tends to progress independently of the status of the underlying neoplasm.

An unusual syndrome of "subacute motor neuronopathy" occurs in patients with Hodgkin disease, non-Hodgkin lymphomas, and thymoma (86). Patients develop subacute progressive weakness, often asymmetric or patchy and predominantly involving the legs, without pain or significant sensory loss. Examination reveals a pure lower motor neuron picture with moderate to severe weakness, fasciculations, atrophy, and diminished deep tendon reflexes. CSF is acellular, but occasionally shows mildly elevated protein. The syndrome may occur prior to the discovery of a neoplasm, as well as after attainment of complete remission. The neurologic deficits generally worsen over weeks to months independent of the course of the malignancy, but usually spontaneously stabilize or improve (sometimes dramatically) after a period of months to years. Autopsy shows patchy degeneration and loss of anterior horn cells, occasional inflammatory infiltrates, secondary thinning of ventral nerve roots, and widespread patchy segmental demyelination of spinal roots and brachial and lumbosacral plexuses.

Diagnostic workup

The CSF in delayed radiation myelopathy is usually normal, but may contain slightly elevated protein or a mild pleocytosis.

MRI scans in the majority of patients with chronic progressive radiation myelopathy show widening of the affected cord and abnormally high signal intensity on T2-weighted images (56; 98; 55; 47). Most patients have abnormal intramedullary contrast enhancement, in either a streaky or, less commonly, a ring-enhancing pattern. The MRI abnormalities often extend beyond the radiation ports and generally persist for several months, followed by spinal cord atrophy in long-term survivors. The MRI abnormalities of radiation myelopathy cannot be absolutely distinguished from those of recurrent spinal cord glioma, intramedullary spinal cord metastasis, or paraneoplastic myelopathy. Review of the diagnostic MRI by the radiation oncologist and comparison to radiotherapy plans may aid in determining if the signal changes on MRI are in a region of high dose and, thus, more likely to indicate radiation myelopathy.

Management

Treatment options for patients with severe radiation myelopathy are severely limited. Some patients show stabilization or partial improvement on corticosteroids (31; 56; 96). In case reports of radiation myelopathy following immunotherapy administration, pausing the immunotherapy and starting corticosteroids resulted in improvement in neurologic symptoms, with rechallenge reportedly without symptom recurrence with pembrolizumab (08; 48). Anecdotal reports are available of partial improvement following anticoagulation with heparin and warfarin (30) or hyperbaric oxygen (03; 07). Bevacizumab, a monoclonal antibody against vascular endothelial growth factor, which is often beneficial in treating patients with cerebral radiation necrosis, may also provide some benefit for patients with radiation myelopathy (09; 69; 94). In a small, retrospective case study of four patients treated with bevacizumab after their chronic progressive myelopathy progressed on corticosteroids, all four patients had improvement of their radiologic findings, and three patients had clinical stabilization of their disease (69). The rarity of radiation myelopathy means that all of the anecdotal reports of treatment lack a "denominator" to estimate the likelihood of a meaningful response. A reporting bias in favor of "responders" is probable.