Dorsal Column Myelopathy after Intrathecal Chemotherapy for Leukemia

Abstract

Intrathecal chemotherapy with methotrexate, a folate antagonist, is widely used to treat central nervous system malignancies. The mechanisms underlying methotrexate-induced neurotoxicity are unclear but may be related to increased homocysteine levels. Intrathecal methotrexate-induced myelopathy mimicking sub-acute combined degeneration, with normal B12 levels, has been documented. We examined treatment and magnetic resonance imaging (MRI) characteristics of 13 patients with leukemia who received intrathecal methotrexate and developed urinary and bowel incontinence, ascending motor weakness, and sensory loss with dorsal column hyper-intensity on MRI between 2000 and 2016. Cerebrospinal fluid evaluation was negative for leukemia in all patients and positive for elevated protein in 12 patients. Seven of 8 patients with available data had reduced serum folate, increased serum homocysteine, or both, implicating methotrexate as the cause of neurotoxicity. Autopsy of one patient revealed loss of myelinated axons in the posterior columns. These findings suggest that methotrexate neurotoxicity may be mediated by folate antagonism. Awareness and a high index of suspicion of these characteristic clinical and radiographic features in patients who develop myelopathy after intrathecal methotrexate may help to avoid additional neurotoxic therapy in such patients.

Introduction

The combination of systemic chemotherapy and aggressive prophylaxis for central nervous system (CNS) disease represents the cornerstone of therapy for acute leukemias.¹ The antimetabolite, folate antagonist methotrexate is a central component of treatment; however, a well-known risk of intrathecal or intravenous methotrexate is acute, sub-acute and late neurotoxicity.² The most commonly reported type of methotrexate-induced neurotoxicity is leukoencephalopathy, which can be acute or late and transient or permanent and is visualized on T2-weighted magnetic resonance imaging (MRI) as abnormal hyper-intensities in the deep white matter of the brain.^{3, 4} Reports of myelopathy after methotrexate therapy, however, have been rare. We previously published a series based on 7 leukemia patients who developed intrathecal chemotherapy induced myelopathy at our institution between 2010 and 2014⁵. Several case reports have documented myelopathy in patients with leukemia characterized on MRI by T2 signal abnormalities in the dorsal horns; potential etiologies include intrathecal methotrexate, intrathecal cytarabine, nelarabine, low serum concentrations of vitamin B12, folate, or copper, and immune dysfunction.^6-15 This MRI finding is pathognomonic for sub-acute combined degeneration, a rare neurologic consequence of vitamin B12 or folate deficiency that results in loss of vibration, fine touch, and position sense and progresses to lower extremity anesthesias and paraplegia.¹⁶ The similarities in the clinical and radiographic presentation between methotrexate myelopathy and sub-acute combined degeneration suggest a shared pathophysiology. To investigate this possibility, we identified 13 patients with irreversible myelopathy after receipt of CNS-directed therapy for leukemia who were found to have MRI T2 abnormalities in the dorsal columns. We evaluated clinical and radiographic findings in those patients.

Materials and Methods

With the approval of our institutional review board, we used a tool for fast analytic search of all electronic MRI reports of patients presenting at The University of Texas MD Anderson Cancer Center with symptoms of myelopathy after having received methotrexate therapy for leukemia between January 1, 2000 and March 3, 2016.¹⁷ We identified 13 patients with T2 hyper-intensity primarily affecting the dorsal columns who also had myelopathy. A skilled neuro-radiologist (LC) reviewed each case. We reviewed the clinical features, treatment characteristics, laboratory data, and MRI findings for each case.

Postmortem examination of the CNS was performed in one of the patients in this series. The brain and spinal cord were serially examined in thin 5 millimeter horizontal sections and submitted for histologic analysis. Automated Hematoxylin and Eosin (H&E) staining of 4 micron thick sections was performed using a staining platform. A special stain for Luxol fast blue (LFB) and Periodic acid-Schiff (PAS) were performed on sections of cervical, upper thoracic and lumbar cord. An immunostain for neurofilament (Invitrogen, clone FNP7/DA2/RMdO20.11, 1:200 dilution) was performed on a 4 micron –thick section of cervical spinal cord.

Results

Patient and treatment characteristics for the 8 male and 5 female patients that developed dorsal column myelopathy are listed in Table 1. Two patients had received primary treatment at outside institutions and presented to MD Anderson with myelopathy. All patients had received intrathecal methotrexate for the treatment of leukemia, either as single-agent therapy or in combination with cytarabine and a steroid (46%). The median number of intrathecal treatments was 17 (range 3–36).

Table 1. Clinical and toxicity characteristics for 13 patients with dorsal column myelopathy.

Characteristic	Value or No. of Patients (%)
Age, years
Median(range)	34 (13–64)
Sex
Female	5 (39)
Tumor histology
B-cell ALL	8 (61)
T cell ALL	2 (15)
CML	1 (8)
AML	1 (8)
Blasticplasmacytoid dendritic cell	1 (8)
History of CNS leukemia
Yes	8 (61)
Induction chemotherapy regimen
Hyper-CVAD	7 (54)
Other	6 (46)
History of relapsed disease before myelopathy
Yes	12 (92)
Salvage therapy
Yes	13 (100)
Median no. regimens (range)	3 (1–11)
Intrathecal chemotherapy
Yes	13 (100)
Single Agent Cytarabine
Yes	12 (92)
Single Agent Methotrexate
Yes	12 (92)
Triple therapy (mtx, araC, steroid)
Yes	6 (46)
No. of intrathecal treatments, median (range)
All	17 (3–36)
Methotrexate	6 (0–14)
Cytarabine	10 (0–25)
Triple	0 (0–8)
Receipt of nelarabine
Yes	2 (15)
History of radiation before symptoms
Yes	6 (46)
Radiation target
Total body	3
Mediastinum	1
Whole brain	2
Chest wall	1
Neck	2
Testicle	1
History of allo SCT before symptoms
Yes	4 (31)
CSF Analytes
Cytology positive at time of symptoms
No	13 (100)
Protein at time of abnormal MRI(normal= 15-55 mg/dl)
Elevated	12 (92)
Not assessed	1 (8)
Median, mg/dL (range)	95.5 (58-453)
Myelin basic protein (normal=0–5.5 ng/mL)
Elevated	4 (31)
Not assessed	9 (69)
Median, ng/dL (range)	342.6 (34.1–481)
Serum Analytes
Folate (normal=7.3–26.1 ng/mL)
Normal	2 (15)
Decreased	6 (46)
Not assessed	5 (39)
Median, ng/mL (range)	4.6 (2.2–10.3)
Vitamin B12 (normal=211–946 pg/mL)
Normal	9 (69)
Decreased	1 (8)
Not assessed	3 (23)
Median, pg/mL (range)	329.5 (185–549)
Homocysteine (normal=5–12 umol/L)
Elevated	4 (31)
Not assessed	9 (69)
Median, μmol/L (range)	21.65 (13.9–37.7)
Methylmalonicacid (normal= <0.40 nmol/mL)
Normal	6 (46)
Not assessed	7 (54)
Median, nmol/mL (range)	0.25 (0.09–0.28)
Copper (normal=0.75-1.45 μg/mL)
Normal	3 (23)
Decreased	1 (8)
Elevated	1 (8)
Not assessed	8 (61)
Median, μg/mL (range)	1.23 (0.64–1.48)
MRI T2 dorsal hyper-intensity
Yes	13 (100)
MRI reported normal at time of initial symptoms
Yes	7 (54)
CNS-directed therapy given after initial symptoms
Yes	11 (85)

Abbreviations: ALL, acute lymphoblastic leukemia; CML, chronic myelogenous leukemia; AML, acute myelogenous leukemia; CNS, central nervous system; CVAD, cyclophosphamide, vincristine, doxorubicin, and dexamethasone; mtx, methotrexate; araC, cytarabine; allo SCT, allogeneic stem-cell transplantation; CSF, cerebrospinal fluid; MRI, magnetic resonance imaging.

T2 dorsal cord hyper-intensity (Figure 1) was noted in all patients. Initial MRI scans obtained when symptoms arose had been reported as normal in 7 patients (54%); re-review of these scans when subsequent scans showed abnormalities revealed that subtle T2 changes had been present in 4 of those cases. Axial T2 images are not obtained routinely at our institution; rather, only sagittal T2 images are acquired. In several patients whose initial MRI studies had been considered “normal,” subtle T2 changes had not been apparent on the sagittal T2 images.

Figure 1.

A T2 hyper-intense lesion involving the posterior columns of the spinal cord on (a) axial and (b) sagittal T2 weighted magnetic resonance images in a patient with dorsal column myelopathy.

Details of presenting neurologic symptoms were available for 12 patients. In all 12 cases, patients presented with ascending lower extremity parasthesias, urinary and/or bowel incontinence and progressive lower extremity weakness. Saddle anesthesia was reported in 7 of 12 patients. In one patient detailed information regarding presenting neurologic deficits was not available. On examination all 13 patients were noted to have areflexia and paraplegia. The median number of days between the last intrathecal methotrexate administration and onset of symptoms was 15 (range 3-60 days).

Cerebrospinal fluid (CSF) analyses done when patients presented with myelopathy showed no evidence of leukemia in any of the 13 patients, and 12 of 12 evaluated for CSF protein were elevated. CSF levels of myelin basic protein, a myelin sheath marker that is elevated after trauma or active demyelination or in patients with leukemia and CNS toxicity,^18,19 were significantly elevated in all 4 patients assessed for this marker.

Serum folate levels were measured in 8 patients (62%) when symptoms arose and were below normal in 6 patients and normal in 2. Serum vitamin B12 levels, measured in 10 patients, were normal in 9. Elevated levels of homocysteine, an amino acid that requires folic acid for conversion to methionine,²⁰ have been implicated in vascular disease²¹ and in methotrexate neurotoxicity.^{20, 22,23} Deficiencies in vitamin B12 or folic acid can lead to increased homocysteine levels, whereas vitamin B12 deficiency alone leads to elevation of serum methylmalonic acid.²⁴ Serum homocysteine levels were measured in 4 patients before vitamin B12 and folate repletion, and were elevated in all 4; levels of methylmalonic acid, however, were normal, suggesting a deficiency in folate, but not in vitamin B12, in these patients.

Although we attempted repletion with B12 and folate in this patient population, unfortunately it did not result in neurologic improvement. Another approach for reducing symptoms of methotrexate-induced CNS toxicity is the administration of dextromethorphan, a non-competitive antagonist of N-methyl-D-aspartate receptors that are excessively activated by homocysteine and other excitotoxic amino acids.²⁵ Two of the patients in our series received dextromethorphan, but without significant durable improvement.

Eleven patients in our study (85%) had additional CNS treatment after the initial appearance of myelopathy because of concern for leukemia infiltration; two of these patients had received radiation therapy, one to the craniospinal axis and the other to T10 through the sacrum. Postmortem examination of the latter patient revealed degeneration in the dorsal columns in the un-irradiated cervical spinal cord, most prominently within the fasciculus gracilis, an ascending tract carrying sensory fibers from the lower thoracic, lumbar, and sacral levels to the brain (Figure 2a). The fasciculus cutaneous, which carries sensory fibers from the cervical and upper thoracic levels, was largely spared, which was consistent with findings on physical examination (normal upper extremities and normal sensory level above T11). Periodic acid Schiff staining (for neurons) revealed axonal loss of the fasciculus gracilis. Luxol fast blue staining (for myelin) revealed persistent myelin in the residual axons within the fasciculus gracilis, suggesting that the primary process was axonal loss and not demyelination (Figure 2b). Prominent necrosis and parenchymal hemorrhage were noted in the irradiated lower thoracic and sacral cord (Figure 2c). No leukemic infiltrates were identified.

Figure 2.

Post-mortem Spinal Cord Pathology in a Patient with Dorsal Column Myelopathy who received Radiation to T10 through the sacrum prior to recognition of the etiology of paraplegia.

A) Cervical Spinal Cord H&E. This image is a lower power (4×) H&E view of the unirradiated spinal cord at the cervical level. The center of the image shows the left and right fasciculus gracilis. The posterior median septum is the shallow groove centrally. The posterior intermediate sulcus that divides the fasciculus gracilis and fasciculus cuneatus is seen on both sides with exaggeration on the right. The neuropil is vacuolated in both tracts and is most striking in the fasciculus gracilis bilaterally. The fasciculus cuneatus of either side is lesser involved.

B) Cervical Spinal Cord. A Luxol fast blue staining (LFB) for myelin and Periodic acid-Schiff (PAS) for neurons shows the vacuolization due to axonal loss of the fasciculus gracilis. The blue color of myelinated fibers is retained suggesting the primary pathologic process is axonal loss, not demyelination (20×).

C) Lower Thoracic Spinal Cord. This H&E image of the lower thoracic cord at the level of T11 that received radiation shows dead neurons in the ventral gray horn as indicated by their pale eosiniophilic cytoplasm, nuclear shrinkage and pyknosis (20×)

Discussion

Patients with hematologic malignancies, especially leukemia, are at increased risk of neurologic toxicity due to sequelae of central nervous system (CNS) directed therapy, direct cancer effects or from immune dysfunction. The etiology behind neurotoxicity is not always clear, as these factors are often all present in affected patients. There are an increasing number of case reports that describe patients with a distinctive radiographic and clinical pattern of irreversible, severe myelopathy that has a predilection for dorsal columns. Several etiologies have been proposed, including immune dysfunction, intrathecal methotrexate and/or cytarabine, nutritional deficiencies such as decreased vitamin B12, folate or copper, as well as other chemotherapeutic agents such as nelarabine.^6-15 This MRI finding is diagnostic for sub-acute combined degeneration, a rare neurologic complication of vitamin B12, and less frequently folate deficiency^{16, 26}. Degeneration of the posterior and lateral columns manifests as sensory ataxia and ultimately progresses to spastic paraplegia. The recognition in the similarities in the clinical and radiographic presentation between methotrexate myelopathy and sub-acute combined degeneration has resulted in the postulation of a common pathophysiologic mechanism between the two entities.

Plasma homocysteine levels are increased among patients treated with methotrexate^{23, 27}. In 1997 Quinn et al. hypothesized that excess homocysteine may partially mediate methotrexate neurotoxicity²⁰. In a study of 23 pediatric cancer patients that received methotrexate and 34 control patients who did not, significantly higher levels of CSF homocysteine were present among the methotrexate treated children. Furthermore, patients with neurotoxicity at the time of CSF collection had the highest homocysteine levels. Similar findings of elevated homocysteine correlating with neurotoxicity after methotrexate have been reported based on serum homocysteine levels²². Among 53 pediatric patients with acute lymphoblastic leukemia (ALL) that received high dose methotrexate, the serum homocysteine levels were higher after methotrexate administration than before. Furthermore, patients that experienced seizures tended to have higher plasma homocysteine levels. Although serum homocysteine data was only available for 4 patients in our series, it was significantly elevated in all 4 patients. Interestingly, in a case report of a 41 year old male with T-cell ALL that received both systemic and intrathecal methotrexate and developed irreversible myelopathy characterized by T2 hyper-intensity in the dorsal columns on MRI, the authors attributed his myelopathy to nelarabine¹³. His folate and vitamin B12 levels were reported as normal. In this case report, the serum homocysteine levels were abnormally high at 198 mg/dl (normal 14-58 mg/dL). We propose that this patient had myelopathy induced by methotrexate and not nelarabine, given the history of methotrexate therapy and elevated homocysteine levels, despite the normal folic acid levels. Indeed, states of folate deficiency have been reported with normal serum level of folic acid²⁸. Other reports of dorsal column myelopathy attributed to nelarabine have been published^{14, 29}. In each of these reports intrathecal and/or systemic methotrexate were also given. This raises the question of whether nelarabine is actually the causal agent in these cases. In our series only 2 patients had ever received nelarabine therapy.

Because low pretreatment serum folate levels (less than 10 nmol/L) have been shown to increase homocysteine levels, screening patients for low folate levels before methotrexate administration could be considered, although this practice must be weighed against the risk of “folate over-rescue,” which is thought to reduce the anti-neoplastic effectiveness of methotrexate^{30, 31}. Unfortunately, the administration of B12 and folate did not result in neurologic improvement in our series. Administration of dextromethorphan to patients with severe methotrexate induced CNS toxicity has been shown to result in symptomatic improvement in several case series^{25, 32}. We did not appreciate significant, durable improvements in neurologic symptoms among the two patients that received dextromethorphan therapy. The postmortem examination of the un-irradiated cervical cord revealed that axonal loss had occurred and not simply demyelination. This suggests that early intervention is essential to avoid likely irreversible axon death. It is plausible that the folate replacement and dextromethorphan administration were not given early enough in the disease process to halt axonal deterioration. Interestingly, another autopsy case of dorsal column myelopathy following methotrexate treatment for B-cell ALL in a 23 year old patient with Down syndrome revealed marked axonal loss in the posterior column of the cervical spinal cord, consistent with our findings¹¹.

Treatment related myelopathy should be in the differential diagnosis of leukemia patients that present with neurologic abnormalities. In our study the majority of patients received additional CNS directed therapy prior to recognition of toxicity as the etiology behind the neurologic changes. The autopsy performed truly illustrates the potential for further devastating injury when radiation is administered in the setting of pre-existing neurotoxicity. It is now standard procedure at our institution for axial T2 images to be acquired among leukemia patients undergoing MRI where treatment related toxicity is a possibility. Classically, the focus is on post-contrast images to identify potential enhancement associated with leptomeningeal disease. In the majority of the cases in our series, re-review of initial MRI studies that were read as unremarkable, actually did demonstrated subtle T2 dorsal changes on the sagittal image. Likely if axial T2 images were performed, even these subtle changes would be recognized and may have spared several patients additional neurotoxic therapy.

There are limitations to our study and several questions remain unanswered. Serum folate and homocysteine levels were available only for a fraction of the patients in this series, therefore it is unknown if the other patients had the same serum findings. Furthermore CSF homocysteine levels were not obtained in any patient. It is unclear if elevation of homocysteine in the spinal fluid may be of greater diagnostic value due to increased sensitivity. Also due to the heterogeneity in the chemotherapy drugs given to these 13 patients, it is plausible that another drug may be the causative agent. Other factors may have also contributed to the dorsal myelopathy in these patients, such as immune dysfunction as has been reported previously¹⁵. Given the potential for this radiographic finding to be overlooked by radiologists it is likely that there have been other cases of dorsal myelopathy that were not recognized.

In conclusion, we assert that treatment-related myelopathy should always be considered when patients with leukemia present with neurologic abnormalities. At our institution, standard procedure now includes obtaining both axial and sagittal T2 MR images from patients with leukemia in whom treatment-related toxicity is suspected. CSF levels of protein and myelin basic protein should be measured; if they are elevated, methotrexate-mediated toxicity may be causing the myelopathy. Future studies of genetic causes of excessive methotrexate toxicity may help to identify patients at high risk of developing severe dorsal myelopathy.