Gadolinium enhancement of cauda equina after combined spinal–epidural anaesthesia

Abstract

The occurrence of neurological symptoms after spinal anaesthesia has been reported with several local anaesthetics including lidocaine, prilocaine, mepivacaine, tetracaine and bupivacaine. Although hyperbaric bupivacaine is known to induce neurological symptoms less frequently than lidocaine, a few cases of cauda equina syndrome (CES) following the intraspinal injection of bupivacaine have been reported in the English literature. We describe lumbar MRI findings for a 29-year-old woman presenting with CES after caesarean section.

Cauda equina syndrome (CES), a rare complication of spinal or epidural anaesthesia, results from injury to sacral roots in the neural canal. Because the nerves in the cauda equina lack a protective sheath as they pass through the distal end of the dural sac, they are particularly prone to injury from high concentrations of local anaesthetics. Although laboratory findings have suggested that lidocaine, mepivacaine and ropivacaine may have greater potential for neurotoxicity than bupivacaine [1], several cases of CES have been reported in association with the intrathecal injection of bupivacaine [2]. To the best of our knowledge, no spinal MRI findings have been presented for nerve root injury caused by spinal anaesthesia.

Case report

A 29-year-old multiparous woman was admitted to the hospital in the 38th week of pregnancy for her second caesarean delivery. Her past medical history was unremarkable. On the second day of hospitalization, she went to the operating room. Combined spinal–epidural anaesthesia (CSEA) was performed in the left decubitus position with 2.2 ml 0.5% hyperbaric bupivacaine administered intrathecally at the L3–4 interspace, followed by placement of a multiorifice epidural catheter at the T8–9 interspace. An 18-gauge epidural catheter was positioned 5 cm into the epidural space oriented in the cephalad direction to optimise cephalic spread of the anaesthetic. Surgery was allowed to proceed after sensory block from T5 to S5 was achieved. A total dose of 5.1 ml of 0.2% ropivacaine was administered using a patient-controlled epidural analgesia device over 75 min. Oxygen 3 l min^–1 was given through a face mask; 20 min elapsed between induction of anaesthesia and uterine incision. The surgical procedure was completed without incident and the estimated blood loss was 800 ml.

The patient did not fully recover from the motor and sensory blockade induced by CSEA, and at about 48 h post-partum was unable to walk because of lower extremity weakness. Neurological assessment confirmed sensory deficit to pinprick and vibration in the T3 to L5 distribution bilaterally with preservation of position sense. On the following day, she still exhibited hypaesthesia bilaterally in the T6 to L2 distribution with lower extremity weakness, which had shown minor improvement. Results of lumbosacral MRI without contrast medium were within normal limits, ruling out the possibility of spinal epidural haematoma or spinal infarction. Because the levels of sensory and motor dysfunction were nearly identical to the extent of the CSEA, a neurological complication related to lumbar epidural or subarachnoid infusion of anaesthetic was deemed the most likely cause of her symptoms; 7 days postoperatively, she continued to be unable to walk, void or defaecate spontaneously. A repeat MRI study with contrast medium was performed, which detected enhancement of the nerve roots from the T11–T12 to L3–L4 level (Figure 1). At the level of T12–L1 and L1–L2, the ventral nerve roots exhibited higher signal intensity than the dorsal nerve roots on post-contrast axial T₁ weighted images. Neither thickening of the nerve roots nor spinal cord abnormalities were demonstrated. Results of testing for autoantibodies such as antinuclear antibody, anti-ds-DNA antibodies, anti-Sjögren's syndrome A, anti-Sjögren's syndrome B and anti-neutrophil cytoplasmic antibodies were negative. An autoimmune process with nerve root involvement was therefore considered unlikely. Haematological examination revealed no abnormality suggestive of infection (white blood cell count, 6500 mm^–3; C-reactive protein, 0.2 mg dl^–1). The patient was treated with high-dose corticosteroids, with gradual improvement of signs and symptoms. By the end of treatment, she was able to walk without assistance, but 6 months later she still had mild sensory disturbance of the lower extremities and pelvic area.

Figure 1.

A 29-year-old woman with cauda equina syndrome after combined spinal–epidural anaesthesia. Contrast-enhanced sagittal (A) T₁ weighted magnetic resonance (MR) image reveals abnormal enhancement in the cauda equina. Contrast-enhanced axial T₁ weighted MR images show enhancement of the ventral nerve roots at the level of T12–L1 (B) and the level of L1–L2 (C).

Discussion

Persistent neurological injury is a rare but significant complication of spinal anaesthesia. Four cases of CES after continuous spinal anaesthesia with 5% lidocaine or 1% tetracaine were reported in 1991 and generated concern regarding the neurotoxic potential of local anaesthetics administered intrathecally [3]. Although bupivacaine may be less neurotoxic than other local anaesthetics [1], 11 patients with CES attributable to bupivacaine were reported among the 1 710 000 cases of spinal and epidural blockade reviewed by Moen et al [4].

Several mechanisms have been proposed for neurological sequelae of spinal and epidural administration of local anaesthetics [1]. Sakura et al [5] intrathecally infused normal saline, 10% lidocaine or 2.13% bupivacaine in rats, and found that the specimens from rats given lidocaine or bupivacaine included axonal degeneration, disintegrated myelin lamellae and myelin ovoid. Yamashita et al [6] compared the effects of local anaesthetics on glutamate concentrations in cerebrospinal fluid microdialysate and neurological and histopathological outcomes in rabbits. They observed characteristic types of histopathological injury to the cauda equina and the spinal cord, with chromatolytic injury of motor neurons and vacuolation in the dorsal funiculus.

It is unclear why intense contrast enhancement was observed in the ventral nerve roots in our patient, despite the presence of dorsal root symptoms as well as motor disturbance. One possible explanation is unmyelinated fibres in the ventral nerve roots. Ko et al [7] reported that approximately 20% of human spinal ventral root fibres were unmyelinated. This suggests the possibility that the ventral nerve roots may be more severely affected than the dorsal roots by anaesthetics. Another possible explanation is that some agents injure the ventral nerve roots more strongly than the dorsal roots. Anderson et al [8] reported the case of a 3-year-old girl diagnosed with acute lymphoblastic leukaemia who developed subacute anterior lumbosacral polyradiculopathy following intrathecal methotrexate. Her spinal MRI revealed lumbosacral ventral root enhancement, which the authors suggested could indicate selective ventral nerve root vulnerability to intrathecal chemotherapy. Their patient's electrodiagnostic studies demonstrated motor axonal loss, rather than demyelination. An effect of uneven dispersion of local anaesthetics in the subarachnoid space cannot be excluded as another possible cause, although the puncture level of spinal anaesthesia, which was more caudal than the levels of increased enhancements of the ventral roots, and the absence of the denticulate ligament below the level of T11, which could hamper homogeneous spread of anaesthetics, make this unlikely in our patient. The possibility that CSEA has an intrinsic potential to cause nerve root enhancement not correlated with our patient's clinical findings could not be completely excluded either.

The differential diagnosis of our patient's findings includes the pure motor and axonal variant of Guillain–Barré syndrome (acute motor axonal neuropathy (AMAN)). Byun et al [9] reported MRI findings for the spine in eight patients with Guillain–Barré syndrome and concluded that enhancement of the anterior spinal nerve roots alone is strongly suggestive of AMAN. However, the absence of an antecedent event, the short duration of time between the onset of symptoms and CSEA, the concordance in distribution of neurological disturbances with the extent of local anaesthetic delivery, the sphincteric involvement, and the prolonged neurological impairment did not support a diagnosis of AMAN in the present case.

Radwan et al [1] assessed the neurotoxicity of local anaesthetics including bupivacaine using dorsal root ganglia of chick embryos, and found that all local anaesthetics tested produced growth cone collapse and neurite degeneration. Although their findings cannot be directly extrapolated to in vivo conditions, the detrimental effects of local anaesthetics on growing or regenerating neurons may have played a role in prolonging the clinical course in this case.

In conclusion, we have described a case of CES following the intrathecal injection of bupivacaine with contrast enhancement of nerve roots on spinal MRI. Although the reason for this finding is unclear, it may have involved ventral root vulnerability to bupivacaine.