Diplopia: A Complication of Dural Puncture

To compile case reports of EOMP associated with dural puncture, we performed a computerized search of the medical literature in English, Spanish, French, German, and Japanese from 1966 through December 20, 2002 using PubMed §and the OVID search engine (Ovid Technologies, New York, NY). Key words used included dural puncture , lumbar puncture , spinal anesthesia , spinal puncture , spinal injection , epidural anesthesia , myelography , diplopia , ophthalmoplegia , abducens nerve , oculomotor nerve , and trochlear nerve . Forty-four related articles were identified, and 41 reports were obtained. Four reports were excluded from the review because of either insufficient descriptions of the cases or confounding factors, such as medicine or underlying disease, that could have otherwise contributed to EOMP. 3–6An additional 12 reports were obtained after hand-searching reference lists of retrieved reports and review articles. A total of 94 reported cases and one of our own were analyzed for this review (table 1).

The reported incidence of EOMP after dural puncture varies from 1 in 400 to 1 in 8,000. 7–9These incidence reports were either from retrospective reviews of spinal anesthesia in 1947 7and 1961 8or diagnostic lumbar punctures in which larger spinal needles were often used. 9Spinal anesthesia was found to be the most frequently reported procedure involved (47%), followed by myelography (18%), diagnostic lumbar puncture (12%), epidural anesthesia/injection (11%), continuous spinal anesthesia (4%), and other dural puncture procedures (9%).

Although other cranial nerve palsies can occur after lumbar puncture, the abducens nerve (cranial nerve VI) is affected in the majority of cases (92–95%). 7,10Nearly 80% of the cases are unilateral. 7Abducens palsy can coexist with oculomotor (cranial nerve III) or trochlear (cranial nerve IV) nerve palsies. Multiple coexisting cranial nerve palsies can be masked by a large esotropia, making the exact diagnosis of these cranial nerve palsies difficult. 11 

Extraocular muscle paralysis has been reported in patients aged 17–69 yr (mean age, 42 yr). Perhaps the largest survey of EOMP associated with spinal anesthesia was by Thorsen 7in 1947. He reported that 80% of the patients were younger than 50 yr and 30% of the patients were younger than 30 yr, although most of the patients who had spinal anesthesia were older than 30 yr.

The incidence of PDPH in women has been reported to be twice as high as in men. 12This sex pattern does not hold true for EOMP: Thorsen 7(1947) reported a predilection of abducens nerve palsy for men, whereas Hayman and Wood 10(1942) stated that women were more susceptible. In our review, no significant sex predilection of this complication was found (male vs.  female: 55%vs.  45%, respectively).

The window period for EOMP to manifest is 1 day to 3 weeks after dural puncture, but it most often presents 4–10 days after dural puncture (mean, 7 days; median, 6 days) (table 1and fig. 1). This finding is consistent with classic reports. 7,10EOMP associated with dural puncture is almost always preceded by PDPH, but EOMP can occur either before or after the headache subsides. If the cranial nerve (III, IV, or VI) palsy is isolated, is preceded by PDPH, and occurs within 3 weeks after dural puncture with no other neurologic deficits, it is likely that cranial nerve palsy is a postdural puncture complication. Diagnosis of this complication is based purely on clinical presentation, and there is no specific test for its accurate diagnosis.

Magnetic resonance imaging of the brain has occasionally shown signs of cerebrospinal fluid (CSF) volume depletion and intracranial hypotension, such as diffuse pachymeningeal enhancement, descent of the brainstem, and subdural fluid collections. 9,13,14However, these findings are not specific for EOMP after dural puncture and can be seen in spontaneous intracranial hypotension. 15–17There may be no abnormality found 18–21when magnetic resonance imaging is performed after PDPH resolution.

The differential diagnosis of acquired EOMP is varied, such as neoplasm, ischemia, trauma, aneurysm, multiple sclerosis, encephalitis and myasthenia gravis. The greatest proportion of abducens nerve palsies are of unknown origin; however, the overall spontaneous recovery rate is close to 80%. 22,23Despite the good prognosis of the nerve palsy as well as the low yield and low specificity of diagnostic studies for this complication, magnetic resonance imaging of the brain may still be of value to rule out other serious conditions that require treatment. Indeed, subdural hematoma or hygroma can rarely occur after dural puncture, from tearing of the bridging dural veins associated with acute intracranial hypotension. 24–28Subdural hematoma should be included in the differential diagnosis if prolonged PDPH loses the postural dependence of symptoms and/or is accompanied by other neurologic signs.

An epidural blood patch is highly effective for PDPH, with the reported success rate being as high as 93% for the first attempt, 29but it has consistently failed to show efficacy in treating EOMP after dural puncture. 13,18,30–32Abducens palsy is associated with a favorable outcome in general, and its prognosis after dural puncture is even better. In our review, 80 of 90 patients (89%) fully recovered in 2 weeks to 8 months (table 1). The majority of those recovered within 6 months (figs. 2 and 3), consistent with classic reports. 7,10 

Conservative treatment (such as an eye patch or prism glasses) is generally adequate to minimize the patient’s discomfort. Isolated abducens palsy in the absence of other neurologic signs or symptoms should be observed for improvement for 8 months. Further investigation is unwarranted if the deficit resolves spontaneously.

In our review, EOMP cases that lasted more than 8 months were found to be permanent. No apparent trend was found among the cases of permanent EOMP, probably because of a small number of cases and the lack of detailed description in some of the reports (e.g. , size and type of dural puncture needle). Corrective surgery on the extraocular muscles, such as recession of the medial rectus muscle, was performed in some patients, allowing them to resume social activities or previous occupations. 13,33,34However, some suggest that surgical correction be postponed until at least 18 months have lapsed, considering cases of protracted recovery. 35,36Indeed, some patients became asymptomatic after several months to years, although residual hyperdeviations persisted. 37–39The decision should be individualized based on the duration (at least longer than 8 months) and severity of symptomatic EOMP as well as the risks and benefits of the corrective surgery.

Similar to PDPH, intracranial hypotension due to CSF leakage is the generally accepted cause of cranial nerve palsies after dural puncture because this complication can occur after diagnostic lumbar puncture where no medication is injected into the intrathecal space. Spontaneous intracranial hypotension with orthostatic headache that occasionally presents as diplopia (commonly abducens nerve palsy) supports the hypothesis. 15–17The brain descends caudad with upright posture, and the CSF “cushion” for the brain is displaced. This downward traction could damage some of the cranial nerves that anchor the brain in the skull.

The time course (of weeks to months) for the cranial nerve palsies with good prognosis suggests neurapraxia (focal segmental demyelination) or axonotmesis (axonal interruption and Wallerian degeneration with preservation of supporting tissue framework) as a potential pathologic mechanism. 40Even after axonotmesis, electrical activity and conductivity may be present in the axonotmetic distal stump for a day or two, but the axon then quickly becomes unresponsive with degeneration. 41,42This may explain why an epidural blood patch is not effective in treating the cranial nerve palsies. There are both functional disturbances and structural lesions in the nerve by the time the nerve palsy manifests.

Preferential damage to the abducens nerve can be explained by its anatomic course. As the nerve emerges into the subarachnoid space from the caudal pons, it immediately ascends the clivus, crosses branches of the basilar artery, and pierces the dura mater. Then, the nerve bends at nearly a right angle over the petrous apex of the temporal bone. 43,44The abducens nerve runs in the direction of the typical caudad displacement of the brain with intracranial hypotension. As a result, traction associated with changes in intracranial pressure is fully transmitted to the nerve. The nerve can be stretched by caudal displacement of the pons, and it also may be compressed at the dura, petrous apex, or basilar artery if (1) the penetration aperture for the nerve in the dura or petrous apex is sharply edged or (2) branches of the basilar artery are well developed.

Ikeda et al.  45,46have shown that a combination of nerve stretch and compression even to a mild degree can be more detrimental to a nerve than either alone, causing severe axonotmesis, with not only Wallerian degeneration but also retrograde degeneration from the injured site (“dying-back degeneration”). The wide variation in duration of the nerve palsies may be associated with varying degrees of nerve injuries from mild neurapraxia with conduction block to severe axonotmesis with extensive degeneration.

After EOMP occurs, little can be done to change its course. Therefore, prevention is of great importance. Vandam and Dripps 47reported in their survey of 10,098 patients undergoing spinal anesthesia that diplopia (secondary to probable abducens nerve palsy) occurred only in patients who underwent continuous spinal anesthesia with a 16-gauge needle. The incidence of EOMP was high (1:140). Because intracranial hypotension associated with CSF leakage seems to play a major role in the pathogenesis, minimizing CSF leakage with smaller, pencil-point needles should reduce the risk of EOMP. However, EOMP can occur after otherwise uncomplicated spinal anesthesia using a 25-gauge Whitacre needle. 48 

Bed rest has been advocated in cases of dural puncture by some clinicians. However, a recent meta-analysis failed to show that bed rest after dural puncture was better than immediate mobilization in reducing the incidence of PDPH. 49Bed rest can be associated with a higher incidence of PDPH in particular patient groups. 50,51In theory, however, upright posture may exacerbate compression–stretch injury of the abducens nerve by promoting further caudad displacement of the brain. Because of the low incidence of EOMP, it seems unlikely that the value of routine bed rest in an effort to prevent EOMP will ever be determined.

Does early application of an epidural blood patch after dural puncture prevent cranial nerve palsy from occurring by restoring intracranial pressure? There are no studies to support this idea, nor is such a study feasible because of the low incidence of this complication. EOMP could occur shortly after an epidural blood patch procedure for PDPH because of the slow manifestation of EOMP. In this circumstance, the blood patch procedure may be blamed for abducens nerve palsy if diplopia was not recognized as a complication of dural puncture. On the other hand, a blood patch can actually cause EOMP and/or aggravate neurologic symptoms in a patient with PDPH or spontaneous intracranial hypotension when the mass effect of the coexisting subdural hematoma can no longer be compensated by the disappearance of CSF leak by the patch. 28,52A thorough history and physical examination assessing signs or symptoms suggestive of subdural hematoma is mandatory before an epidural blood patch procedure, and close observation after the procedure is strongly recommended. Should mental status changes or any neurologic signs/symptoms manifest, the patient will need immediate medical attention and imaging studies.

Although the current incidence of EOMP after dural puncture is unknown, it can occur with smaller pencil-point spinal needles. Abducens nerve involvement is most often unilateral. EOMP seems to be very rare in elderly patients, and male and female patients seem to be equally vulnerable. EOMP usually occurs 4–10 days after dural puncture but can manifest as late as 3 weeks. Full recovery can generally be expected in 2 weeks to 8 months, although permanent cases have rarely been reported. Anesthesiologists, emergency physicians, neurologists, and ophthalmologists should be aware of this complication and communicate the information so that early diagnosis can alleviate patient anxiety. The exact pathophysiology is unclear, but a nerve lesion such as neurapraxia or axonotmesis caused by stretch and/or compression secondary to intracranial hypotension due to CSF leakage is the generally accepted mechanism. Treatment is supportive except for persistent or permanent cases, for which corrective surgery may be necessary. An epidural blood patch does not seem to be an effective treatment, whereas the benefit of a prophylactic blood patch is unknown. Avoiding, if possible, or minimizing CSF leakage associated with dural puncture may be the only measure for now to potentially minimize the risk of this rare but distressing complication.