Epidemiology and Pathophysiology

Traumatic optic neuropathy is an acute injury of the optic nerve with disruption of visual function. Specific causes include motor vehicle and bicycle accidents, head trauma from falls and falling debris, assault, stab and gunshot wounds, endoscopic sinus surgery mishap, recreation-associated trauma, and seemingly trivial injuries. Traumatic optic neuropathy is part of the spectrum of head trauma. Historic series suggest that traumatic optic neuropathy is seen in 0.5% to 2% of head injuries ; however, traumatic optic neuropathy is more common in the setting of craniofacial fractures. In the first 6 years of operations Iraqi and Enduring Freedom, 30 484 troops were wounded in action. Of these, 523 had globe or orbital adnexal combat injuries that required tertiary care. This included 103 cases of traumatic optic neuropathy, accounting for nearly 20% of the ocular combat injuries. Almost half of these cases (48%) were indirect optic nerve trauma. The incidence of loss of consciousness at the time of traumatic optic neuropathy is reported as 20% to 75% depending on the series. Prognosis for visual recovery is poor for patients presenting with no light perception vision and orbital fractures, afferent pupillary deficits that exceed 2.1 log units, and flash visual evoked potentials that are less than 50% of the unaffected optic nerve.

Traumatic optic neuropathy is classified as direct when the nerve is injured directly by a projectile, knife, or other object that penetrates the orbit to damage the optic nerve. Indirect optic neuropathy is diagnosed when the injury to the nerve results from the nonpenetrating effects of trauma, including hemorrhage, edema, and concussion. Indirect traumatic optic neuropathy has been the subject of more clinical research than direct injury. This may reflect the increased opportunity for visual recovery associated with indirect traumatic optic neuropathy.

Depending on the nature of the event, impact forces are loaded into the optic nerve within milliseconds (range, 1.5 to 19 ms). The shock wave can fracture the optic canal. Static loading studies suggest that compression of the superior orbital rim is transferred and concentrated in the orbital roof and optic canal. The optic nerve is fixed within the optic canal, which contributes to the vulnerability of the nerve in this location. Coup-contrecoup forces whip mobile portions of the optic nerve against fixed structures, causing injury. Violent rotation of the globe can result in partial or complete optic nerve avulsion. Finite analysis modeling suggests that rapid globe rotation is associated with an abrupt rise in intraocular pressure, a factor in globe rupture at the optic nerve head.

Traumatic optic neuropathy is also classified anatomically based on distinct clinical presentations associated with injury location. Partial or complete optic nerve head avulsion presents with intraocular hemorrhage at the optic nerve head and disruption of optic nerve head anatomy. Anterior optic nerve trauma is presumed when there is loss of vision with disruption of circulation at the optic nerve head. The trauma affects the nerve at a site anterior to where the ophthalmic artery enters the optic nerve. Unilateral or asymmetric bilateral posterior traumatic optic neuropathy is a clinical diagnosis that should be considered when visual loss following trauma cannot be explained by other ocular injuries and there is a relative afferent pupillary defect even though the fundus appears normal. In bilateral injury, absence of a relative afferent pupillary defect suggests that the nerves are comparably damaged. Posterior injuries involve the optic nerve proximal to where the ophthalmic artery enters the optic nerve in the orbit to the optic chiasm. The most common sites of injury are the foramina of the optic canal, the optic canal, and under the falciform dural fold that drapes the edge of the anterior clinoid process.

Walsh described the process of contusion necrosis that results from shearing injury to the axons and microvasculature, and this process is thought to be the basis for injury in these cases. What follows are postinjury biochemical cascades that exacerbate the initial damage. Most cases of visual loss are immediate; however, delayed visual loss is documented in 10% of cases. Treatment with surgery and corticosteroids is intended to limit secondary injury.

High-Dose Corticosteroids

In the neurosurgical literature, the term high-dose corticosteroids refers to the treatment protocol of intravenous methylprednisolone used to treat acute spinal cord injury in the Second and Third National Acute Spinal Cord Injury Studies (NASCIS II and NASCIS III). The protocol consisted of a loading dose of 30 mg/kg followed by a continuous intravenous infusion of 5.4 mg/kg per hour for 24 or 48 hours. The ophthalmic literature refers to this protocol as megadose corticosteroids. Investigators have used a range of methylprednisolone regimens to treat traumatic optic neuropathy. As classified by Levin and associates, methylprednisolone regimens include megadose (greater than 5400 mg/d), very-high-dose (2000 to 5400 mg/d), high-dose (500 to 1999 mg/d), moderate-dose (100 to 499 mg/d), and low-dose treatment (below 100 mg/d). For purposes of clarity and consistency with the neurosurgical literature, in the discussion to follow, intravenous methylprednisolone of 30 mg/kg loading followed by a continuous intravenous infusion of 5.4 mg/kg per hour for 24 or 48 hours will be referred to as high-dose rather than megadose treatment. In the discussion of other protocols, the actual dose will be cited (eg, 250 mg/d of methylprednisolone rather than moderate dose).

Treatment of traumatic optic neuropathy with high-dose methylprednisolone was widely embraced by the ophthalmic community with the publication in 1990 of the results of the Second National Acute Spinal Cord Injury Study. This study was a randomized, double-blind, placebo-controlled study of patients with acute spinal cord injury, which was published in the New England Journal of Medicine. Patients were randomly assigned to 1 of 3 treatment arms within 12 hours of injury: high-dose methylprednisolone, naloxone, or placebo. The study concluded that patients with acute spinal cord injury treated with high-dose methylprednisolone within 8 hours of injury have a statistically significant improvement in motor and sensory function compared with placebo, naloxone, or high-dose methylprednisolone treatment started more than 8 hours after trauma. Critics of the Second National Acute Spinal Cord Injury Study argue that stratifying the corticosteroid-treated patients based on the 8-hour time frame was chosen as a result of post hoc data analysis. When the 8-hour stratification is removed, corticosteroid-treated patients have worse outcomes overall compared with those receiving placebo.

Since 1990, there have been 43 studies of traumatic optic neuropathy that have enrolled at least 10 patients, for a total of 1906 patients. Of these, 1442 were treated with corticosteroids alone or corticosteroids plus optic canal surgery. Unfortunately, each of these studies is relatively small. In some cases, conclusions are limited by the changing approach to patient recruitment evident in the literature. Compared with older studies, these reports have larger numbers of patients identified and treated more quickly after trauma.

Two studies from the Massachusetts Eye and Ear Infirmary highlight the effect of early diagnosis. The first is by Lessell, who reported 33 cases of traumatic optic neuropathy seen from 1976 to 1987. In this series, 6 (18%) of the patients initially had no light perception vision and 9 (27%) had visual improvement. Contrast this with a study from the same institution published just 1 year later by Joseph and associates. They reported 14 unilateral cases of traumatic optic neuropathy seen in a 16-month period between April 1987 and October 1989. Of these, 5 patients (36%) presented with no light perception vision. All patients were seen within 1 week, and 5 (36%) underwent optic canal surgery within 24 hours of their trauma. Overall, 11 patients (79%) had visual improvement. These studies, from a single institution, demonstrate the complexity of interpreting the traumatic optic neuropathy literature. It took 11 years to accumulate 33 cases of traumatic optic neuropathy in Lessell’s series, but just 16 months to accumulate another 14 cases in the series by Joseph and associates.

This disparity raises several questions. Did the community incidence or referral patterns of traumatic optic neuropathy change? It is more likely that study investigators made a greater effort to identify patients with acute traumatic optic neuropathy presenting to their institution. Did the treatment in the second series account for the improvement in outcome, or was it simply the result of identifying a cohort with traumatic optic neuropathy closer in time to the initial injury, thereby increasing the opportunity for spontaneous visual improvement? It is likely that early diagnosis biased the second study toward a more favorable outcome irrespective of treatment. Conversely, as time from injury increases, the probability of visual recovery decreases.

This set of articles is not unique in presenting conflicting information that creates confusion in the literature. In the study by Chou and associates, the authors described 58 patients with traumatic optic neuropathy. They concluded that patients treated with high-dose corticosteroids had better outcomes than untreated patients. Untreated patients, however, did not present to the authors until 3 weeks after injury on average. In contrast, patients treated with corticosteroids alone presented, on average, within 2 weeks of trauma; and those who had surgery plus corticosteroids presented, on average, within 3 days of trauma. Time to study recruitment rather than treatment protocols may account for the difference in outcomes in this particular study.

The International Optic Nerve Trauma Study analyzed 133 patients with traumatic optic neuropathy who had vision assessed within 3 days of their trauma. The striking feature of this study is the bias toward treatment. Of 133 patients, 125 received corticosteroids in varying doses, and 33 of those also had optic canal surgery. This bias for treatment means that we do not know the natural history of traumatic optic neuropathy and the likelihood of spontaneous visual recovery. Without a control group of untreated patients with traumatic optic neuropathy identified at a time frame comparable to that of treated patients, it is impossible to conclude that medical or surgical intervention makes a difference. The authors of the International Optic Nerve Trauma Study stated there was “sufficient evidence to conclude that neither corticosteroids nor optic canal surgery should be considered the standard of care for patients with traumatic optic neuropathy.” They suggested it was reasonable for clinicians to make empirical treatment decisions with their patients. Since the findings of the International Optic Nerve Trauma Study were released in 1999, several new lines of evidence now suggest that treating traumatic optic neuropathy with high-dose corticosteroids is harmful rather than helpful.

The Corticosteroid Randomization After Significant Head Injury (CRASH) trial was a multicenter, randomized, placebo-controlled study of high-dose methylprednisolone therapy for acute head trauma, published in the journal Lancet in 2004. Patients were randomly assigned within 8 hours of trauma to treatment with either placebo or high-dose methylprednisolone (30 mg/kg loading followed by an infusion of 5.4 mg/kg per hour) for 48 hours. The goal was to enroll 20 000 patients, but the study was halted at 10 008 patients. Safety monitoring data revealed a higher risk of death from all causes 2 weeks after trauma in the corticosteroid-treated patients (21% vs 18% mortality, P = .0001). This is 1 excess death for every 31 patients with head trauma treated. This finding has prompted neurosurgeons to abandon the use of high-dose corticosteroids after head trauma. The CRASH trial has immediate implications for treating traumatic optic neuropathy given the high incidence of concomitant head trauma.

Not every traumatic optic neuropathy patient will have a brain injury, so the mortality rate associated with corticosteroid treatment will be lower than that seen in patients with primary head trauma. No study of high-dose methylprednisolone therapy for traumatic optic neuropathy has used death as a study endpoint because death in this setting has always been attributed to the underlying head trauma. Traumatic optic neuropathy studies exclude these patients. Since the CRASH trial, informed consent for high-dose corticosteroid therapy for traumatic optic neuropathy must include information on the increased mortality to be expected from this treatment. Recognizing the existence of a higher mortality rate associated with treating traumatic optic neuropathy with high-dose corticosteroids, ophthalmologists should look to affirmative evidence balancing the potential risks of treatment.

Not only do we lack evidence for the efficacy of high-dose corticosteroids, there are now several animal studies that show high-dose corticosteroids to be directly toxic to injured optic nerve. Steinsapir and associates studied the treatment of experimental optic nerve injury with methylprednisolone in doses ranging from 30 to 120 mg/kg, compared with a saline-treated group. The saline-treated animals retained statistically significantly greater numbers of axons than the corticosteroid-treated animals. The second research study was a double-insult study by Ben Simon and associates. They found that head trauma preceding optic nerve injury was neuroprotective. Furthermore, treatment of the animals with high-dose methylprednisolone (30 mg/kg) after the optic nerve injury resulted in a loss of the protective effect associated with the head trauma. This neuroprotective effect was preserved when the animals were treated with low-dose methylprednisolone (1 mg/kg). Finally, Diem and associates demonstrated that methylprednisolone (20 mg/kg) was detrimental in an animal model of inflammatory optic neuropathy. Optic nerve inflammation is a different entity than traumatic optic neuropathy but may share common postinjury biochemical cascades. Corticosteroids may paradoxically interfere with adaptive mechanisms that limit the spread of secondary injury. No animal models of indirect traumatic optic neuropathy demonstrate a benefit for corticosteroids at any dose.

Clinical evidence does not support the use of high-dose corticosteroids for traumatic optic neuropathy. This treatment increases mortality in patients with head trauma. Animal studies show the treatment to be toxic to injured optic nerve. There is no reason to subject our patients to this treatment. The potential loss of life associated with high-dose corticosteroids in the setting of head trauma is not counterbalanced by clinically meaningful benefits. Without new and compelling research data, high-dose corticosteroids for traumatic optic neuropathy should be abandoned.

This naturally raises the question about whether lower dosages of methylprednisolone could be safe and effective. The answer is that no clinical or animal studies have shown any convincing evidence that methylprednisolone is effective for treating traumatic optic neuropathy at any dose. Clinical enthusiasm to use corticosteroids for treatment of traumatic optic neuropathy without any evidence that it may be helpful must be weighed against our new understanding that the treatment can actually be harmful. Whether methylprednisolone is safe when administered in a dosage below 20 mg/kg every 6 hours cannot be answered. The absence of data needed to answer this question does not provide a rationale for conducting human experimentation, given the risk of potential harm. Further clinical experimentation with alternate corticosteroid protocols would be difficult to support ethically without compelling, new animal data justifying such a treatment approach or, at the minimum, establishing that such treatments do not cause harm. Ben Simon and associates found that high-dose methylprednisolone (30 mg/kg) interfered with an optic nerve neuroprotective effect seen following head trauma. Methylprednisolone given at 1 mg/kg did not block the neuroprotective effect. This does not imply that methylprednisolone even at this lower dose is safe or effective for treating traumatic optic neuropathy.

Following trauma, clinicians may have reasons other than the treatment of traumatic optic neuropathy for considering corticosteroid therapy. To cite just one hypothetical example, we lack evidence to conclude that a kidney transplant recipient with a diagnosis of traumatic optic neuropathy who requires long-term prednisone therapy to control graft rejection should have the prednisone regimen discontinued. In the absence of better information, it seems prudent to limit such treatments to the equivalent of 1 mg/kg every 6 hours of methylprednisolone in the presence of concomitant acute traumatic optic neuropathy.