METHYLPREDNISOLONE AND OTHER CORTICOSTEROIDS IN SPINAL CORD INJURY

Steroids in various forms have been used in the treatment of SCI for many years. Historically, the rationale for the use of corticosteroids in the management of neural trauma extended from their use in decreasing edema in the management of brain tumors. In addition, their anti-inflammatory effect was thought to be beneficial to the secondary injury pathophysiology of SCI. Studies in dogs supported these hypotheses and demonstrated a modest improvement in neurologic outcome with steroid treatment and a modified anti-inflammatory response.¹⁶ Subsequently, other animal studies have provided support for improved neurologic recovery after SCI in animal models when methylprednisolone is administered and have provided evidence for inhibition of lipid peroxidation, protection of energy metabolism, reduced post-traumatic ischemia, maintenance of neurofilament structure, and decreased post-traumatic ionic shifts.¹⁷

The role of steroids in human SCI management became more rigorously considered after publication of the Second National Acute Spinal Cord Injury Study (NASCIS II). Unfortunately, the initial enthusiasm for an apparent positive effect of methylprednisolone in SCI demonstrated by NASCIS II has not stood up to the extensive scrutiny that ensued.^18–21 However, despite significant criticism, this medication continues to be prescribed by many physicians, and a 2002 study suggests that most practitioners prescribe it because of peer pressure or fear of litigation, rather than a firm belief that it is indeed efficacious.²²

The first NASCIS study compared low- (100 mg/day × 10 days) and high-dose (1000 mg/day × 10 days) methylprednisolone, and did not include a placebo group.²³ It failed to demonstrate a difference between the doses tested. The high-dose group exhibited an increased risk for complications. That study was followed by a randomized, controlled trial comparing a 24-hour protocol (30 mg/kg methylprednisolone bolus followed by 5.4 mg/hg/hr until 24 hours) with placebo in NASCIS II.^24,25 The dose selected in NASCIS II was greater than that of the original study because of further animal work that suggested a therapeutic threshold of 30 mg/kg.²⁵ NASCIS II concluded that improved neurologic recovery was seen when the methylprednisolone treatment protocol was initiated within 8 hours of injury. That study was then followed by NASCIS III, which compared patients randomized to the 24-hour NASCIS II protocol with those randomized to a 48-hour protocol (5.4 mg/kg/hr methylprednisolone after the 30-mg/kg bolus).^26,27 That study concluded that patients for whom therapy was initiated within 3 hours did not gain any benefit from extending treatment to 48 hours, whereas those for whom therapy was initiated between 3 and 8 hours did benefit further. No benefit has been shown if therapy is initiated beyond 8 hours in NASCIS II.

Both the NASCIS II and NASCIS III trials were well designed and executed. However, closer scrutiny demonstrates that the primary analyses of methylprednisolone treatment effect were negative in both studies. The stated conclusions were based on post hoc analyses that suggested minor treatment effects on motor scores at 1 year and when therapy was initiated in the 8- and 3- to 8-hour windows identified in NASCIS II and III, respectively (statistical probability was slightly greater than 0.05 for 1-year motor scores in the NASCIS III 48-hour steroid group). None of the sensory scores was different between treatment groups in either study.

Several concerns have arisen regarding the post hoc analyses of NASCIS II and III. The left-sided motor scores were not published but reported “similar” to right-sided scores. Thus, half the available data were excluded. The statistical analyses failed to correct for multiple statistical comparisons, and it is unclear whether the repeated-measures design was considered. More than 65 methylprednisolone-related t tests were performed in NASCIS II, and more than 100 t tests in NASCIS III. There was, therefore, a high likelihood of type I error (erroneously detecting a statistical difference that does not exist) through random chance. The rationale for an 8-hour subanalysis (NASCIC II) is unclear. It has been claimed that this subgroup was selected based on median time to treatment. However, by definition, 50% of patients should have initiated treatment before the median time of treatment initiation. In fact, only 38% of patients (183/487) were included in this post hoc analysis. The justification for the 3- and 8-hour windows in NASCIS III is similarly obscure. Other observations raise concern about imposing these artificial timerelated stratifications. For example, in NASCIS II, the incompletely injured placebo group when separated into <8- and >8-hour gr-oups differs in recovery, with the latter group showing improved recovery comparable with the <8-hour incompletely injured methylprednisolone group.¹⁷ This implies that these patients could be treated with placebo beyond 8 hours and a similar result to treatment with methylprednisolone initiated within 8 hours can be expected. Finally, another common criticism of the NASCIS studies has been the lack of outcomes assessing functional recovery meaningful to the patient’s expected activities.

In addition to the NASCIS studies, Otani and colleagues²⁸ published a prospective, randomized trial investigating the NASCIS II methylprednisolone dosing protocol. The investigators were not blinded to treatment, and the control group was allowed to receive alternate steroids at the physicians’ discretion. Of 158 patients entered, 117 were analyzed. The primary outcome measures (American Spinal Injury Association [ASIA] motor and sensory scores) were not different between treatment groups. Post hoc analyses suggested that more patients improved on the NASCIS II steroid regimen compared with control patients. However, for a greater number of steroid-treated patients to improve, the fewer control patients who also improved must have demonstrated a larger magnitude of recovery (because overall ASIA motor and sensory scores were no different between groups). Thus, such post hoc analyses become difficult to interpret in the face of a negative overall effect.

A retrospective study with concurrent case controls also suggested a benefit with corticosteroid administration.²⁹ This study investigated the use of dexamethasone initiated within 24 hours of injury with the specific dose left to the discretion of the attending physicians. Length of follow-up was not specified, and a new but unvalidated neurologic grading system was used for outcome assessment. This study reports that the percentage of patients who improved was significantly greater in the steroid-treated group. However, there was a much greater mortality rate within the control group, suggesting a selection bias to more severely injured patients in the control arm. The magnitude of the mortality rate is also a concern and suggests that the study population may not be representative and that the results are not generalizable.

A randomized, controlled trial designed to examine the potential therapeutic benefit of nimodipine (a calcium channel antagonist) included an NASCIS II methylprednisolone regimen and a placebo group.^30,31 This study, which included approximately 25 patients in each group, failed to show any difference between any of four groups (placebo, nimodipine, methylprednisolone, methylprednisolone and nimodipine) using ASIA scores and ASIA grade outcomes. However, this study was remarkable for an increase in infectious complications in the methylprednisolone group.

Most recently, the data from the five randomized trials of methylprednisolone (NASCIS, NASCIS II, NASCIS III, Otani and colleagues,²⁸ and Pointillart and coworkers³¹) were subject to a meta-analysis and concluded that high-dose methylprednisolone given within 8 hours of acute SCI is safe and modestly effective. This article estimated a treatment effect of 4.1 motor score points (from ASIA score) over placebo treatment (using the NASCIS II 24-hour protocol). The findings and limitations of the individual articles included in this review are discussed earlier in this chapter (Table 1-1).

It must also be recognized that corticosteroid administration comes with increased risk for several potential adverse events including pneumonia, sepsis, and steroid-induced myopathy, all of which may negatively impact outcome in patients with SCI, potentially overshadowing any unproven beneficial effect.³² Galandiuk and colleagues³³ demonstrate that patients with SCI treated with corticosteroids exhibited a greater rate of pneumonia (79% vs. 50% with placebo treatment) and required a longer hospital stay (44.4 vs. 27.7 days with placebo treatment). Matsumoto and coauthors³⁴ specifically investigated the complication rate in patients treated with the NASCIS II protocol compared with placebo in a small randomized trial. They found a nonstatistically significant trend to greater complication rates with steroid treatment overall, but a significant difference in respiratory and gastrointestinal complications. The majority of these complications were pneumonia and gastric ulcer, respectively. NASCIS II did not find an increased complication rate with steroid treatment, but NASCIS III reported a greater rate of sepsis and pulmonary complications particularly with the 48-hour infusion. The Corticosteroid Randomization after Significant Head Injury (CRASH) trial investigated the use of a corticosteroid regimen similar to that used in NASCIS II in the setting of closed head injury.³⁵ It demonstrated increased mortality with steroid use in that population. One must certainly recognize the possibly of an increased mortality risk in patients with SCI as well. However, it must also be noted that Sauerland and colleagues³⁶ performed a large meta-analysis review of approximately 2500 patients treated with preoperative high-dose steroids (>15 mg/kg methylprednisolone) and found no evidence of increased risk for gastrointestinal bleeding, wound complications, pulmonary complications, or death. They suggest that some reports that claim increased complication rates with high-dose steroid use were subject to selection bias.

CONCLUSIONS

In summary, although well-designed and executed studies have been performed, they have failed to demonstrate convincingly a beneficial effect of methylprednisolone or other corticosteroids in the management of SCI. Post hoc analyses have been used to argue a small effect on motor function in three randomized trials. However, all of these analyses contain significant flaws rendering conclusions of efficacy dubious. These observations have led two national organizations to publish guidelines recommending methylprednisolone administration as a treatment option rather than as a standard of care or recommended treatment (Table 1-2).^15,37

TABLE 1-2 Summary of Recommendations

^*Although data are available from level 1 publications, this item was not the main focus of these articles and the studies were not necessarily powered to address this issue.