Methylprednisolone for the Treatment of Acute Spinal Cord Injury

The published literature pertaining to methylprednisolone (MP) administration for acute spinal cord injury has recently and comprehensively been reviewed by the American Association of Neurological Surgeons/Congress of Neurological Surgeons Joint Section on Disorders of the Spine and Peripheral Nerves as part of their Guidelines for the Management of Acute Cervical Spine and Spinal Cord Injuries.1 For a more detailed analysis, the audience is referred to that publication. There have been 5 prospective randomized studies published in 8 articles examining the effect of MP on neurological recovery following acute spinal cord injury (SCI).2-9 Three of these studies comprise the work of the National Acute Spinal Cord Injury Study group (NASCIS), a fourth from Bordeaux Cedex, France, and a fifth translated from a Japanese journal. In all 5, the primary outcome analyses were indisputably negative. Three class II studies10-12 and 10 class III studies13-22 have examined the effect of MP on acute SCI, the majority of which (9/13) have failed to show clinical benefit. Taken in perspective then, the literature appears quite decisive in that 100% of the class I data, 100% of the class II data, and 60% of the class III data have failed to establish a benefit from MP administration. Indeed, the literature is much more robust in establishing harmful effects from the administration of this drug. Where is the controversy then? Why is it that in 2013 a debate apparently still rages with respect to MP use in SCI? Is it because reasonable evidence can be uncovered if one digs deep enough? Is it because money is to be made? Or is it a misguided attempt to provide some type of treatment and hope within the devastating circumstances of an acute SCI? Let's begin by looking in more detail at the few "positive" studies that seem to have gained immortal favor and acclaim, so that we might properly evaluate their miraculous evidence. CLASS I "POSITIVE" DATA Steroid proponents like my esteemed colleague Dr Fehlings have capitalized on "positive" results from both the NASCIS II and III trials, happy to let the scientific rigor of randomized trial data dazzle and obscure the highly biased retrospective (post hoc) analyses best characterized as fishing expeditions. Intensive massage of the original data set becomes necessary to demonstrate a beneficial effect from MP. If we take NASCIS II as an example, within the original research hypothesis there was no sign that MP administered to patients within 12 hours of SCI improved neurological function in any of the primary outcome measures.4,5 These included motor, light touch, and pinprick scores, American Spinal Injury Association (ASIA) grade, and propensity to change from -plegic to -paretic or from quad- to para-. However, if only patients treated within 8 hours from the time of injury were studied (n = 66 instead of the original 161), pinprick, light touch, and motor scores showed statistically greater improvement in the MP group than in controls at 6 weeks and 6 months after injury (Figure 1). Findings worthy of publication in the New England Journal of Medicine.4 Or were they? Beyond the NEJM 6-month report, statistical significance was lost for both sensory modalities at 1 year, remaining only for motor improvement. In placing these motor results in perspective, however, one must remember that this is a retrospective analysis of <40% of the original study population, at an arbitrary 8-hour cutoff neither previously proven nor physiological in nature, incompletely reported data (right-sided motor scores only), of a noncompelling magnitude and negligible practical relevance (Figure 1), and which, even in this highly selected population, all other outcome measures were negative at a year; caveats that steroid proponents like Dr Fehlings stubbornly forget to acknowledge. But in reality, these selected results hardly have the glamour or impact of a randomized clinical trial. Indeed, the isolated post hoc partial motor scores reported as significantly improved are more akin to an event of random chance.FIGURE 1: Motor and sensory scores over the 1-year follow-up period of NASCIS II. Light touch outcomes were statistically better in MP patients compared with placebo at 3 weeks and 6 months in comparison with baseline. However, the effect was lost at 1 year. Pinprick results (not shown) demonstrated a similar trend. When compared with baseline, motor scores were also statistically increased in MP patients vs placebo even at 1-year follow-up (P = .03). However, examining the data in graphical form (not provided in the NASCIS publications), one can appreciate how statistical significance does not necessarily represent clinical significance; the curves overlap substantially and are not compellingly different from each other. Furthermore, these "positive" results are from a small subset of the study population discerned through post hoc analyses, and only one-half of the available motor data were reported. NASCIS, National Acute Spinal Cord Injury Study; MP, methylprednisolone; sem, standard error of the mean.There will be little surprise to the audience that the "positive" results from NASCIS III fall into exactly the same category. All primary preplanned comparisons were negative. The only interesting results came from 151 (30%) of the original 499 patients, 80 of whom were treated with 48 hours of MP between 3 and 8 hours of injury, yet another arbitrary nonphysiological "therapeutic" window in which sensory scores, propensity to change ASIA grade, and functional outcomes were negative. Motor scores were "significantly" (statistically) improved at 6 months. However, by normal standards, this statistical significance was lost at 1 year (P > .05). Hence, in NASCIS III even the retrospective analyses were universally negative at 1 year. The only other "class I" published data proposed to show a beneficial effect of MP in acute SCI appeared in the Japanese journal Sekitsui Sekzui in 1994.8 Otani and colleagues reported their experience with 117 patients (70 MP, 47 control) randomly assigned to receive NASCIS II doses of MP or saline placebo. An undisclosed number of patients in the control arm were allowed to receive "discretionary" steroids. Six-month follow-up was provided. Preplanned primary outcomes of ASIA motor, pinprick, and light touch scores were no different between the treatment groups. Retrospective analysis of the data set suggested that a greater percentage of patients in the MP group demonstrated "some degree" of neurological improvement, statistically significant for sensory improvement but not motor function. For example, 30 of 44 patients (68%) demonstrated an improvement in pinprick scores compared with only 6 of 19 patients (32%) in the control group (P < .05). However, when the audience thinks carefully about this, they will realize that if twice as many patients in the MP group improved compared with controls, but the primary outcome measure (mean score) was the SAME for each group, then it follows that the magnitude of recovery in the 6 control patients who improved was far greater than the magnitude of recovery seen in the 30 MP patients who improved. Not only is this counterintuitive, but it is also nonsensical, and demonstrates one of the biggest pitfalls of retrospective subanalyses in prospective clinical trials: failure to consider the other side of the coin. CLASS II "POSITIVE" DATA None. Three class II studies representing cumulative experience in 119 patients have found no beneficial effects of MP in acute SCI.10-12 CLASS III "POSITIVE" DATA Six of 10 retrospective class III studies reported no benefit to MP administration in acute SCI. Four studies declared benefit to steroid use, but none make a compelling argument. All suffer from inherent limitations of retrospective case series, including selection bias, nonstandardized treatments, and suboptimal outcome measures. As an example, in 1993, Kiwerski published what is still the largest retrospective review on 620 patients with acute SCI treated in Poland between the years of 1976 and 1991.13 Two hundred ninety patients received MP while 330 did not. However, the dose given to MP patients varied according to age, weight, medical conditions, and the attending physician. The usual dose was in the order of 8 mg 3 times a day for several days, substantially less than the NASCIS II 30 mg/kg loading dose and 5.4 mg/kg/h maintenance dose. The details of the neurological examinations were not disclosed; patients were simply classified as "improved" or "not improved." Kiwerski observed that, across all age groups of patients, a higher percentage of those treated with MP improved than those who did not receive MP. The greater the age, the greater the effect. What Kiwerski failed to point out, though, was that within his cohort of patients mortality rates were consistently twice as high in the non-MP patients, ranging from 15% to nearly 40%. Higher mortality rates were observed in older patients. Hence, it is clear that the "control" patients (no MP) sustained more severe systemic injuries (likely due to selection bias) and were therefore in a less favorable prognostic category to begin with. In summary then, if one examines the available literature there exists no class I evidence that MP at any dose is beneficial in the treatment of acute SCI. Small-scale benefits have been reported only in post hoc/retrospective (class III) analyses. These small-scale benefits are inconsistent from 1 study to the next and are isolated events, lacking confirmation from other outcome measures within the same group of patients. REBUTTAL It is appropriate at this time, on behalf of my worthy opponent, to disclose that he is an investigator and author for the NASCIS III trial. I would like to also refresh the audience's memory as to what exactly the NASCIS III authors concluded about their study. I quote: "Patients started [on MP] between 3 and 8 hours….should be maintained for 48 hours." But when it came to adverse events, the same authors concluded "Possible increased mortality rates….call for caution." On the one hand, a definitive practice recommendation advocating for aggressive steroid use, and, on the other, a passing reference to the possibility of death. A strong recommendation for 48-hour steroid administration based on a single observation in a post hoc analysis of a small subset of patients. Yet only vague allusion to mortality rates 6 times higher across the entire cohort of 48-hour MP patients compared with controls–a primary outcome measure. The audience will be interested to learn that the strong recommendation pertaining to neurological benefit was based on a post hoc significance level of P = .053, while the watered-down warning of death was based on an a priori significance level of P = .056. This statistical evidence differs by three-tenths of a percentage point; significant in a retrospective subanalysis, yet not significant in a preplanned comparison of an entire study population. If anyone is beginning to feel like this might represent biased reporting, I would be inclined to agree. To my colleague opposite me and the other NASCIS III investigators I would say "Shame on you." More recently, Dr Fehlings has published his work on the surgical timing of decompression in acute spinal cord injury (STASCIS), a noteworthy study providing evidence toward early decompression of the spinal cord in the setting of acute traumatic injury. For reasons as of yet undisclosed, he and his coauthors have used their data to make yet another argument for steroid administration, claiming that "The combination of early surgery and steroids results in the best neurological outcomes with the lowest rate of complications" based on a multivariate analysis adjusted for preoperative neurological status and steroid administration. Unfortunately, we (the readers) are denied seeing the actual neurological scores for ourselves, so it is impossible to estimate their magnitude or validity. However, Dr Fehlings fails to mention the fact that significantly more patients administered MP underwent early surgery compared with controls. Hence, the lower incidence of complications observed in the MP group may simply be related to selection bias, the nemesis of most nonrandomized studies. In the same interest of citing unpublished STASCIS data to bolster the position for MP use in SCI, I would like to share some unpublished data of my own. For a number of years, the Rick Hansen Institute has been collecting registry data on patients with SCI across Canada. Part of this registry involves a data field indicating whether or not MP was administered as part of the acute care management. The database involves 31 major trauma and rehabilitation hospitals in 15 cities across the country. We recently extracted MP status and ASIA motor scores in the database and came up with results for 855 patients over the 5-year period from 2006 to 2011 (Figure 2). It is clear from a national cohort larger than (and likely including many of) Dr Fehling's STASCIS patients that MP administration was not associated with improved ASIA motor outcomes.FIGURE 2: ASIA motor improvement in 855 patients from the Rick Hansen Institute spinal cord injury registry, a national Canadian database (unpublished data). Improvement was calculated by subtracting the patients' hospital discharge ASIA motor score from their hospital admission ASIA motor score. The bar graphs indicate that there was clearly no evidence of motor benefit from MP administration in this large cohort of SCI patients. Similar trends were observed when patients were further stratified for ASIA grade and comorbidity index (not shown). MP, methylprednisolone; SCI, spinal cord injury; ASIA, American Spinal Injury Association.Finally, it is worthwhile to highlight the most consistent effects reported with high-dose steroid administration: complications. Level I evidence from randomized clinical trials has shown higher risks of wound and respiratory infections, hyperglycemia requiring insulin administration, and gastrointestinal hemorrhage in patients treated with MP for acute SCI (P < .05).2,9,23 Level II evidence demonstrates a higher incidence of respiratory, urinary, and wound infections (P = .028) and steroid-induced myopathy (80% vs 0%) in high-dose steroid patients.11,12 Similarly, level III publications indicate MP-treated SCI patients experience statistically higher rates of infection (all types), respiratory infection, hyperglycemia requiring insulin infusion,22 pneumonia, increased ventilator days, and prolonged intensive care unit stay.24 Although not directly related to SCI, the CRASH trial results are interesting because NASCIS III doses of MP were administered in 239 hospitals across 49 different countries to head-injured patients. Enrollment was targeted for 20000 but the study was terminated early by the ethics committee when an interim analysis of 10008 patients revealed higher death rates in the MP-treated patients compared with saline placebo. For every 29 patients treated with 48-hour MP, one died of drug-related morbidity (P = .001). The evidence of morbidity and mortality from high-dose steroids is more consistently reported than any evidence of therapeutic benefit. In summary, there should be no debate at all with respect to MP administration in SCI. Even through analyses of literature now span 30 years, the evidence for a therapeutic effect remains sporadic, inconsistent, of marginal magnitude, and the result of hundreds of retrospective comparisons, hence, likely representing random chance alone. The heightened risks of complication including infection and death are very real and have been substantiated on numerous occasions. If, like Dr Fehlings, for some obscure reason you continue to prescribe MP for acute SCI, then you are not only risking your patient's health, you are risking their lives. More promising treatment modalities such as Minocycline, Riluzole, and stem-cell therapy are on the horizon. We owe it to ourselves and our patients to focus on these new strategies and leave MP behind, allowing it to become, as it deserves, of historical interest only. Disclosure The author has no personal, financial, or institutional interest in any of the drugs, materials, or devices described in this article.

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.