The University of Chicago Journal Club
Participating faculty: Roitberg B, Frim D, Lam S.
Residents: Monim-Mansour N; Dey M; Khader-Eliyas J; Hobbs J; Stamates M
Topic: Methylprednisolone for spinal cord injury. Is the controversy over?Introduction: March 2013 saw the publication of new guidelines for the treatment of acute spinal cord injury, which included a clear recommendation against the use of corticosteroids, and specifically high dose methylprednisolone, for this indication. Here we review a few milestones from acceptance to rejection of methylprednisolone for acute spinal cord injury. Among the many publications on the topic we chose three. First - the NASCIS III trial, that refers to previous NASCIS studies and to me personally was a trigger to start routinely using methylprednisolone in acute spinal cord injury situations. Then we review the recommendations from the Neurosurgery March special issue – the article on “Pharmacological Therapy for Acute Spinal Cord Injury”. This guideline is unusually firm in rejecting methylprednisolone as a treatment of acute spinal cord injury. Finally, we review a large study of the use of methylprednisolone for head injury, which served as a source of data about the risks of high dose steroids.
Article 1: Administration of Methylprednisolone for 24-48 Hours or Tirilazad Mesylate for 48 Hours in the Treatment of Acute Spinal Cord Injury: Results of the Third National Acute Spinal Cord Injury Randomized Controlled Trial. Bracken MB, Shepard MJ, Holford TR, Leo-Summers L, Aldrich EF, Fazl M, Fehlings M, Herr DL, Hitchon PW, Marshall LF, Nockels RP, Pascale V, Perot PL Jr, Piepmeier J, Sonntag VK, Wagner F, Wilberger JE, Winn HR, Young W. JAMA, 1997;277;1597-1604.
Hobbs: This article, also known as NASCIS III, together with previous NASCIS studies, led to widespread adoption of methylprednisolone (MP) for acute spinal cord injury (SCI). This was a double-blinded, randomized clinical trial with 499 patients that presented to one of sixteen NASCIS centers in North America within 8 hours of injury with a goal of randomization within 6 hours of injury and drug administration within 8 hours of injury. After diagnosis of acute spinal cord injury, all patients received a MP bolus (20-40mg/kg) and then randomized into one of three treatment arms (MP 5.4 mg/kg/hr x 24 hours, MP 5.4 mg/kg/hr x 48 hours, or tirilazad mesylate 2.5mg/kg every 6 hours x 48 hours). Tirilazad was chosen as control because it is anon-glucocorticoid 21-amino steroid that had evidence of benefit in randomized trials for patients with stroke and subarachnoid hemorrhage. The authors evaluated patients’ neurologic function by comparing baseline examinations upon admission, then at follow up at 6 weeks and 6 months. They evaluated motor function, sensory function, and Functional Independence Measure (FIM). The primary endpoint was overall change in neurological function. Demographically, most patients were between 14-34 years old, male, conscious on admission, no overt comorbidities that would confound results, and who were not already on methylprednisolone. 49.7% had complete spinal cord injuries. Of note, the tirilazad group had worse neurologic function upon admission compared to MP groups.
Subgroup analysis was performed investigating drug administration within 3 hours vs. 3-8 hours after injury. Motor function analysis did not show significant differences between 24hr vs. 48hr of MP administration at 6 weeks (change in scores 9.0 vs. 11.8 [P=.09]) or 6 months (13.4 vs. 16.8 [P=.07]) when all patient’s considered. However, upon distinguishing time of drug administration, it found that at 3-8hours after injury, the 48 hour methylprednisolone had significant improvement of motor function at 6 weeks and 6 months (7.6 vs. 12.5 [P=.04] and 11.2 vs. 17.6 [P=.01]). Tirilazad group improved at a rate between the two MP groups. Sensory function followed similar patterns. FIM was only significant in self care [P=.02] and sphincter control [P=.01] after 48 hours of MP. Given these findings, the authors suggested 48hours of MP is beneficial if initial drug administration is between 3-8 hours after injury. If administration of MP is within 3 hours of injury, 24 hours of MP is appropriate. Of note, the rate of sever sepsis and pneumonia were higher in the 48 hour MP group compared to the 24 hour MP group (2.8% vs. 0.6% [P=.07] and 5.8% vs. 2.6% [P=.02], though overall morbidity and mortality was no different.
There are a few areas of contention regarding this study. All participants received initial MP dosing, confounding the findings in the trilizad groups (i.e. how much treatment effect is simply due to initial MP dose). There was no placebo group in this paper (though in NASCIS II, there was a placebo group present that was found to be inferior, it still weakens the current study as the overall benefit of treatment is assumed without a control group). No imaging was included, so we do not know the extent of the initial injuries in those patients that actually displayed neurological improvement. Most importantly, the subgroup analysis of investigating the 3-8 hour group is post-hoc analysis. Their initial findings were inconclusive regarding 24 hour vs. 48 hour MP use, since they originally only used an 8 hour timeline. It wasn’t until the post-hoc analysis that they found the time- to-treat correlations. This decreased the strength of the study.
This NASCIS studies are a fundamental part of neurosurgical literature that led to practice changes regarding treatment of acute spinal cord injury. However controversy persisted, and whether or not to implement their suggestions was up to the individual practitioner and should have been used on a case-by-case basis, with the understanding of the increased risks associated with MP use.
Frim: Though retrospective, the 3-8 hour window was not arbitrary. Hall et al from Upjohn studied animal models, where effect on outcome was seen 3-8 hours after injury, but earlier was better. Interestingly, pre-injury methylprednisolone was potentially detrimental in a similar model.
Monim Mansour: NASCIS III had no placebo group, like many surgical studies. No data was collected on imaging, which is always a part of our evaluation of the patient with acute spinal cord injury. Therefore it was always difficult to be sure how this study relates to our actual practice.
Roitberg: The practice of medicine is difficult – we face the need to treat very ill or gravely injured patients, who are desperate for help. Physicians in turn are eager to find clarity and guidelines that will reduce the uncertainty inherent in our practice. Large randomized trials are often seen as great beacons that light our way; their findings and recommendations tend to be incorporated into practice despite significant limitations and often justified criticisms. Methylprednisolone administration is a pharmacological intervention that ER physicians can start before the arrival of the neurosurgeon, and feel that they have done something good for the patient. This is a powerful incentive to believe that MP is helpful and discount evidence of harm. A thought crossed my mind – the physicians who order MP in the acute period are not always the same ones who manage the long term medical complications in that patient.
Article 2: Pharmacological Therapy for Acute Spinal Cord Injury. Hurlbert RJ, Hadley MN, Walters BC, Aarabi B, Dhall SS, Gelb DE, Rozzelle CJ, Ryken TC, Theodore N. Neurosurgery. 2013 Mar;72 Suppl 2:93-105.
Stamates: This article by Hurlbert et al is part of a wider supplement in Neurosurgery regarding recommendations for the treatment of acute spinal cord injury, and an update of the the 2002 guidelines released by AANS and CNS on the use of methylprednisolone(MP) and GM-1 gangliosides in the setting of acute SCI. The authors performed a PubMed search from 1966-2011 on the topic of steroids & GM-1 ganglioside in the setting of SCI or neurological deficit with approximately 680,000 citations acquired; 641 of these titles & abstracts were reviewed. The original articles supporting the use of MP were NASCIS I, II and III. This review revisits the NASCIS trials and discusses their limitations as well as the evidence they provide for complications of MP use. The benefit in motor score in NASCIS II was designated by the review group as Class III, because it was discovered only in post-hoc analysis of a partial dataset. However, complications like wound infection, GI hemorrhage and thrombophlebitis were more common in the MP-treated group. The authors indicate that NASCIS were prospective, blinded RCTs that nevertheless failed to show Class I evidence supporting MP use in the treatment of 980 SCI patients (with 280 control subjects). Additional weaknesses of NASCIS were the lack of a placebo group in NASCIS III, and motor score reporting of only right-sided extremities.
The remainder of this extensive review article quotes multiple studies that used a protocol similar to NASCIS—8 hr onset of treatment, 24 or 48 hr duration and high-dose methylprednisolone; however, the benefits were mixed and not clearly clinically significant (motor v. sensory improvement.) The benefits of MP were also not robustly reproducible across multiple populations of patients with acute spinal cord injury. Despite the lack of similarities in benefit, the harm of steroid treatment was seen in all studies, including: acute pneumonia, increase in ventilated days, longer ICU stay, and higher incidence of UTI.
Lastly, the authors observed that no neurological condition score gives the “best” clinical picture of outcomes, and we as a community continue to use multiple scoring scales, like ASIA, Frankel scores and the specific motor score developed for NASCIS.
Although extensive and thorough, this review and guideline article may itself be criticized. The lead author criticized the use of MP for acute spinal cord injury a decade ago (1,2), and thus could be considered a party to the controversy rather than a neutral observer. Nevertheless, the evidence in the many articles presented in this review is enough to strongly suspect that the risk to benefit ratio of MP for spinal cord injury is not beneficial. After years of popularity, MP for spinal cord injury may now be falling out of favor. We now need to go back to more basic and clinical research to find better options for our patients.
Roitberg: This guideline represents expert opinion, supported with review of selected literature. It is different from a Cochrane review in terms of methodology. The review is not presented from a neutral point of view. However, the data in favor of MP use for SCI has never been robust. The concern about side effects has been front and center in many clinical decisions about MP use. This review article changes the focus of the discourse from possible benefit to the harm of MP, and will probably result in declining use of MP by neurosurgeons.
Frim: Is the guideline itself more than Class IV evidence? The group that developed the guideline does not present new data or new analysis. This is a pronouncement of expert opinion, rather than a meta-analysis or a presentation of data pro and con, according to “clinical pathway guidelines."
------------------------- Article 3: Final results of MRC CRASH, a randomized placebo-controlled trial of intravenous corticosteroids in adults with head injury-outcomes at 6 months. CRASH trial collaborators. Edwards P, Arango M, Balica L, Cottingham R, El-Sayed H, Farrell B, Fernandes J, Gogichaisvili T, Golden N, Hartzenberg B, Husain M, Ulloa MI, Jerbi Z, Khamis H, Komolafe E, Laloë V, Lomas G, Ludwig S, Mazairac G, Muñoz Sanchéz Mde L, Nasi L, Olldashi F, Plunkett P, Roberts I, Sandercock P, Shakur H, Soler C, Stocker R, Svoboda P, Trenkler S, Venkataramana NK, Wasserberg J, Yates D, Yutthakasemsunt S; CRASH trial collaborators. Lancet. 2005 Jun 4-10;365(9475):1957-9.
Khader-Eliyas: Corticosteroid Randomization in Adults in Severe Head injury (CRASH) trial was a multi-centered international trial that looked at effectiveness of corticosteroids in head injury. Before this study, literature review had showed slight advantage of using corticosteroids in head injury but the evidence was not compelling. The authors of the study wished to find out if there would be any change in mortality and morbidity with the use of steroids in head injured patients.
The study was conducted at 239 hospitals across 49 countries enrolling subjects for a little more than 5 years. To be included in the study, subjects had to be over 16 years of age presenting within 8 hours of injury with a GCS of 14 or less. Additionally there had to be no indication or contraindication to steroid use. Primary outcome measures included death from any cause within 14 days and death & disability at 6 months. Six-month follow up data was obtained using questionnaires, telephone interviews, home and hospital visits. Data was stratified using injury to randomization time interval and GCS subgroups.
There were more than ten thousand patients (10,008) in the study with an average age of 37 years (SD-17). Mean time to randomization was 3 hours. Though 98% adhered to the treatment allotted, only 83% received at least 24 hours of treatment. Six-month follow up results were available in 9673 patients (96%), 4854 belonged to the steroid arm and 4819 to the placebo group. There was a significant increase in relative risk of death (1.15%) in the steroid treated patient population (CI-95%, P=0.0001) that mirrored the risk at 2 weeks from injury. Risk of death and disability was also increased in the steroid group but did not reach statistical significance. When results were compared between stratified groups (both according to 1. time to randomization and 2. GCS score) no difference in outcome was found.
In summary the MRC CRASH trial was performed with the primary objective to establish the role of corticosteroids in head injury. The study was well designed and executed with excellent follow up percentage. Unfortunately outcomes especially in trauma can be difficult to attribute to one factor in-spite of rigorous randomization. Often an issue with studies performed in multiple centers and countries is the variation seen in standard treatment offered to patients. These can sometime be significant enough to skew results. Also the study does not talk about cause of death in these patients nor attempt to explain the increased mortality and morbidity associated with steroids. Apart from these minor drawbacks, the CRASH study provides evidence against using steroids in head injury thus making one wonder about the need for similar large effort in the context of spinal cord injury.
Roitberg: This large study provided data regarding the additional risk of using steroids, in the head injury population. The higher risk associated with steroid use in this study was a major factor in the recommendation against the use of steroids also in spinal cord injury. Although tempting, such a conclusion is not self-evident. All treatments are based on a balance of risks and benefits, so the higher early mortality in the head injury study may have reflected either lack of benefit specifically for head injury, or a special increase in risk of steroids in head injured patients.
Hobbs: Ito et all (3) demonstrated a higher risk of pneumonia with steroids, in their work from 2009. Looks like the treatment adds expected infectious complications from the immune suppressant effect of steroids, without enough benefit to counterbalance it.
Dey: A key question when deciding whether steroids are good or bad for these patients is the relationship of the outcome and hyperglycemia. Hyperglycemia can lead to a worse outcome by itself. Could we achieve better glucose level control in patients treated with methylprednisolone? Would the risk/benefit ratio change then?
Monim-Mansour: It is difficult to generalize treatment recommendations. There may be some patients who experience a substantial benefit from methylprednisolone, and others who develop serious complications. The average results in a large population obscure the individual effect. Ideally was should be able to find those patients who can benefit the most, and avoid treating those who would develop complications. Maybe we can try to anticipate complications and treat those with strict glucose control, antibiotics, etc.
Frim: After reading articles and guidelines, the question is – what would you do next time in the ER? Would you now try to actively dissuade the ER physicians from starting methylprednisolone even before calling your consultation? Are there populations for whom empirical methylprednisolone makes sense? Some of the complications associated with the use of methylprednisolone occurred in patients with injuries that were treated operatively, making the postoperative course more complicated. Does the risk extend to all groups? For example, the young athletes who had “zingers” and some residual deficits? There is no operation, the patients are younger and presumably can tolerate steroid side effects better. Is there a place for a study of methylprednisolone in selected populations? Would there be support for such a study given the current negative guidelines?
Roitberg: We chose this set of articles for discussion because it is relevant to our practice, but also because it is an example of the poor quality of the science that guides many medical decisions. The issue of MP for spinal cord injury is arguably among the better studied topics. It was the subject of large and expensive trials, and now has been the topic of a dramatic change in the guidelines. Sadly, the original science is far from perfect. As we noted in the discussion of NASCIS III, it had no real control group and no imaging data was included. So, is the data valid? Should we have ever started using MP for spinal cord injury? We do not know. Absence of good evidence that MP works is not evidence that it does not.
Now, we have a new set of recommendations, but the quality of the science is again limited. The new guideline against the use of steroids in acute spinal cord injury is based on two types of data – pointing out the weakness of the data for the use of MP, and additional data on its complications that accumulated in the past decade. The CRASH trial is an example of the latter. In that study, adherence to at least 24 hours of MP was limited, and the applicability of head injury data to spinal cord injury is questionable. We can clearly see that steroids have side effects, but we have always known that.
Better designed studies are needed to properly answer the question of MP efficacy for spinal cord injury. In our discussion we could see the outline of such a study: select patients at low risk for complications and no planned surgery (based on imaging); include proper placebo control group; treat hyperglycemia, use detailed and widely accepted outcome scales. Such as study is important because it can provide key information on efficacy of MP in a best case scenario. If the medication is effective for some patients, but has side effects, we can try to select the patients better and use strategies to reduce side effects. If it is ineffective in the best case scenario, the medication should simply not be used.
1) Hurlbert RJ. The role of steroids in acute spinal cord injury: an evidence-based analysis. Spine (Phila Pa 1976). 2001 Dec 15;26(24 Suppl):S39-46. Review.
2) Hurlbert RJ, Moulton R. Why do you prescribe methylprednisolone for acute spinal cord injury? A Canadian perspective and a position statement. Can J Neurol Sci. 2002 Aug;29(3):236-9.
3) Ito Y, Sugimoto Y, Tomioka M, Kai N, Tanaka M. Does high dose methylprednisolone sodium succinate really improve neurological status in patient with acute cervical cord injury?: a prospective study about neurological recovery and early complications. Spine (Phila Pa 1976). 2009 Sep 15;34(20):2121-4.