Friday, May 3, 2013

Steroids for Spinal Cord Injury: is it beneficial?


Summary

  1. There is no evidence to support the use of steroids in the management of spinal cord injury.

  2. Administration of high dose steroids to trauma patients can have significant adverse effects on patient outcome.

  3. There is evidence to suggest a worse outcome in patients with penetrating injury who receive steroids.

  4. There are no studies of the use of steroid therapy in the paediatric population.

Summary of Relevant Studies

Prospective, Randomized Controlled Trials

StudyDesignOutcome
NASCIS 1USA, 1984Prospective, randomized, double-blind.
Methylprednisolone, 2 dose regimens.
Negative.
NASCIS 2 USA, 1990Prospective, randomized, double-blind.
Methylprednisolone, Naloxone, Placebo
Negative.
Otani Japan, 1994Prospective, randomized*, un-blinded.
Methylprednisolone vs ?Placebo
Negative.
NASCIS 3 USA, 1997Prospective, randomized, double-blind.
Methylprednisolone, Tirilizad
Negative.
Petitjean, Pointillart France, 1998, 2000Prospective, randomized, single-blind.
Methylprednisolone, Nimodipine, Placebo
Negative.

Retrospective Studies

StudyDesignOutcome
Prendergast USA, 1994RetrospectiveNegative
Gerhart USA, 1995RetrospectiveNegative
George USA, 1995RetrospectiveNegative
Gerndt USA, 1995RetrospectiveN/A (looked at adverse events)
Poynton Ireland, 1997RetrospectiveNegative

Prospective, Randomized Studies

NASCIS 1, USA 1984

Design:Multicentre, prospective, randomized, double-blind trial. Patients: 330 patients with acute spinal cord injury. 179 6-month follow-up. Randomization:
  • Treatment 1: 100mg bolus methylprednisolone, then 25mg every 6 hours for 10 days.
  • Treatment 2: 1000mg bolus methylprednisolone, then 250mg every 6 hours for 10 days.
Neurologic outcome: No significant differences in neurologic outcome between the two groups. Safety & Adverse events: Statistically significant increase in wound infections in high dose group (9.3% vs 2.6%, p=0.01) Non-significant trend towards increased incidence of sepsis, pulmonary embolism and death in the high-dose group. CommentsAnimal studies subsequently determined that dose of methylprednisolone used in NASCIS 1 was below the therapeutic threshold to observe any potential beneficial effect. Subsequent trials used higher dosing regimens. Nevertheless, NASCIS I demonstrated an increased incidence of adverse events even at these dose levels.

NASCIS 2, USA 1990

Design Multicentre, prospective, randomized, double-blind trial. Patients: 487 patients with acute spinal cord injury. Follow-up of 427 patients (95%) at 1 year (deaths excluded). Exclusions: Injuries below L1 Children Randomization:
  • Treatment 1: Methyprednisolone 30 mg/kg bolus, then 5.4 mg/kg/hr for 23 hours
  • Treatment 2: Naloxone 5.4 mg/kg bolus, then 4.5 mg/kg/hr for 23 hours
  • Treatment 3: Placebo
Neurologic outcome: Patients were examined at admission, 6 weeks, 6 months and 1 year. Motor strength was measured using the American Spine Injury Association (ASIA) 0-5 scale in 14 muscle groups, giving a maximum possible score of 70. Pin prick & touch sensation were assessed in 29 dermatomes. Maximum sensory score was 58. Analysis only used scores from the right side of the body. There was no mention of left-sided scores, although both sides were examined.
There was no difference in motor score between treatment groups at any time point. There was a statistically significant improvement in pinprick (3.4/58) and light touch (3.8/58) scores at 6 months which was lost at 1 year.
Safety & Adverse events: Wound infection and pulmonary embolus were doubled in the steroid group (non-significant, study not sufficiently powered). Post-hoc analyses All the reported positive results from the NASCIS 2 trial are from post-hox subgroup analysis.
When patients were stratified by time to treatment, patients receiving steroids within 8 hours had a statistically significant improvement of 5 points on the motor score at 6 months and 1 year (p=0.03). Patients treated with steroids more than 8 hours after injury had a worse neurological outcome, but this did not achieve significance.
Comments:
Statistical significance:
NASCIS 2 was a negative study. The Level 1 evidence from this data is that there is no difference between methylprednisolone and placebo in the outcome of spinal cord injury. Data from post-hoc analyses cannot be classed as Level 1 or 2 evidence (or even level 3).
The 8-hour cut-off point for the post-hoc analysis appears completely arbitrary. This splits the patients into 183 (38%) that randomized before 8 hours and 304 after. Control patients in the pre-8 hour administration category had significantly worse outcomes than those in the post-8 hour category, a difference which could entirely account for the statistically significant improvement seen in the steroid group.
NASCIS 2 allows for 78 potential discrete post-hoc subgroup analyses based on time of administration. By chance, 1 in 20 of these would be expected to be statistically significant at a p of 0.05. Furthermore, the NASCIS 2 statistical analysis includes over 60 t-tests for comparing neurological outcomes. There are no corrections for multiple comparisons, and no analysis of variance or multivariate statistical techniques were employed. Additionally, much of the data is thought to be non-parametric, and hence the t-test is not appropriate. It is unlikely that any statistical significance would be observed if correct statistical methodology was used.
Clinical significance:
The post-hoc analysis identified a statistically significant improvement of 5 motor points improvement at 1 year. Is this clinically significant? An improvement of 5 motor points in one muscle group is unlikely - but even so would not confer any increased functionality on a spinal cord injured patient. Similarly, and increase of 1 motor point across 5 different muscle groups will also have little impact on functional ability and independence. The mean motor and sensory scores between steroid and placebo groups in NASCIS 2 are shown below:

Otani, Japan 1994

Design Multicentre, prospective, randomized*, un-blinded trial. Patients: 158 patients. 116 of 117 patients wiere available for 6-month follow-up. Exclusions: Patients <15 or >65. Randomization:
  • Treatment 1: Methyprednisolone 30 mg/kg bolus then 5.4 mg/kg/hr for 23 hours
  • Treatment 2: Routine medical management. This included in some patients, steroid therapy, subsequently excluding 29 patients.


Neurologic outcome: Neurological assessment was almost identical to the NASCIS 2 study.
There was no significant difference in outcome between treatment groups.
Safety & Adverse events: There was a trend towards an increase in septic complications in the methylprednisolone group (66% vs. 45%) although this did not achieve statistical significance.
Post-hoc Analyses: Post-hoc subgroup analyses identified a statistically significant increase in the number of patients who experienced an improvement in sensation for those patients who received steroids (68% vs 32%).
Comments: 41 Exclusions after randomization (primarily for protocol violations) make it impossible to clearly assess this study as Level 1 evidence.
Admission characteristics between the two groups were different in terms of severity of spinal cord injury (Frankel grade) and neurologic scores for motor and pinprick.
If the post-hoc analyses identified more patients in the steroid arm experienced a sensory improvement, then more patients in the control arm must have experienced a motor improvement, as overall neurological outcome was unchanged between the two groups.

NASCIS 3, USA 1997

Design Multicentre, prospective, randomized, double-blind trial. No placebo arm.
Patients: 499 patients with acute spinal cord injury. Follow-up of 459 patients (92%) at 1 year (439 / 88% including deaths). Treatment initialized within 8 hours. All patients received a bolus of 20-40 mg/kg methylprednisolone.
Exclusions: Children
Randomization:
  • Treatment 1: Methyprednisolone 5.4 mg/kg/hr for 24 hours
  • Treatment 2: Methyprednisolone 5.4 mg/kg/hr for 48 hours
  • Treatment 3: Tirilizad 2.5 mg/kg every 6 hours for 48 hours
Neurologic outcome: Patients were examined at admission, 6 weeks, 6 months and 1 year. Motor strength was measured using (ASIA) 0-5 scale in 15 muscle groups, although the extra group was subsequently excluded, giving a maximum possible score of 70. Pin prick & touch sensation were assessed in 29 dermatomes. Maximum sensory score was 58. Disability was scored using the Functional Independence Measure (FIM) to try to interpret the functional significance of any improvement in motor score. Only right sided deep pain & pressure, left sided light touch & motor scores used in analyses, despite both being examined.
There was no significant difference in motor score between treatment groups at any time point.
Safety & Adverse events: Mortality due to respiratory complications was 6 times higher in the 48-hour group (p=0.056). There was a 2x increase in the incidence of severe pneumonia and a 4x increase in severe sepsis in the 48-hour group compared to the 24-hour group. This was not statistically significant but the study was underpowered for this analysis.
Post-hoc analyses All the reported positive results from the NASCIS 3 trial are from post-hox subgroup analysis.
Patients were stratified by time to treatment into those treated less than 3 hours from injury and 3-8 hours from injury. There was no difference in outcome for those patients receiving steroids within 3 hours of injury. Only for patients treated between 3 and 8 hours of injury was there a 5-point improvement in motor scores at 1 year (p=0.053). There was no difference in sensory outcome. Disability as measured by FIM was also unchanged at 1 year (there was a statistically significant improvement in sphincter control of 1/14 points at 6 months which was lost at 1 year).
Comments:
Statistical significance:
NASCIS 3 was a negative study. As with NASCIS 2, The Level 1 evidence from this data is that there is no difference between methylprednisolone and placebo in the outcome of spinal cord injury. Data from post-hoc analyses cannot be classed as Level 1 or 2 evidence (or even level 3).
The post-hoc analysis of subgroups receiving steroids in 0-3 hours or 3-8 hours is completely arbitrary. Patients who received steroids early did not gain any benefit from steroids (70% of the study population). Ensuing statistically significant results are almost certainly due to random events.
NASCIS 3 allows for 36 potential discrete post-hoc subgroup analyses based on time of administration. By chance, 1 in 20 of these would be expected to be statistically significant at a p of 0.05. Again, as with NASCIS 2, NASCIS 3 statistical analysis includes over 100 t-tests for comparing neurological outcomes. There are no corrections for multiple comparisons, and no analysis of variance or multivariate statistical techniques were employed. Additionally, much of the data is thought to be non-parametric, and hence the t-test is not appropriate. It is unlikely that any statistical significance would be observed if correct statistical methodology was used.
Clinical significance:
The post-hoc analysis identified a statistically significant improvement of 5 motor points improvement at 1 year. As with the NASCIS 2 results, the clinical significance of this improvement is doubtful. Indeed, the FIM measurements showed no difference in the level of disability. The mean motor and sensory scores between steroid and placebo groups in NASCIS 3 are shown below:

Petitjean, Pointillard, France 1998, 2000

Design Single centre, prospective, randomized, double-blind trial.
Patients: 106 patients, hospitalised within 8 hours of injury. 100 patients available for follow-up at one year. Policy of early surgery. 76% operated on within 24 hours, 46% within 8 hours of injury.
Exclusions: Patients <15 or >65.
Randomization:
  • Treatment 1: Methyprednisolone 30 mg/kg bolus then 5.4 mg/kg/hr for 23 hours
  • Treatment 2: Nimodipine 0.5 mg/kg/hr for 2 hours then 0.03 mg/kg/hr for 7 days
  • Treatment 3: Methylprednisolone & Nimodipine
  • Treatment 4: Placebo
Neurologic outcome: Patients were examined at admission and 1 year by a trained neurologist blinded to the treatment..
There was no significant difference in ASIA score between treatment groups at one year. There was also no significant difference overall in those patients who received steroids (54) and those who did not (52). Two-way ANOVA showed no evidence of interaction between methylprednisolone and nimodipine.
Safety & Adverse events: There was a trend towards an increase in septic complications in the methylprednisolone group (66% vs. 45%) although this did not achieve statistical significance.
Comments: Not only was there no significant difference in outcome in this study, but not even a trend to improved outcome in this study.

Retrospective Studies

Prendergast, USA 1994

Design: Retrospective review
Patients: 54 patients with acute spinal cord injury. 31 patients with penetrating injury 29 patients received steroid therapy
Neurologic 'Outcome:' Neurologic assessments were performed at 4 days and 1, 2, 4 & 8 weeks.
There was no difference in neurologic improvement in blunt trauma patients.
Patients with penetrating trauma who received steroids had a significantly worse outcome within the first week of injury which persisted throughout the study (p<0.05).

Gerhart, USA 1995

Design: Retrospective, population-based study
Patients: 363 spinal cord injury survivors from two time periods, pre- and post- NASCIS 2. 188 with NASCIS protocol, 90 no steroids.
Neurologic Outcome: Frankel grade was assessed by neurological documentation at admission and at the end of rehabilitation or discharge home.
There was no significant difference in Frankel grade between those patients who received steroids and those who did not.

George, USA 1995

Design: Retrospective review
Patients: 130 patients with acute spinal cord injury. 9 patients with penetrating injury. 75 patients received steroid therapy. The steroid group was younger and has lower Injury Severity scores
Neurologic Outcome: There was no difference in disability outcome between those patients who received steroids and those who did not.
Safety & Adverse events: There was a trend towards an increase in infectious complications in the methylprednisolone group although this did not achieve statistical significance.
Comments: There was equivalent outcome in steroid and no-steroid groups, despite the control group being older and more severely injured.

Poynton, Ireland 1995

Design: Retrospective review
Patients: 71 consecutive patients with acute spinal cord injury. 63 patients available for follow-up, mean 30 months. Patients admitted more than 8 hours following injury were not given steroids and were used as controls.
Neurologic Outcome: Patients were assessed with ASIA scores on admission, on transfer to rehabilitation and at follow-up.
There was no difference in disability outcome between those patients who received steroids and those who did not.
Safety & Adverse events: There was a trend towards an increase in infectious complications in the methylprednisolone group although this did not achieve statistical significance.
Comments: There was equivalent outcome in steroid and no-steroid groups, despite the control group being older and more severely injured.

Levy, USA 1996

Design: Retrospective review of penetrating injury patients.
Patients: 252 patients with single penetrating spinal cord injury. 71 received steroid therapy, 55 NASCIS protocol.
Neurologic Outcome: Frankel Scores at admissionm & discharge from rehabilitation. Discharge assessment conducted at mean of 174 days. No difference in neurological outcome.

Gerndt, USA 1997

Design: Retrospective review. Study was aimed towards idntifying adverse events from steroid therapy
Patients: 231 acute blunt spinal cord injured patients. 91 received steroid therapy.
Neurologic Outcome: Not detailed
Safety & Adverse events: There was significant increase in the incidence of pneumonia and in the duration of ventilation and ICU stay in the methylprednisolone group.
Comments: There was equivalent outcome in steroid and no-steroid groups, despite the control group being older and more severely injured.

Heary, USA 1997

Design: Retrospective review of penetrating trauma patients (gunshot wounds).
Patients: 254 patients with gunshot wounds to the spine or spinal cord. 61 received steroid therapy (methylprednisolone or dexamethason).
Neurologic Outcome: ASIA grade & Frankel scores. Mean follow-up was 53 months. No statistically significant difference in neurologic outcome at any stage.
Safety & Adverse events: There was significant increase in the incidence of infections in the steroid group. The dexamethasone group had a statistically significant increase in the number of gastrointestinal complications, and the methylprednisolone group had a statistically significant increase in the incidence of pancreatitis.

Safety and Adverse Effects of Steroids

Potential adverse effects of high-dose steroid administration in trauma patients include:
  • Increased incidence of infectious and septic complications
  • Increased incidence & severity of respiratory complications
  • Increased incidence of pulmonary embolism
  • Worsening of head injury outcome
  • Increased incidence of gastrointestinal haemorrhage
  • Increased incidence of pancreatitis
  • Possibility of missed hollow viscus injury due to 'masking' of abdominal signs