(Stroke. 2001;32:2433.)
© 2001 American Heart Association, Inc.
Comments, Opinions, and Reviews |
From the Department of Neurology (J.H., M.V.), the Department of Clinical Epidemiology and Biostatistics (R.J.d.H.), and the Department of Medical Informatics (M.L.), Academical Medical Center, University of Amsterdam, and the Department of Animal Physiology (P.G.M.L.), University of Groningen, Groningen, the Netherlands.
Correspondence to J. Horn, MD, Department of Neurology, AMC, PO Box 22700, 1100 DE, Amsterdam, Netherlands. E-mail jhorn{at}wxs.nl
| Abstract |
|---|
|
|
|---|
Methods We performed a systematic review of animal experiments with nimodipine in focal cerebral ischemia. Studies were identified by searching Medline and Embase. We assessed whether these studies showed a beneficial effect of active treatment. In-depth analyses were performed on infarct size and amount of edema, and subgroup analyses were performed on the length of the time window to the initiation of treatment and the methodological quality of the studies.
Results Of 225 identified articles, 20 studies were included. The methodological quality of the studies was poor. Of the included studies, 50% were in favor of nimodipine. In-depth analyses showed statistically significant effects in favor of treatment (10 studies). No influence of the length of time to the initiation of treatment or of the methodological quality on the results was found.
Conclusions We conclude that the results of this review did not show convincing evidence to substantiate the decision to perform trials with nimodipine in large numbers of patients. There were no differences between the results of the animal experiments and clinical studies. Surprisingly, we found that animal experiments and clinical studies ran simultaneously.
Key Words: animal models calcium channel blockers cerebrovascular disorders nimodipine meta-analysis
| Introduction |
|---|
|
|
|---|
Similar discrepancies between initial successful experimental animal data and lack of effect in clinical application have been reported for a variety of potentially neuroprotective drugs. These agents either evoked unacceptable adverse reactions or did not have any beneficial effect.912 These disappointing results have raised doubts about the interpretation and validity of the results of animal stroke models with respect to subsequent clinical research.
The aim of the present investigation was to determine whether the evidence from animal experiments was in favor of nimodipine, which would mean that these models were inadequate in the prediction of treatment response in patients or that the data were inconclusive, which would mean that the clinical trials were based on insufficient evidence.
| Materials and Methods |
|---|
|
|
|---|
Studies were included if they fulfilled the following criteria: (1) the study assessed the effect of nimodipine on focal cerebral ischemia; (2) a group of control animals was described; (3) nimodipine was administered after the induction of ischemia; (4) the effect of nimodipine was assessed in animals or in whole brains, not in slices or samples of brain tissue; (5) nimodipine was not tested in combination with other neuroprotective agents; (6) there was a restriction to the original study results (reviews and duplicate publications were excluded); and (7) the studies were published in English, French, or German. The selection of trials (unblinded) was performed (by J.H.) on the basis of title and abstract. In case of doubt, the whole publication was evaluated.
Data Extraction
From the included studies, the following data were extracted: animal species; number of animals in treated and control groups; method of allocation to treatment group; method to induce ischemia; method, dosage, and time of drug administration; method to assess efficacy (blinded assessment); and results of treatment.
Methodological Quality of Studies
On the basis of recommendations published in 1999,13 we designed an 8-point rating system to assess the methodological quality of the included animal experimental studies. One point was attributed for each of the following characteristics: (1) the dose/response relationship that was investigated, (2) randomization of the experiment, (3) optimal time window of the treatment investigated, (4) monitoring of physiological parameters, (5) blinded outcome assessment, (6) assessment of at least 2 outcomes (infarct size and 1 functional outcome), (7) outcome assessment in the acute phase (1 to 3 days), and (8) outcome assessment in the chronic phase (7 to 30 days). The points were granted when in the study report these items were mentioned. Studies scoring <4 points were graded as "poor methodological quality," and studies with
4 points were scored as "good methodological quality."
Outcome Assessment and Statistical Analyses
For each study, we defined whether a positive (nimodipine beneficial) or negative (no difference between active and placebo treatment or deleterious effect of nimodipine) result was reported.
Analysis in depth was possible only for a limited selection of trials, which reported data on the impact of nimodipine on infarct size or amount of edema. Per study, we calculated the effect size (the mean of the treatment group minus the mean of the control group divided by the pooled standard deviation of the 2 groups) and pooled the individual effect sizes accordingly. Because our data was shown to be heterogeneous, we used a random-effects model.14 Statistical uncertainty was expressed in 95% CIs.
Subgroup Analyses
In view of the common opinion that treatment has to be started as soon as possible after the induction of ischemia,15 we performed a subgroup analysis on data from studies in which nimodipine was started within or later than 1 hour after ischemia. Difference in the frequency of treatment results was analyzed with the Fisher exact test. Finally, we investigated whether study methodology influenced the results of the experiments.
| Results |
|---|
|
|
|---|
|
Twenty studies fulfilled the inclusion criteria; the characteristics of these studies are listed in Table 2, in which the studies are ordered by increasing length of the time interval between the induction of ischemia and the initiation of treatment. In total, 234 animals were treated with nimodipine (216 animals served as controls) after the induction of focal cerebral ischemia. In all studies, the animals were anesthetized during the surgical procedures, only 1 study explicitly reported that the animals were allowed to recover before the induction of ischemia.25 Parameters such as body temperature, blood glucose levels, and blood pressure were maintained within strict limits in most animal experiments.
|
In 12 of 19 studies, treatment was started within 1 hour after the induction of ischemia.16,21,2534 One study did not state the length of this interval.35 The methodological quality of the included studies was poor (median 2.85 points, range 1 to 5). Only 2 studies mentioned randomization of the animals,27,33 and in 6 studies, the outcome was assessed by a blinded assessor.21,25,26,28,32,36 One study reported "double-blind" assessment of effectiveness.37 Although 9 studies assessed 2 outcome measurements (functional and histopathologic),18,25,27,32,3539 only 3 studies assessed outcome in the chronic phase.18,32,36
Outcome
Of the 20 included studies, 10 reported a positive effect of nimodipine, and 10 did not. In 7 studies, exact data about infarct size were presented21,2831,36,37 (Figure 1). The pooled effect size in favor of nimodipine was -1.2 (95% CI -1.7 to -0.7).
|
Extracting data regarding the amount of edema was possible in 3 studies18,27,34 (Figure 2), revealing a pooled effect size of -0.6 (95% CI -1.2 to -0.1) in favor of nimodipine.
|
Subgroup Analysis
Additional analysis of the studies in which treatment was started within 1 hour after the induction of ischemia resulted in 4 positive and 8 negative studies. Treatment that was started
1 hour after the induction of ischemia resulted in 5 positive and 2 negative studies (P=0.17).
In the 4 studies with a good methodological quality,25,32,36,37 2 showed an effect in favor of nimodipine, and 2 did not.
| Discussion |
|---|
|
|
|---|
We found that the methodological quality of the included studies was poor. Issues such as randomization, masked treatment allocation, blinded outcome assessment, and intention to treat analyses, which are very important issues and are now generally required in clinical studies, were especially neglected in these animal experiments.40 Surprisingly, 1 study reported a double-blind assessment of effectiveness.
We realize that systematic reviews carry hazards such as publication bias and a bias for good quality studies.41,42 Evaluating "old" studies from the early 1980s increases these hazards. Because most experiments were performed >10 years ago, we did not approach the authors for detailed information about their studies. We considered it to be unlikely that the authors either possessed or would remember the data required. By limiting our search strategy to the electronic databases and the reference lists of articles, we were unable to estimate the number of unpublished manuscripts. For clinical trials with calcium antagonists in ischemic stroke, we found that at least 18% of all studies remained unpublished.7 In these unpublished trials, results were significantly worse for patients in the treatment group. One might assume that in experimental animal research, this percentage would not be different. If such a publication bias for negative animal experiments exists, this would result in an even more "negative" conclusion of the present review. Other possible sources for bias are selection of studies and data extraction. The fact that this was done by 1 author might have increased this risk; however, criteria for study inclusion and data extraction were strict. The number of manuscripts written in a "foreign language" (ie, Chinese, Japanese, Russian, or Polish) was limited (2%). We assumed that translating these manuscripts would not result in more useful data.
Was the animal model used suitable to predict the effectiveness of nimodipine in stroke patients? Several authors have discussed the discrepancies between the results of animal models in stroke and the results of clinical studies.4348 A possible explanation for these discrepancies was the difference in the age of the patients. Older stroke patients often have comorbidity, such as diabetes and hypertension. Furthermore, in animal experiments, physiological parameters (blood pressure, [brain] temperature, and blood glucose levels) can be kept within strict limits, and neuroprotective treatment can be started immediately or shortly after the induction of ischemia. Finally, major differences exist in the assessment of the effectiveness of animal experiments measuring histopathologic parameters compared with trials measuring functional ability in stroke patients.
Results of animal experiments with nimodipine in focal cerebral ischemia were reported until 1997, which was 15 years after the first clinical study. It is surprising that these animal experiments ran a course parallel to several clinical studies, because it is reasonable to assume that the clinical studies are preceded by animal studies. However, as early as 1982, the first results of clinical studies on the effect of nimodipine on cerebral blood flow in stroke patients were reported.49 On the basis of the hypothesis of the calcium-dependent final common pathway in cell death and his previous experiments, an author started a "single-blind" pilot study, which showed that treatment with nimodipine was promising.50 In 1988, the same author reported the results of a placebo-controlled, double-blind, randomized clinical trial.2 In the introduction to his article, results of animal experiments with nimodipine were quoted. Only 1 experiment dealt with focal cerebral ischemia28; others concerned global ischemia or the use of nimodipine before the induction of ischemia. The authors concluded that nimodipine was statistically significantly better than placebo regarding mortality and changes in neurological deficit (modified Mathew scale). However, repeat analysis of data on the functional outcome of patients in this trial, by dichotomizing the score on the disability item in the modified Mathew scale, shows a difference that was not statistically significant.7 The supposedly positive results of this first randomized controlled trial were followed by a small trial,51 which showed similar results. In the introduction to that article, the authors refer to an animal experiment by Germano et al.52 Subsequent reports on larger clinical trials3,4,53,54 refer in the introduction to a single positive animal experiment, to the proven effectiveness of nimodipine in the treatment of subarachnoid hemorrhage, and primarily to the positive results of the first randomized controlled trial with nimodipine in ischemic stroke.
We conclude that the results of the animal experiments reviewed in the present investigation did not show convincing empirical evidence to substantiate the decision for trials with nimodipine in stroke patients. In fact, there were no differences between the results of animals experiments and clinical studies. Surprisingly, we found that animal experiments and clinical studies ran simultaneously.
| Acknowledgments |
|---|
Received December 21, 2000; revision received May 22, 2001; accepted June 15, 2001.
| References |
|---|
|
|
|---|
This article has been cited by other articles:
![]() |
J. Minnerup, J. Heidrich, J. Wellmann, A. Rogalewski, A. Schneider, and W.-R. Schabitz Meta-Analysis of the Efficacy of Granulocyte-Colony Stimulating Factor in Animal Models of Focal Cerebral Ischemia Stroke, June 1, 2008; 39(6): 1855 - 1861. [Abstract] [Full Text] [PDF] |
||||
![]() |
N. A. Crossley, E. Sena, J. Goehler, J. Horn, B. van der Worp, P. M.W. Bath, M. Macleod, and U. Dirnagl Empirical Evidence of Bias in the Design of Experimental Stroke Studies: A Metaepidemiologic Approach Stroke, March 1, 2008; 39(3): 929 - 934. [Abstract] [Full Text] [PDF] |
||||
![]() |
E. Sena, P. Wheble, P. Sandercock, and M. Macleod Systematic Review and Meta-Analysis of the Efficacy of Tirilazad in Experimental Stroke Stroke, February 1, 2007; 38(2): 388 - 394. [Abstract] [Full Text] [PDF] |
||||
![]() |
R. Lemon and S. B Dunnett Surveying the literature from animal experiments BMJ, April 30, 2005; 330(7498): 977 - 978. [Full Text] [PDF] |
||||
![]() |
L. E. Mignini, P. M. Latthe, J. Villar, M. D. Kilby, G. Carroli, and K. S. Khan Mapping the Theories of Preeclampsia: The Role of Homocysteine Obstet. Gynecol., February 1, 2005; 105(2): 411 - 425. [Abstract] [Full Text] [PDF] |
||||
![]() |
M. R. Macleod, T. O'Collins, D. W. Howells, and G. A. Donnan Pooling of Animal Experimental Data Reveals Influence of Study Design and Publication Bias Stroke, May 1, 2004; 35(5): 1203 - 1208. [Abstract] [Full Text] [PDF] |
||||
![]() |
P. Pound, S. Ebrahim, P. Sandercock, M. B Bracken, and I. Roberts Where is the evidence that animal research benefits humans? BMJ, February 28, 2004; 328(7438): 514 - 517. [Full Text] [PDF] |
||||
![]() |
P. Pound and S. Ebrahim Supportive evidence is lacking for report on animal studies BMJ, November 2, 2002; 325(7371): 1038 - 1038. [Full Text] |
||||
![]() |
D. J. Gladstone, S. E. Black, and A. M. Hakim Toward Wisdom From Failure: Lessons From Neuroprotective Stroke Trials and New Therapeutic Directions Stroke, August 1, 2002; 33(8): 2123 - 2136. [Abstract] [Full Text] [PDF] |
||||
![]() |
I. Chalmers, L. V. Hedges, and H. Cooper A Brief History of Research Synthesis Eval Health Prof, March 1, 2002; 25(1): 12 - 37. [Abstract] [PDF] |
||||
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
Stroke Home | Subscriptions | Archives | Feedback | Authors | Help | AHA Journals Home | Search Copyright © 2001 American Heart Association, Inc. All rights reserved. Unauthorized use prohibited. |