Calcium Antagonists for Ischemic Stroke
A Systematic Review
Background and Purpose—Stroke is a common disease, and many trials with calcium antagonists as possible neuroprotective agents have been conducted. The aim of this review is to determine whether calcium antagonists reduce the risk of death or dependency after acute ischemic stroke.
Methods—Acute stroke trials were identified with help of the Cochrane Collaboration Stroke Group and personal contacts. All randomized trials (published and unpublished) investigating a calcium antagonist (acting on voltage-sensitive calcium channels) were included. Poor outcome, defined as death or dependency in activities of daily living, was used as main outcome. Analyses were, if possible, “intention-to-treat”; pooled relative risks with 95% CIs were calculated.
Results—Forty-seven trials were identified, of which 29 were included (7665 patients). No effect of calcium antagonists on poor outcome at the end of follow-up (relative risk, 1.04; 95% CI, 0.98 to 1.09) or on death at end of follow-up (relative risk, 1.07; 95% CI, 0.98 to 1.17) was found. Sensitivity analyses on route of administration and time interval between stroke and start of treatment showed no effect on outcome. In subgroups of unpublished and methodologically sound trials, a statistically significant negative effect for calcium antagonists was found. This contrasts with results of published trials and trials of moderate or poor methodological quality.
Conclusions—The presented evidence rules out a clinically important effect of calcium antagonists after ischemic stroke. The large amount of data leads to narrow CIs with no significant heterogeneity, and the overall results are therefore likely to be statistically robust.
Acute ischemic stroke is a major cause of death and disability. Despite many experimental studies, there is as yet no effective, generally accepted, specific treatment in the acute phase of stroke. Massive calcium influx into hypoxic cells is a final common pathway, leading to cell death.1 Animal experiments have indicated that calcium antagonists administered after cerebral ischemia are effective in reducing infarct volume and lead to improvements in neurological outcome.2 3 Calcium antagonists may act as neuroprotective drugs by diminishing the influx of calcium ions through voltage-sensitive calcium channels.4
A clinical trial with the calcium antagonist nimodipine suggested a beneficial effect,5 but none of the many subsequent trials confirmed these results. However, sample sizes may have been too small to demonstrate a modest but perhaps clinically significant effect. Some meta-analyses with a limited scope (only nimodipine data included) have been performed.6 7 8 The aim of this systematic review was to analyze all available clinical evidence to determine whether calcium antagonists reduce the risk of death or poor outcome after ischemic stroke.
Identification of Trials and Criteria for Inclusion
This review has drawn on the search strategy developed for the Cochrane Collaboration Stroke Group, and relevant trials were identified in the Specialized Register of Controlled Trials.9 This register was last checked in May 1999. Some unpublished trials were found through contacts with principal investigators and company representatives. Both authors independently selected the trials to be included in the review. Studies were included if they fulfilled the following criteria: true randomization, randomization of patients within 14 days after ischemic stroke, and investigation of the effect of calcium antagonists (defined as agents whose principal mode of action is to inhibit influx of calcium into cells by way of voltage-sensitive calcium channels).
The unblinded trial reports were read and assessed independently by both authors, and information needed on methodological quality was extracted. Disagreement was resolved by discussion. Protocols of studies were sought for methodological information, and data collection forms were sent to all principal investigators.
Types of Outcome Measures
As main outcome we used “poor outcome,” defined as all-cause case fatality or dependency from others in activities in daily living, at the end of follow-up. For this purpose the available functional health scale of each study was used in a dichotomized fashion (Table 1⇓).10 11 12 13 14 Because patients with a Rankin score of 4 or 5 almost certainly cannot live independently,15 this cutoff point for the Modified Rankin Scale was chosen. The cutoff points for the other scales were derived from this cutoff point. These cutoff points were established before the analysis of data started. If more than one scale was available, the one with the smallest number of missing values was selected.
Mortality at end of treatment and at the end of follow-up and adverse events were assessed as secondary outcome. Adverse events were recorded if they were mentioned as such in the original article, and therefore the definitions of the investigators were used.
The primary analysis concerned the main end point of poor outcome. Several trials only reported data on mortality.
Sensitivity analyses were performed for the following: (1) Route of drug administration (intravenous or oral) was analyzed. In contrast to oral administration, intravenous administration of nimodipine seemed to cause serious hypotension.16 (2) Time interval between stroke and start of treatment, in view of evidence that time elapsed between stroke onset and initiation of treatment is important for therapeutic success, was analyzed. Treatment started within 12 hours after stroke onset was considered early treatment. In view of previous meta-analyses we separately addressed nimodipine trials (120 mg/d, oral administration, started within 12 hours after stroke onset) in this analysis. (3) Trial methodology was analyzed. We determined methodological quality of the studies17 : 1 credit was given if the word randomized was mentioned, and 1 credit was given if the method of randomization was properly described and sufficient. If the title or article contained phrases such as double-blind or concealed, this yielded another credit; proper description of the method of treatment concealment led to another credit. The last credit was given when all included patients were accounted for in the article. Trials scoring 5 points were considered good quality trials, trials scoring 3 or 4 points were considered moderate quality trials, and trials scoring <3 points were considered poor quality trials. (4) Publication status was analyzed to investigate the effect of publication bias. In all these sensitivity analyses, poor outcome was assessed.
In several cases both the original data set and the published article were available. The numbers in the data sets were usually slightly different, and in some cases differences were extensive. Many authors report data of patients valid for efficacy analysis instead of reporting intention-to-treat analyses. In Table 2⇓⇓18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 we indicate which data set is used for each trial. Thus, the numbers used in this review may differ from those in original publications. In case of incomplete follow-up data on functional status, the last known value was carried forward. If only functional status at trial inclusion was available, these data were not used. The present analysis can be considered a “best case” analysis: missing values are handled as if they represent good outcomes. Most missing values regarded dependency. We performed a “worst case” sensitivity analysis by considering all missing values to represent poor outcomes.
Results of analyses are presented as relative risks (RRs) with 95% CIs. Relative risks were computed with the Mantel-Haenszel method (fixed effects model).
All analyses were performed by the authors with Revman software, developed and provided by the Cochrane Collaboration.
We tested for statistical heterogeneity between trial results using a standard χ2 test. P<0.05 indicated presence of statistical heterogeneity.
Full details of the methods can be found in the full Cochrane Library electronic publication, on which this article is based.42
Description of Studies
We identified 47 studies using calcium antagonists in patients with acute ischemic stroke. Eighteen studies were excluded, of which 9 did not fulfill inclusion criteria and 9 lacked crucial information (Table 3⇑).43 44 45 46 47 48 49 50 51 52 53 54 55 56 Characteristics of the 29 included studies (with data of 7665 patients) are presented in Table 2⇑. In 25 studies treatment was started within 48 hours after stroke onset. Length of follow up was <3 months in 10 included trials, approximately 3 months (12 weeks) in 5 trials, and >3 months (6 months to 1 year) in 14 trials.
Results are presented in the Figure⇓. Data of 22 trials could be used (6877 patients) for the analysis of poor outcome. No overall effect was found, and none of the separate drugs showed any statistically significant effect. No major differences arose in the analysis in which patients with missing functional outcome were considered to have a poor outcome. Data of 7522 patients were included in the mortality at end of follow-up analysis; no effect of any calcium antagonist was found. In this analysis the 3 flunarizine trials showed a statistically significant unfavorable effect (RR, 1.3; 95% CI, 1.0 to 1.8). No difference was present in the analysis on mortality at the end of treatment. Adverse events were reported more often in patients treated with calcium antagonists than in the control groups. Again, with flunarizine the results were even worse (RR, 3.2; 95% CI, 1.9 to 5.2). However, in this analysis only data from 1 trial could be included, and the main adverse event was thrombophlebitis.23
No difference was found between oral or intravenous administration of calcium antagonists in indirect comparisons (oral versus placebo and intravenous versus placebo), although intravenous administration tended to have worse results. Only 1 small trial (144 patients) directly compared these 2 routes of administration,22 demonstrating a nonsignificant trend in favor of intravenous administration (RR, 7.1; 95% CI, 0.4 to 135.0).
In contrast to previously reported results,8 we did not find a beneficial effect of early treatment (within 12 hours), nor was any effect present for treatment started after 12 hours. The separately analyzed data from trials using nimodipine (120 mg, oral administration) did not show a beneficial effect of early treatment. For these analyses data of 11 trials (660 patients in treatment arm, 619 in placebo group) were used (G. Lowe and C. Forbes, unpublished data, 1989).5 25 27 32 33 34 36 38 40 41 For poor outcome the RR was 1.0 (95% CI, 0.9 to 1.2), and for mortality it was 0.9 (95% CI, 0.8 to 1.2).
With the criteria of Jadad et al,17 12 studies were graded as good quality trials, 12 as moderate quality trials, and 3 as poor quality trials. The methodological quality of 2 (unpublished) trials could not be assessed because information was lacking. The comparison between active and placebo treatment in the good quality trials yielded a statistically significant negative effect for active treatment. In the moderate and poor quality trials, no effect of active treatment was found.
Data of 18 published or presented trials were included, and we identified 4 unpublished trials from which we obtained data. Comparison between treatment and placebo in published trials yielded no difference, whereas in unpublished trials a statistically significant unfavorable effect of treatment was found.
No statistical heterogeneity was found in the main analyses (poor outcome, mortality) (P=0.14 and P=0.68, respectively). We found statistically significant heterogeneity (P=0.003) in the analysis of adverse events. This was mainly caused by the Flunarizine in Stroke Treatment (FIST) trial, in which more statistically significant adverse events were present.23 After this trial was deleted from the adverse events analysis, no heterogeneity was present. Clinical heterogeneity was addressed in the sensitivity analyses.
This systematic review of all available data, published and unpublished, failed to demonstrate a reduction of death and dependency in acute stroke after treatment with calcium antagonists. The CIs are fairly narrow, and the overall result is therefore subject to limited statistical uncertainty.
In the sensitivity analyses we found no differences between oral or intravenous administration and early or late start of treatment. This last result is intriguing, since Mohr et al8 reported a statistically significant beneficial effect of early administration of nimodipine. In their meta-analysis no effect of nimodipine was found in the overall analysis, but in the subgroup analysis of patients treated within 12 hours of stroke onset, the odds ratio was 0.62 (95% CI, 0.44 to 0.87) for neurological impairment in favor of nimodipine. Similar results were reported for functional outcome assessments. By separate analysis of nimodipine trials, we tried to confirm the results found by Mohr et al but were unable to do so. We included more trials because at the time of the former meta-analysis data of 2 trials were not available,34 40 and data of 2 trials were excluded by Mohr et al because data of these trials were not available from the Bayer database (G. Lowe and C. Forbes, unpublished data, 1989).27 Differences between structured reviews may occur as a result of differences in the design of the meta-analysis.57 58 The methods section of the meta-analysis by Mohr et al did not describe all decisions taken. Moreover, the exact numbers of patients with poor outcome of each included trial were not reported and could not be retrieved from the authors or from the pharmaceutical company involved. As a result, we could not compare the analyses in detail.
Trial methodology can cause considerable bias in a systematic review.59 60 Although scoring the quality of trials can be hazardous,61 we decided to address this issue in a sensitivity analysis. Sensitivity analyses are subgroup analyses, and the conclusions should therefore be regarded with caution. Nevertheless, we found strong indications for the presence of publication bias in our sensitivity analyses. While the published trials showed no overall effect on death and dependency, unpublished trials were associated with a statistically significant worse outcome in the active treatment group. It is likely that more “negative” trials have been performed that have not been published. In fact, we have some indications for the existence of these trials, of which we could not acquire further information.
From this review we conclude the following: (1) the use of calcium antagonists is not justified in patients with ischemic stroke; (2) positive results of subgroup analyses within a meta-analysis should be judged with caution; and (3) there is a clear difference in results from published versus unpublished trials, and this effect is also found for good quality trials compared with moderate or poor quality trials.
This study was supported by Janssen Pharmaceuticals, Stichting De Drie Lichten. Dr Limburg is a clinical investigator of the Netherlands Heart Foundation (D 93.014). Janssen Pharmaceuticals supported this study with a grant without any restrictions. We wish to express our gratitude to the Cochrane Stroke review group (especially Dr Peter Sandercock and Hazel Fraser) for its enduring support.
- Received July 1, 2000.
- Revision received October 26, 2000.
- Accepted October 31, 2000.
- Copyright © 2001 by American Heart Association
Gelmers HJ, Hennerici M. Effect of nimodipine on acute ischemic stroke: pooled results from five randomized trials. Stroke. 1990;21(suppl IV):IV-81–IV-84.
Mohr JP, Orgogozo JM, Harrison MJG, Hennerici M, Wahlgren NG, Gelmers HJ, Martinez-Vila E, Dycka J, Tettenborn D. Meta-analysis of oral nimodipine trials in acute ischemic stoke. Cerebrovasc Dis. 1994;4:197–203.
Cochrane Stroke Group. Specialised register (Cochrane review). In: The Cochrane Library, issue 1, 2000. Oxford, UK: Update Software.
Norris JW. Comment on “Study design of stroke treatments.” Stroke. 1982;4:527–528.
de Haan R, Limburg M, Bossuyt P, van der Meulen JHP, Aaronson N. The clinical meaning of Rankin “handicap” grades after stroke. Stroke. 1995;26:2027–2030.
Azcona A, Lataste X. Isradipine in patients with acute ischaemic cerebral infarction: an overview of the ASCLEPIOS programme. Drugs. 1990;40:52–57.
Bridgers SL, Koch G, Munera C, Karwon M, Kurtz NM. Intravenous nimodipine in acute stroke: interim analysis of randomized trials. Stroke. 1991;22:153. Abstract.
Chandra B. A new form of management of stroke. J Stroke Cerebrovasc Dis. 1995;5:241–243.
Kramer G, Tettenborn B, Schmutzhard E, Aichner F, Schwartz A, Busse O, Hornig C, Ladurner G. Nimodipine in acute ischemic stroke: results of the German-Austrian Stroke Trial. Cerebrovasc Dis. 1994;4:182–188.
Kaste M, Fogelholm R, Erila T, Palomaki H, Murros K, Rissanen A, Sarna S. A randomized, double-blind, placebo-controlled trial of nimodipine in acute ischemic hemispheric stroke. Stroke. 1994;25:1348–1353.
Kornhuber HH, Hartung J, Herrlinger JD, Hertel G, Hulser PJ, Prange H, Bernhard H, Birkner B, Buske A, Elpelt B, Kohler W, Kohne M, Krapf H, Marbach-Kirchner C, Plenio A, Schneider E, Schwenk B, Wissel J, Huss HJ, Peters T, Seiler KU. Flunarizine in ischemic stroke: a randomised, multicentre, placebo-controlled, double blind study. Neurology Psychiatry Brain Res.. 1993;1:173–180.
Martinez-Vila E, Guillen F, Villaneuva JA, Matias-Guiu J, Bigorra J, Gil P, Carbonell A, Martinez-Lage J. Placebo-controlled trial of nimodipine in the treatment of acute ischemic cerebral infarction. Stroke. 1990;21:1023–1028.
The American Nimodipine Study Group. Clinical trial of nimodipine in acute ischemic stroke. Stroke. 1992;23:3–8.
Hennerici M, Kramer G, North PM, Schmitz H, Tettenborn D, for the Nimodipine European Stroke Trial Group. Nimodipine in the treatment of acute MCA ischemic stroke. Cerebrovasc Dis. 1994;4:189–193.
Infeld B, Davis S, Donnan G. Nimodipine and perfusion after acute stroke. Stroke. 1999;30:1417–1423.
Oczkowski WJ, Hachinski VC, Bogousslavsky JB, Barnett HJM, Carruthers SG. A double-blind, randomized trial of PY 108-068 in acute ischemic cerebral infarction. Stroke. 1989;20:604–608.
Sherman DG, Easton JD, Hart RG. Nimodipine in acute cerebral infarction: a double blind study of safety and efficacy. In: Battistini N, ed. Acute Brain Ischaemia: Medical and Surgical Therapy. New York, NY: Raven Press; 1986:257–262.
Uzuner N, Ozdemir G, Gucuyener D. The interaction between nimodipine and systemic blood pressure and pulse rate. In: Proceedings of the WHO Pan-European Consensus Meeting on Stroke Management; November 8–10, 1995; Helsingborg, Sweden.
Horn J, Limburg M, Vermeulen M. VENUS—Very Early Nimodipine Use in Stroke: final results from a randomised-controlled trial. Cerebrovasc Dis. 1999;9(suppl 1):127.
Wimalaratna HSK, Capildeo R. Nimodipine in acute ischaemic cerebral hemispheric infarction. Cerebrovasc Dis. 1994;4:179–181.
Horn J, Limburg M. Calcium antagonists for ischaemic stroke (Cochrane review). In: The Cochrane Library, issue 1, 2000. Oxford, UK: Update Software.
Ameriso SF, Wenby RB, Meiselman HJ, Fischer M. Nimodipine and the hemorheological variables after acute ischemic stroke. J Stroke Cerebrovasc Dis. 1992;2:22–25.
Dalal PM, Dalal KP. Use of calcium channel blockers in acute ischemic cerebrovascular disease. J Assoc Phys India. 1995;43:394–397.
Davalos A, Cendra E, Gonzalez B, Genis D, Teruel J, Ruibal A. Double-blind randomized clinical trial of nicardipine vs placebo in acute ischemic stroke: clinical, radiological and biochemical evaluation of the ischemic area: preliminary results. Neurol India. 1989;37(suppl):283. Abstract.
Davalos Errando A, De Cendra E, Genis D, Teruel J, Ruibal A, Musoles S. Double-blind controlled trial of nicardipine versus placebo in the treatment of the acute phase of cerebral infarction. Neurologia. 1992;7:157. Abstract.
Fagan SC, Gengo FM, Bates V, Levine SR, Kinkel WR. Effect of nimodipine on blood pressure in acute ischemic stroke in humans. Stroke. 1988;19:401–402.
Lamsudin HR, Misbach HJ, Andradi, Tobing L, Cahyadi P, Soedomo H, Fauziah, Sumaranto. A randomized, double-blind, placebo-controlled multicenter trial. Indonesian J Clin Epidemiol Biostatistics. 1995. Abstract.
Matias-Guiu J, Molto JM, Galiano L. Pilot double-blind placebo controlled trial of nicardipine versus placebo for cognitive impairment in minor stroke patients. J Neurol. 1992;2(suppl):S39. Abstract.
Petrogiannopoulos G, Zaharof A, Tzoumani A, Poulikakos J. Efficacy of long term treatment with nimodipine on brain function after acute ischemic stroke. Eur J Neurol. 1996;3:35. Abstract.
Rosselli A, Landini G, Castagnoli A, Vannucchi L, Mugnaini C, Calacoci S, Regoli G, Francesconi A, Turchi A, Agostino I. Therapy with nimodipine of acute focal cerebral ischaemia (minor stroke): clinical study evaluating the regional cerebral blood flow with single photon emission computed tomography (SPECT). Recenti Prog Med. 1992;83:85–88.
Rosenbaum DM, Zabramski JM, Fry J, Yatsu F, Marler J, Spetzler RF, Grotta JC. Early treatment of ischemic stroke with a calcium antagonist. Stroke. 1991;22:437–441.
Szczechowski L, Wajgt A. Treatment efficiency of acute hemispheric strokes using nimodipine infusions. Neur Neurochir Pol. 1994;28:299–306.
Yao J. Preliminary study of nimodipine in acute cerebral infarction. Chin J Nerv Ment Dis. 1991;17:47. Abstract.
Lindbaeck M, Hjortdahl P. How do two meta-analyses of similar data reach opposite conclusions? BMJ. 1999;318:873–874.
Jadad AR, Cook DJ, Browman GP. A guide to interpreting discordant systematic reviews. Can Med Assoc J. 1997;156:1411–1416.