1,026 experimental treatments in acute stroke

Victoria E. O’Collins, B.Sci,1 Malcolm R. Macleod, MRCP, PhD,3 Geoffrey A. Donnan, MD, FRACP,2 Laura L. Horky, MD, PhD,2 Bart H. van der Worp, MD, PhD,4 and David W. Howells, PhD1 Objective: Preclinical evaluation of neuroprotectants fostered high expectations of clinical efficacy. When not matched,
the question arises whether experiments are poor indicators of clinical outcome or whether the best drugs were not taken
forward to clinical trial. Therefore, we endeavored to contrast experimental efficacy and scope of testing of drugs used
clinically and those tested only experimentally. Methods: We identified neuroprotectants and reports of experimental
efficacy via a systematic search. Controlled in vivo and in vitro experiments using functional or histological end points
were selected for analysis. Relationships between outcome, drug mechanism, scope of testing, and clinical trial status were
assessed statistically. Results: There was no evidence that drugs used clinically (114 drugs) were more effective experi-
mentally than those tested only in animal models (912 drugs), for example, improvement in focal models averaged
31.3 ؎ 16.7% versus 24.4 ؎ 32.9%, p > 0.05, respectively. Scope of testing using Stroke Therapy Academic Industry
Roundtable (STAIR) criteria was highly variable, and no relationship was found between mechanism and efficacy.
Interpretation: The results question whether the most efficacious drugs are being selected for stroke clinical trials. This
may partially explain the slow progress in developing treatments. Greater rigor in the conduct, reporting, and analysis of
animal data will improve the transition of scientific advances from bench to bedside.
A common perception of neuroprotection research is neity, inappropriate dose, and time window). These re- that everything works in animals but nothing works in views have provided a solid framework for sharpening people. This perception has been reinforced again and experimental and clinical trial design, but, by and again by reports of unsuccessful or mixed outcomes in large, they have been qualitative rather than quantita- trials of candidate neuroprotectants in acute stroke pa- tive in nature. Additionally, only recently has the im- tients. If animal experiments are indeed unable to in- pact of study design and quality been examined in re- form clinical decision making, then serious doubts are lation to efficacy in experimental models of stroke.8–10 raised about the utility of animal models of stroke and We set out to examine whether it is indeed true that about the ethics of continuing current animal experi- everything produces neuroprotection in animals, and to compare the experimental efficacy of those interven- In response to this challenge, several excellent re- tions taken forward to clinical use with that of drugs views and commentaries have tackled the issue of the tested only in the laboratory. A broad definition of apparent failure to translate neuroprotection successes neuroprotection was adopted, encompassing any agent from the laboratory to the clinical setting (for exam- or treatment tested in an animal model of stroke, ples, see previously published studies2–7). Such com- whether administered with the intent of preventing mentaries have questioned the appropriateness of the neuronal death, restoring blood flow, or merely inves- experimental animals (eg, age, sex, comorbidities, com- tigating the mechanisms of damage in stroke. Thus, of parative anatomy, sample sizes), the stroke model (eg, the many ways one can define neuroprotection (see Ta- anesthesia, hypothermia, glucose, reperfusion), the out- ble 1), primacy was given to the evaluation of the final come measures (histology vs functional deficits and effect of a drug on the preservation of brain function, death), study quality (blinding, randomization), and irrespective of intended cellular target, purpose of ad- clinical patient selection and dosing (patient heteroge- ministration, or clinical classification of the drug.
From the 1Neuroscience Lab, Department of Medicine, University of Melbourne, Austin Health, Heidelberg; and 2National Stroke Re- (www.interscience.wiley.com). DOI: 10.1002/ana.20741 search Institute, Heidelberg Heights, Australia; 3School of Molecu-lar and Clinical Medicine, University of Edinburgh, Edinburgh, Address correspondence to Dr Howells, National Stroke Research United Kingdom; and 4Department of Neurology, University Med- Institute and University of Melbourne, Department of Medicine, ical Centre Utrecht, Utrecht, The Netherlands.
Austin Health, Studley Road, Heidelberg, VIC 3084, Australia.
Received Jul 25, 2005, and in revised form Oct 10. Accepted for 2006 American Neurological Association Published by Wiley-Liss, Inc., through Wiley Subscription Services Table 1. Definition of a Neuroprotectant Agent administered after acute stroke to preserve neurons, in contrast with agents that restore blood flow (thrombolytics).
Limitation: Absence of strong logical argument as to why thrombolytics should be considered a dis-crete class, whereas the neuroprotectives class should include a myriad of agents themselves withother clinical uses (eg, antidepressants, fluid replacement). Furthermore, whereas the immediate tar-get of thrombolytics might be blood flow, the ultimate effect of interest is normalization of brainfunction and behavior. The direct effect of thrombolytics on neural tissue is also attracting greaterattention.
Agent containing chemical or physical properties known to be associated with neuroprotective effects.
Limitation: Such properties are generally yet to be elucidated.
Agent targeting cellular pathways that control neuronal fate to preserve brain function.
Limitation: Greater understanding required regarding which cells must survive to preserve brainfunction (eg, neurons, oligodendroglia, microglia) and of how the fate of individual cells relates tothe overall survival of brain tissue. More information is also needed regarding how the preservationof processes governing the recruitment and disengagement of brain networks relates to the behaviorthought to be subserved by these networks.
Agent administered with the purpose of preserving brain function.
Limitation: Drugs administered for other purposes, eg, fluid replacement, antidepressants, thrombo-lytic, may also affect the preservation of brain function.
Agent whose final effect is to preserve brain structure or function.
Limitation: Nonmechanistic, empirical approach does not facilitate further drug development.
A further aim was to examine the scope of preclini- experimental phase (experimental agents); and (3) to cal testing to which drugs have been subjected, with a compare the experimental neuroprotective efficacy of the view to assessing the adequacy of our procedures for main classes of neuroprotective agents. Because neuro- selecting drugs to carry forward to clinical trial. Here, protective agents were being evaluated primarily for their scope of testing refers to the range of preclinical exper- application to stroke, greater emphasis was placed on the imental procedures to which an intervention has been identification and analysis of the efficacy demonstrated subjected (eg, range of species and stroke models in in models of focal ischemia (rather than global ischemia which the drug was tested). Scope of testing, together or cell culture methods) because focal models are with quality control within individual experiments (eg, thought to more closely mimic human stroke.15 via blinding and randomization) form the twin pillarsof the Stroke Therapy Academic and Industry Round- Materials and Methods
table recommendations.11 These recommendations re- Identification of Putative Neuroprotective Drugs late to the benchmarks for preclinical evidence that Drugs or treatments used in experimental stroke were iden- should be met before an experimental treatment is con- tified by (1) PubMed search for “neuroprotection”; (2) sidered for clinical trial and represent an important PubMed search for “cerebral ischemia” from 1960 to 1980 consensus on preclinical standards reached between (to capture agents tested prior to the use of the keyword several distinguished researchers in the field of stroke.
“neuroprotection”), (3) search of the test and reference lists Neuronal injury in stroke is thought to result from a of articles identified by (1) and (2). Drugs used in clinical surge in activation of interlinking pathophysiological stroke were identified by searching of clinical trial data- pathways, with different pathways possibly predomi- bases16,17 and clinical trial reviews18,19 and from the text and nating in the core and at the cusp of ischemic dam- reference lists of articles identified in the methods described age.12–14 Neuroprotective treatments in stroke typically above. Because the focus was on whether the interventionhas been trialed in humans and not whether the clinical trial target one or more of these mechanisms of damage; was well conducted, reports of any trial in humans were used hence, we have also investigated whether efficacy in ex- to demarcate the intervention clinical trial status.
perimental stroke was substantially different in any ofthe major categories of therapeutic agents.
The main aims of the review were therefore (1) to Reports of efficacy in experimental stroke were identified via identify agents tested in animal neuroprotection models PubMed and by hand-searching of relevant journals. For and those treatments given to acute stroke patients; (2) each drug, a PubMed search was conducted using the search to compare the overall quality of evidence and experi- criteria “(drug name) AND (cerebral ischemia OR stroke OR mental efficacy of those treatments that have been given neuroprotection).” Where no results were found, the search to acute stroke patients (clinical treatments) and those was repeated using only the drug name. Information on the agent or strategies that have not progressed beyond the physiological effects of drugs was obtained from the Merck Table 2. Models of Cerebral Ischemia Clip, ligation by external filament, internal filament, inflatable cuff Microsphere, macrosphere, autologous clot, fibrin clot, polyvinyl acetate Endothelin-1, Rose Bengal photochemical dye, arachidonic acid, adenosine 5Ј Potassium cyanide (KCN), asphyxia, carbon dioxide, nitrogen Clip, tourniquet, balloon compression, decapitation, ligation, intracranial pres- Staurosporine, C2-ceramide, paclitaxel, low potassium, etoposide, tunicamycin, serum deprivation, Bleomycin sulfate, 3-morpholinosydnonimine, SIN-1,sodium nitroprusside (SNP), S-nitroso-N-acetylpenicillamine (SNAP) Anoxia, hypoxia, oxygen, and glucose deprivation, veratridine, sodium cya- nide, iodoacetic acid, sodium azide (cytochrome oxidase inhibitor),endothelin-1 H2O2, iron, heme, photochemical stress, glutathione depletion, superoxide, naphthazarin, antimycin A, rotenone, 3-morpholinosydnonimine, sodiumnitroprusside (SNP), S-nitroso-N-acetylpenicillamine (SNAP) Index, online chemical databases, and experimental and re- emia experiments, with the following information also ex- view papers. Foreign language reports were included only tracted: animal (species sex, age, comorbidities); stroke model when published with an English abstract.
(method of induction); intervention (dose, time treatment,mode of delivery), and outcome (end point, time measured).
Information regarding focal ischemia experiments was taken Drug efficacy studies in established in vivo and in vitro mod- from full publications (where available) but for other exper- els of neuroprotection were considered for inclusion (see Ta- iment types was generally extracted from published abstracts.
ble 2). Both focal and global models of ischemia were in- Each experiment was given a score of Ϫ1, 0, or 1 depend- cluded, but experiments performed in nonstroke models of ing on whether the treatment resulted in an outcome that neuroprotection (eg, Parkinson’s disease and epilepsy) were was significantly worse, the same, or better than that in the excluded. In vitro models of neuroprotection designed to control group, respectively. Where the drug had also been model the pathophysiology of stroke were also included and tested in focal models of ischemia and the infarct volume typically involved the application of a toxic stimulus (eg, glu- (IV) reported, the average level of protection (%) was also tamate, oxygen and glucose deprivation, hydrogen peroxide) to a neuronal culture or coculture of neurons and astrocytes,to hippocampal or cortical slices or to retinal explants. Con- trolled experiments measuring outcome using a functional(behavioral), histological, survival, electrophysiological, or imaging end point were retained. For focal ischemia studies, we included only those studies reporting outcome as infarct volume (measured histologically or by brain imaging) or a be-havioral score. For global ischemia, we included studies report-ing survival; the extent of cell death; neurobehavioral scores; or Scoring System for Scope of Testing normalization of electroencephalogram. In keeping with the A scoring system was designed to measure the diversity of evi- Stroke Therapy Academic Industry Roundtable (STAIR) rec- dence supporting each drug, based on the STAIR recommen- ommendations,11 reports restricted to consideration of mech- dations. Each drug was assigned a score (0 –10) to reflect how anism of drug action, gene expression patterns, cerebral water widely it has been tested in preclinical models of stroke (Table content, blood–brain barrier function, blood flow changes, 3). Although the criteria largely reflect the STAIR recommen- and restoration of metabolism were excluded because these dations, several departures were made; testing in nonhuman pri- represent surrogate end points having complex relationships mates was not included as a criterion because the superior va- with neuronal survival and neurological impairment.
lidity of such models is not well established. Additionally,because the focus of the review was the performance of multiple drugs over multiple experiments, the scoring system did not in- For each study, the following information was collected: type clude measures relating to quality control within individual ex- of experimental model, drug or intervention, and outcome.
periments (eg, randomization and blinding) but instead focused A more comprehensive data set was compiled for focal isch- on the scope of testing across experimental models.
O’Collins et al: Neuroprotective Treatments Table 3. Quality of Evidence in Experimental Stroke Scale Focal model tested in two or more laboratories Focal model tested in two or more species Focal model tested in old or diseased animalsa Focal model tested in male and female animals Tested in temporary and permanent models of focal ischemia Drug administered at least 1 hour after occlusion in focal model Drug administered using at least two doses in focal model Tested using a feasible mode of delivery (eg, not intracisternal or intraventricular, Both behavioral and histological outcome measured Outcome measured at 4 or more weeks after occlusion in focal models Not given in order of priority.
aDiabetic, hypertensive, aged, hyperglycemic.
Excel and Systat 11 were used to analyze the data. All means The initial search identified 1,150 interventions; 124 are presented as mean Ϯ standard deviation (SD). A 10(drug were excluded from further analysis: 56 because reports mechanism) ϫ 2(clinical trial status) analysis of variance of their effects related only to mechanism or used non- (ANOVA) was undertaken to evaluate the effect of clinical eligible end points, and 68 because the literature re- trial status and mode of drug action on outcome in focal lated to neuroprotection in models of disease other models of ischemia. Scope of testing was analyzed separately than stroke. This analysis therefore is based on data for because it did not meet the assumptions for ANOVA.
1,026 candidate stroke drugs, of which 912 have been tested in animal models only, 97 have been tested inboth animal models and clinical trials, and 17 have Number of Neuroprotection Experiments and been tested only in humans. For 51% of the 114 drugs tested in humans, the first report of its clinical use was A total of 8,516 experimental results were extracted published before the first report of its use in a focal from approximately 3,500 papers (published between ischemia model of stroke, and 42% of these drugs were 1957 to 2003). Of these, 962 experimental results were used clinically before the first report of testing in a excluded because they related to nonstroke neuropro- tection models; thus, the analysis is based on 7,554 ex-perimental results from models of focal ischemia (3,867 results), global ischemia (1,546 results), and Of the 1,026 agents, 603 (59%) had been tested in culture (1,341 results). The 1990s showed a marked focal ischemia models, 256 (25%) had been tested in growth in reports of neuroprotection experimentation both focal and global ischemia models, and 24 (2%) (Fig 1). At the same time, there was a less marked in- had been tested in focal, global, and cell culture mod- crease in reports of the first clinical application of can-didate neuroprotective drugs (Fig 2).
Fig 1. Neuroprotection experiments identified from published Fig 2. First reported clinical trials of inventions in acute els. Neuroprotective efficacy was superior to the con- sea snail peptides.24 More than 1,000 such drugs and trol condition in 62% of focal ischemia experimental nonpharmacological strategies have been tested for results, and no different or inferior to the control con- their neuroprotective efficacy in animal models, and at dition in 34% and 4% of cases, respectively. In global least 114 have been administered to acute stroke pa- ischemia models, neuroprotective efficacy was superior tients. Efficacy in focal ischemia ranged from 94% pro- to the control condition in 70% of experimental re- tection to a twofold increase in damage in the inter- sults, and no different or inferior to the control condi- vention versus the control condition.
tion in 26% and 4% of cases, respectively. In culture Those drugs given to acute stroke patients were not models, neuroprotective efficacy was superior to the distinguished by superior efficacy in animal models of control condition in 74% of experimental results, and focal ischemia compared with all drugs that have been no different or inferior to the control condition in tested only experimentally. Given that the purely ex- perimental class of agents included drugs used to in- Table 4 shows the analysis of animal experiment duce comorbid disease states in animals (eg, diabetes, data from these comparisons for those drugs tested in inflammation), this raises the issue of whether we are clinical trial. For each drug, the number of positive, in fact selecting the best drugs to carry forward to clin- neutral, and detrimental experimental outcomes are re- ported for each of the culture, focal, and global mod- The unremarkable experimental track records of some drugs given to patients may be explained by sev-eral factors. Historically, less emphasis may have been Experimental Efficacy in Focal Models of Ischemia placed on animal experimentation, and where it was Because ANOVA is very sensitive to extreme results, undertaken, the outcomes used may have related to one study using a hamster model of focal ischemia was metabolism or blood flow rather than neuroprotection.
excluded because it showed an unusual 1,250% in- Second, the conditions in clinical trials often fail to crease in damage using tissue plasminogen activator (t- replicate the conditions under which drugs have been PA) and 2,500% increase in damage using streptoki- found to work in animals, especially with regard to the nase.20 The average level of neuroprotection in focal time of administration of treatments. Attempts to de- ischemia models was not significantly different sign clinical trials having greater regard to the perfor- (F[1,416] ϭ 2.684, p Ͼ 0.05) between purely experi- mance of drugs in animal models may assist in the mental agents (mean, 24.4%; SD, 32.9%, n ϭ 351) translation of results. Third, more extensive animal ex- and treatments also used clinically (mean, 31.35%; SD, perimentation may have been undertaken only after 16.7%, n ϭ 66). Furthermore, the outcome in focal clinical use in an attempt to explain negative clinical ischemia experiments was not related to the primary results. Finally, this review encompassed all drugs given hypothesized mechanism of drug action (Fig 3; to patients in the context of acute stroke with the pur- F[9,416] ϭ 1.210, p Ͼ 0.05).
pose of improving brain function: some of these drugsmay not have been considered “neuroprotectants” in Scope of Testing and Its Relationship to the narrow sense, and for this reason they may not have been tested in animal models of ischemia.
Scope of testing and its relationship to experimental Recent developments in the formulation of guide- efficacy was examined separately because it did not sat- lines for design and interpretation of animal experi- isfy the assumptions for analysis of variance. Figure 4 mentation may assist the selection of candidate drugs shows the average level of neuroprotection (maximum for clinical trial. Using a scale based on the STAIR 100%) versus the number of STAIR criteria satisfied recommendations for preclinical testing, it is clear that (maximum 10 points) for each drug. As the quality of experimental efficacy ratings must be viewed in light of evidence increased with more STAIR standards being the scope of evidence supporting the use of a drug. An met, the average level of neuroprotection of drugs intervention should be considered for clinical trial only tended to regress toward the overall mean of approxi- when there is both a high level of experimental efficacy mately 25%. Interventions tested in a broad range of and a diverse body of evidence supporting its clinical experimental paradigms and demonstrating a superior application. In this review, NXY-059 and hypothermia level of efficacy included NXY-059 and hypothermia performed well against both these criteria, and early re- ports from the clinical trial of NXY-059 suggest a po-tential alleviation of stroke-induced disability.25 Other Discussion
considerations relating to cost, drug safety, regulatory In the search for an effective treatment for stroke and approval, intellectual property status, and putative drug other neurological conditions, researchers have drawn mechanism are also clearly relevant, but it may be that from sources as diverse as Malaysian pit viper venom,21 such issues have overshadowed neuroprotective efficacy Polynesian ceremonial beverages,22 aged garlic,23 and O’Collins et al: Neuroprotective Treatments Table 4. Preclinical Evidence for Drugs Administered to Acute Stroke Patients Vascular insufficiencyImmunostimulatoryNootropic AntioxidantReduces leukocyte activationIncreases cerebral blood flow O’Collins et al: Neuroprotective Treatments gen demand (CMRO2),Metabolic and synaptictransmission inhibitor.
aFirst trial ϭ the year in which the drug was first reported to have been given to acute stroke patients.
bLevel of protection ϭ average neuroprotection from infarct volume changes in focal ischemia studies.
cN ϭ number of studies from which the level of neuroprotection has been calculated dϩ ϭ number of experimental contrasts with a significant improvement in outcome in the treatment group vs control; 0 ϭ number of experimental contrasts showing no significant in outcome in the treatment group vs control; Ϫ ϭ Number of experimental contrasts showing a worse outcome in the treatment group vs control.
eSTAIR quality ϭ number of STAIR criteria met by the drug (maximum 10: see Table 3).
fAnalyzed with other group because of low numbers.
No particular drug mechanism distinguished itself Limitations: Identification of Relevant Data on the basis of superior efficacy in animal models of The search strategy used was broad but not deep. For focal ischemia. This may reflect the multifaceted nature instance, we have identified 19 publications describing of the sequelae of ischemic stroke and suggest a role for the efficacy of FK506 and 46 for recombinant t-PA, combination therapy to target multiple processes. Al- compared with 28 and 94 identified through more sys- ternately, it might suggest that our conception of tematic methods9 (also M. R. Macleod and colleagues, stroke needs reformulation. A tendency to exclusively unpublished observations). Systematic review then may frame drug activity in terms of the dominant schema of be combined with meta-analysis to give a more com- stroke damage (eg. excitotoxicity, free radical damage), plete description of the efficacy, and limits to efficacy, coupled with the sometimes arbitrary attribution of a for individual drugs.8,26 There is also evidence, for at drug mechanism to one of several nonmutually exclu- least some neuroprotectants, of a significant publica- sive groupings, might distract from other paradigms tion bias in favor of positive results (M.R. Macleod, with greater explanatory power, thus hindering the de- unpublished observations), perhaps for commercial or velopment of more effective treatments.
other reasons. Furthermore, a strong emphasis was is not proportional to the amount of tissue damaged,then infarct volume is compromised as a measure ofdamage.
Behavioral outcomes and neurological assessments were not included in the quantitative aspects of thisreview. Assessment and interpretation of animal behav-ior presents additional challenges in developing strokemodels. For instance, because of the difficulties inquantifying responses to tasks, acute assessment scalesfor rats tend to rely heavily on motor effects in contrastwith clinical stroke scales. Even when nonmotor sys-tems are tested, the interpretation of the results musthave regard to potential differences between animalsand humans in terms of the importance of differentmodalities, for example, olfaction and whisker move- Fig 3. Average neuroprotection in focal models of ischemia for ment. For these and other reasons, animal studies tend interventions, classified by their putative primary mode of ac-tion (100% ϭ complete protection, 0% ϭ no difference from to place a heavy reliance upon infarct volume com- pared with functional outcome measures more com-monly used in the clinical context.2 Nevertheless, it isrecommended that preclinical studies use both func- placed on the identification and extraction of data tional and histological outcomes.11 Challenges in the from focal ischemia models because they tend to be interpretation of behavioral outcomes are no greater considered a better model of human stroke.11 than those encountered in understanding histologicalend points in stroke models (eg, comparative differ- Limitations: Changing Experimental Standards ences in gray/white matter balance, frontal lobe vol- The scale devised for scope of testing was based on the ume). Inclusion of behavioral outcomes in the review, STAIR recommendations for preclinical neuroprotec- as we have been doing in meta-analyses,8,9,26 can only tive and restorative drug development. These recom- assist in the understanding of experiments.
mendations were chosen as the gold standard as theyrepresent a crystallization of recent thought on animal experimentation in stroke. Nevertheless, the reason- A good animal model should be both reliable and ableness of assessing past experiments using today’s valid, that is, produce consistent, replicable outcomes, standards may be questioned when considerable have sound theoretical underpinnings, and have the progress has been made in our understanding and in- ability to predict the effect of an intervention on clin- terpretation of animal models of stroke. Researchers’ ical outcome. The paucity of positive results in clinical conception of the theoretical underpinnings of strokemodels has undergone much refinement over the pastdecade, and the standards for testing neuroprotectiveefficacy have been raised accordingly. Furthermore, theSTAIR standards may themselves change as moreknowledge is gained on the ability of particular modelsto predict clinical outcome.
Limitations: Drug MechanismEach drug was assigned a primary hypothesized mech-anism of action. These classes were drawn broadly andmay obscure the complex and multifaceted nature ofstroke pathophysiology. Such categorization needsmuch refinement.
Limitations: End PointsThe conclusions rest upon the validity of infarct vol- Fig 4. When drugs are tested more extensively in focal models ume as a measure of stroke damage. To the extent that of ischemia (maximum scope of testing ϭ 10), then the aver- different brain areas are specialized for the performance age level of neuroprotection tends not to differ greatly from the of different tasks, that recovery of brain function is mean level of protection (25%). OFNE ϭ oxygenated fluoro- plastic and that interference with learning and memory O’Collins et al: Neuroprotective Treatments trials, together with the disparity in clinical and exper- V.E.O. is in receipt of an Australian Postgraduate Award scholar- imental protocols, have hindered the validation of neu- ship. No other funds were used to support this review process; how- rological score and infarct volume in animal experi- ever, G.A.D. and D.W.H. have been the recipients of industry and ments as predictors of outcome in human disease.
government funding for research into neuroprotection.
Although animal data would support the use of NXY- We thank D. Young and N. Spratt for their comments, and M.
059 and t-PA in the clinic, such data might also pro- Goldberg for clarifying the status of several clinical trials.
vide tentative support for the clinical application ofmany of the other 114 therapies that have been tested References
in humans but not undergone further clinical develop- 1. Pound P, Ebrahim S, Sandercock P, et al. Where is the evi- dence that animal research benefits humans? Br Med J 2004;328:514 –517.
2. Gladstone DJ, Black SE, Hakim AM. Toward wisdom from Different animal models may be needed to reflect dif- failure: lessons from neuroprotective stroke trials and new ther-apeutic directions. Stroke 2002;33:2123–2136.
ferences in patients with acute ischemic stroke concern- 3. Cheng YD, Al-Khoury L, Zivin JA. Neuroprotection for isch- ing stroke subtype, reperfusion status, and comorbidi- emic stroke: two decades of success and failure. NeuroRx. 2004; ties. Averaging across all results might mask the potential utility of a treatment within a particular sub- 4. Green AR, Odergren T, Ashwood T. Animal models of stroke: class of patients. For instance, thrombolytics such as do they have value for discovering neuroprotective agents?Trends Pharmacol Sci Vol. 2003;24:402– 408.
t-PA tend to produce a better outcome in embolic 5. Grotta J. Neuroprotection is unlikely to be effective in humans models of stroke compared with thread occlusion mod- using current trial designs. Stroke 2002;33:306 –307.
els.27 Techniques such as meta-analysis and regression 6. Liebeskind DS, Kasner SE. Neuroprotection for ischaemic modeling are better suited to unraveling the complexity stroke: an unattainable goal? CNS Drugs Vol. 2001;15: 7. Del Zoppo GJ. Why do all drugs work in animals but none in stroke patients? 1. Drugs promoting cerebral blood flow. J In- More accurate estimates of neuroprotective efficacy will 8. Macleod MR, O’Collins T, Howells DW, Donnan GA. Pool- be afforded through systematic review and meta- ing of animal experimental data reveals influence of study de- analyses for individual drugs, and this is now the focus sign and publication bias. Stroke 2004;35:1203–1208.
9. Macleod MR, O’Collins T, Horky LL, et al. Systematic review of the Collaborative Approach to Meta-Analysis and and metaanalysis of the efficacy of FK506 in experimental Review of Animal Data in Experimental Stroke, the stroke. J Cereb Blood Flow Metab. 2005;25:713–721.
CAMARADES group. Meta-analysis and regression 10. Van der Worp HB, De Haan P, Morrema E, Kalkman CJ.
modeling of pooled data for different drugs may also Methodological quality of animal studies on neuroprotection in help establish whether there is indeed a “baseline” pos- focal cerebral ischaemia. J Neurol 2005;2525:1108 –1114.
11. STAIR. Recommendations for standards regarding preclinical itive efficacy in such studies. Further, these techniques neuroprotective and restorative drug development. Stroke 1999; will help elucidate the determinants of efficacy in ani- 12. Sharp FR, Lu A, Tang Y, Millhorn DE. Multiple molecular penumbras after focal cerebral ischemia [review]. J Cereb Blood 13. Heiss W-D, Thiel A, Grond M, Graf R. Which targets are Stroke continues to kill 5.5 million people each year,28 relevant for therapy acute ischemic stroke? Stroke 1999;30: and the development of safe and effective treatments is a major challenge to experimental and clinical neuro- 14. Lipton P. Ischemic cell death in brain neurons. Physiol Rev science. The systematic review and analysis of data from neuroprotection experiments may bring us closer 15. Hunter AJ, Mackay KB, Rogers DC. To what extent have functional studies of ischaemia in animals been useful in the to achieving this goal. It has been suggested that drugs assessment of potential neuroprotective agents? Trends Pharma- should be taken forward to clinical trial only if data from animal experiments are valid and precise and in 16. Cochrane Central Register of Controlled Trials (CENTRAL).
the public domain before clinical trials occur1: this re- 2005. The Cochrane Collaboration, John Wiley & Sons, Ltd.
view affirms that position. Drugs taken forward to 17. Stroke Center. Clinical trials directory. Online: http:// clinical trial in the past have not been distinguished by 18. Sharma SS. Emerging neuroprotective approaches in stroke superior efficacy in animal models, and when assessing treatment. Current Research & Information on Pharmaceutical experimental efficacy attention must also be given to magnitude and scope of preclinical testing. Together 19. Labiche LA, Grotta JC. Clinical trials for cytoprotection in with greater rigor in the conduct and reporting of an- 20. Nagai N, Vanlinthout I, Collen D. Comparative effects of tis- imal data, there is every prospect that such an approach sue plasminogen activator, streptokinase, and staphylokinase on will improve the transition of scientific advances from cerebral ischemic infarction and pulmonary clot lysis in hamster models. Circulation 1999;100:2541–2546.
21. Huang R, Zeng J, Pei Z. [Effects of defibrinogenase on photo- 25. AstraZeneca. First results from SAINT I trial show Astra- chemically induced middle cerebral artery occlusion in stroke- Zeneca’s Cerovine (NXY-059) demonstrates a reduction in prone renovascular hypertensive rats]. Zhonghua Yi Xue Za Zhi disability in patients with acute ischemic stroke. http:// www.astrazeneca.com/pressrelease/4979.aspx, 2005.
22. Backhauss C, Krieglstein J. Extract of kava (Piper methysticum) 26. Macleod MR, O’Collins T, Horky LL, et al. Systematic review and its methysticin constituents protect brain tissue against isch- and meta-analysis of the efficacy of melatonin in experimental emic damage in rodents. Eur J Pharmacol 1992;215:265–269.
stroke. J Pineal Res 2005;38:35– 41.
23. Numagami Y, Sato S, Ohnishi ST. Attenuation of rat ischemic 27. Meng W, Wang X, Asahi M, et al. Effects of tissue type plas- brain damage by aged garlic extracts: a possible protecting minogen activator in embolic versus mechanical models of focal mechanism as antioxidants. Neurochem Int 1996;29:135–143.
cerebral ischemia in rats. J Cereb Blood Flow Metab 1999;19: 24. Buchan AM, Gertler SZ, Li H, et al. A selective N-type Ca(2ϩ)-channel blocker prevents CA1 injury 24 h following 28. WHO. The World Health Report 2003. Online at severe forebrain ischemia and reduces infarction following focal http://www.who.int/whr/2003/en/Facts_and_Figures-en.pdf, ischemia. J Cereb Blood Flow Metab 1994;14:903–910.
O’Collins et al: Neuroprotective Treatments

Source: http://www.animalexperiments.info/resources/Studies/Human-healthcare/Stroke.-O'Collins-et-al-2006./Stroke-OCollins-et-al-2006-An-Neur.pdf