Trends in Neurosciences
Volume 30, Issue 9, September 2007, Pages 433-439
Journal home page for Trends in Neurosciences

Opinion
How can we improve the pre-clinical development of drugs for stroke?

https://doi.org/10.1016/j.tins.2007.06.009Get rights and content

The development of stroke drugs has been characterized by success in animal studies and subsequent failure in clinical trials. Animal studies might have overstated efficacy, or clinical trials might have understated efficacy; in either case we need to better understand the reasons for failure. Techniques borrowed from clinical trials have recently allowed the impact of publication and study-quality biases on published estimates of efficacy in animal experiments to be described. On the basis of these data, we propose minimum standards for the range and quality of pre-clinical animal data. We believe the adoption of these standards will lead to improved effectiveness and efficiency in the selection of drugs for clinical trials in stroke and in the design of those trials.

Section snippets

The pathophysiological basis of stroke treatments

We now understand a great deal about the pathophysiology of focal cerebral ischaemia in animal models of stroke [2], and this has spurred the investigation of the potential therapeutic effects of candidate drugs that either block these processes or restore blood flow [3]. Therapeutic approaches include re-instating blood flow (reperfusion), inhibiting pathways promoting neuronal death or augmenting endogenous protective mechanisms (neuroprotection) and promoting plasticity, repair and

A scientific approach to understanding failures in translation

Both laboratory scientists and clinical trialists have recognized translational failure in stroke as an area of major concern, and several recommendations, based on expert opinion, have attempted to set out the circumstances in which translation is most likely to be successful. Foremost among these are the series of recommendations from the Stroke Therapy Academic Industry Round Table (STAIR) [9]. A compound (NXY-059) reported to meet all of their recommendations recently joined the long list

Publication bias

If positive (rather than neutral) studies are more likely to be published [11] (and some journals explicitly favour positive studies), then any conclusions drawn from the published literature will overstate the magnitude of any effect seen. In a group of studies reporting the same phenomenon, the presence of a significant publication bias can be inferred from the relationship between the precision of individual studies and the sizes of the effects seen. In this way, it can be shown that the

Evidence that animal studies might be falsely positive

In clinical trials, it is a universally acknowledged truth that certain aspects of study design can introduce bias to the results of those trials, and this bias usually leads to the overestimation of drug efficacy 14, 15, 16, 17. However, it is less widely accepted that the same truth might apply to experiments in the basic sciences. Although there is very little research on which aspects of study design are most likely to generate bias or the magnitude of the bias thus introduced, several

Evidence that clinical trials might be falsely negative

tPA is one of only three interventions known to be effective for human stroke, so the animal data for tPA can be taken to represent a ’gold standard’ against which the animal data for other drugs can be judged. These animal data are characterized first by their quantity – more than 100 studies involving more than 3000 animals – and second by the observation that tPA was tested in a clinical context [4] under conditions (particularly the interval between stroke onset and the initiation of

Implications for experimental scientists

We believe there have been substantial improvements in the conduct and regulation of animal experiments over the past 50 years. These improvements have occurred through the desire of experimental scientists to continually improve the relevance, generalizability and precision of their work. The current focus on study quality should be seen in the context of this continuing quality improvement.

Certain aspects of study design in experimental stroke – particularly randomisation, allocation

Are animal experiments in stroke different?

It has been possible to measure the impact of potential sources of bias in preclinical studies of drug efficacy in experimental stroke because of the large amount of available data measuring outcome in a broadly similar way (i.e., as infarct volume). Given that the fundamental experimental approach is similar for studies exploring, for instance, stroke pathophysiology or stroke in transgenic animals, it seems highly likely that these experiments will also be susceptible to similar biases.

Implications for drug development in stroke

Those systematic reviews conducted to date have several implications for the successful translation of efficacy from bench to bedside. First, for most candidate neuroprotective drugs, a precise statement of efficacy and the limits to efficacy is not possible because too few experiments have been performed across too narrow a range of circumstances. Second, the quality of individual studies should be considered so that a judgement about the reliability of the contributing data can be made.

Conclusions

Reports of the efficacy of candidate neuroprotective drugs in animal models of stroke are profoundly biased by aspects of study design. Conclusions drawn from individual publications or from narrative reviews cannot provide the basis for selecting drugs for clinical trial or for the design of those clinical trials. A sound judgement on efficacy, the limits to efficacy, the need for any further animal experiments and the design of any ensuing clinical trial can only be made on the basis of a

Acknowledgements

We are grateful to Peter Sandercock and Geoff Donnan for helpful discussions. The work described in this article has been supported by grants from the Scottish Chief Scientist Office, Mrs Dale's Neurology Fund, and the Australian NHMRC (National Health and Medical Research Council).

Glossary

Control group
we use this term to describe the reported untreated or vehicle-treated comparison group. Our recommendations do not relate to other untreated or vehicle-treated comparison groups used, for instance, for model development in pilot studies. Of course, each experiment should include appropriate vehicle-treated controls if meaningful comparisons are to be made.
Random allocation to experimental group
at the start of the experimental treatment, each animal has an equal chance of being

References (30)

  • W. Hacke

    Association of outcome with early stroke treatment: pooled analysis of ATLANTIS, ECASS, and NINDS rt-PA stroke trials

    Lancet

    (2004)
  • Z.M. Chen

    Indications for early aspirin use in acute ischemic stroke: A combined analysis of 40 000 randomized patients from the chinese acute stroke trial and the international stroke trial. On behalf of the CAST and IST collaborative groups

    Stroke

    (2000)
  • I. Kola et al.

    Can the pharmaceutical industry reduce attrition rates?

    Nat. Rev. Drug Discov.

    (2004)
  • U. Dirnagl

    Bench to bedside: the quest for quality in experimental stroke research

    J. Cereb. Blood Flow Metab.

    (2006)
  • Stroke Therapy Academic Industry Roundtable

    Recommendations for Standards Regarding Preclinical Neuroprotective and Restorative Drug Development

    Stroke

    (1999)
  • Cited by (304)

    View all citing articles on Scopus
    View full text