Published online before print March 23, 2006, doi: 10.1161/01.STR.0000217466.30085.82
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
(Stroke. 2006;37:1150-a.)
© 2006 American Heart Association, Inc.
Letters to the Editor |
Hyperglycemia, Insulin and Acute Ischemic Stroke
Department of Geriatric Medicine, University of Newcastle, Sunderland Royal Hospital, Sunderland, United Kingdom
To the Editor:
We read with interest the recent article by Garg et al, highlighting again the increasing interest within the stroke research community in modulation of physiological variables that could affect stroke outcome.1 Poststroke hyperglycemia has long been recognized as a poor prognostic factor in terms of mortality and functional recovery. It remains to be determined whether active intervention to lower glucose levels after acute stroke can modify clinical outcome. The United Kingdom Glucose Insulin in Stroke Trial (GIST-UK) was designed to address this question, using an intravenous glucose-potassium-insulin (GKI) infusion to induce and maintain euglycemia.2
We are disappointed that Garg and colleagues chose to dispute the relevance of the results of GIST-UK before its completion and without appropriate interpretation of the trial protocol. The final patient will be randomized March 31st, 2006, and the results should be available for presentation and publication in early 2007. Garg’s article reflects a basic misunderstanding of the trial methodology and its context. The aim of GIST is to determine the effectiveness of intervention to maintain euglycemia in routine practice. Although guidelines may currently exist for the management of hyperglycemia, there is no evidence to suggest such intervention is either effective or safe. With respect to target blood glucose levels for the control (saline) group, admission plasma glucose of 17 mmol/L represents the upper limit for inclusion in the trial, not target glucose level. The treatment arm of GIST has a target capillary glucose of 4 to 7 mmol/L, with the objective in the control group being not to intervene unless glucose rises above 17 mmol/L. The most recent published data from GIST demonstrates that the majority of randomized patients have only moderate hyperglycemia (mean admission plasma glucose 8.37 mmol/L).3 Furthermore, without specific intervention plasma glucose levels fall spontaneously in the control group. Thus, any suggestion that patients are being managed with glucose levels of 17 mmol/L is both misleading and incorrect. Our work has demonstrated the safety and efficacy of the GIST GKI regime in the maintenance of euglycemia after acute stroke and currently is the only evidence-based approach from which we can describe how poststroke hyperglycemia can be managed.
We designed the GIST GKI regime to be delivered as part of routine care to the maximum number of eligible patients. In the event of a simple, safe and effective treatment for stroke being discovered, it is probable this will be given to patients with significant levels of comorbidity and pre-existing disability. Thus, the currently favored end point in stroke treatment trials of ‘good outcome’ (modified Rankin Scale [mRS] <1 or <2) is not appropriate; the use of a ‘poor outcome’ as a clinical end point becomes much more relevant. The wish of most stroke patients is the avoidance of the ‘poor outcome’ of death or severe disability. We therefore elected to use mortality at 12 weeks as the primary outcome measure for GIST-UK. Furthermore, as a secondary outcome measure we dichotomized the mRS so that a 12-week mRS of 4 to 6 is categorized as a ‘poor outcome’.4 We estimate that given predicted final recruitment to GIST-UK, the trial will have sufficient power of 90% at the 5% level to detect meaningful differences in nonmortality outcomes between groups, as measured using the mRS. Avoidance of a ‘poor outcome’ reflects the findings of the Stroke Unit Trialists’ Collaboration (reduced dependency, institutionalization, mortality), supporting our belief that this end point is highly relevant.5
There are many issues that arise from clinical trial methodology, but it is inappropriate to dismiss the design and implementation of a trial based on inaccurate interpretation and before publication of results. We look forward to sharing the results of GIST-UK with the stroke research community in the near future.
References
1. Garg R, Chaudhuri A, Munschauer F, Dandona P. Hyperglycaemia, insulin and acute ischaemic stroke. A mechanistic justification for a trial of insulin infusion therapy. Stroke. 2006; 37: 267–273.
2. Scott JF, Robinson GM, French JM, O’Connell JE, Alberti KGMM, Gray CS. Glucose potassium insulin infusions in the treatment of acute stroke patients with mild to moderate hyperglycaemia: the Glucose Insulin in Stroke Trial (GIST). Stroke. 1999; 30: 793–799.
3. Gray CS, Hildreth AJ, Alberti KGMM, O’Connell JE. Poststroke hyperglycaemia: natural history and immediate management. Stroke. 2004; 35: 122–126.
4. Gray CS, Scott JF, O’Connell JE, Hildreth AJ. Assessment of outcome in clinical trials. Stroke. 2000; 31: 232.[Medline] [Order article via Infotrieve]
5. UK Stroke Trialists Collaboration. Collaborative systematic review of the randomised trials of organised in patient (stroke unit) care after stroke. BMJ. 1997; 314: 1151–1160.
Division of Endocrinology, Diabetes and Metabolism, State University of New York at Buffalo, Buffalo, NY
Department of Neurology, State University of New York at Buffalo, Buffalo, NY
Response:
O’Connell and colleagues have raised several questions about our review1 while giving some new information about the GIST-UK. Safety data on GIST was published in 1999, and the trial was projected to end by 2002.2,3 It’s good to know that the last patient will soon be randomized and we are likely to see the results by 2007. We understand the GIST protocol to the extent it has been published. It is possible that that there is more to the GIST trial methodology than published so far. However, the fact remains that there is no intervention in the control group unless glucose rises above 17 mmol/L. If GIST were to be planned today, the threshold for intervention in control group will be much lower because the definition of conventional treatment has changed over time. Lower threshold for insulin treatment in the control group would have led to a bigger sample size. That’s why we suggest a nonmortality primary end point. Although we appreciate the effort to determine the effectiveness of insulin infusion in routine practice, it may be helpful to first establish the effectiveness in any setting. Insulin infusion is still not given outside the special units in most US hospitals. This is probably true of UK as well. Finally, no one can dismiss a major trial like GIST. We are eagerly waiting to see the results of this trial, as is the rest of the medical community. However, if it were to take several more years, a trial on the lines that we suggested in our review might have been worthwhile in retrospect.
References
1. Garg R, Chaudhuri A, Munschauer F, Dandona P. Hyperglycemia, insulin, and acute ischemic stroke: a mechanistic justification for a trial of insulin infusion therapy. Stroke. 2006; 37: 267–273.
2. Scott JF, Robinson GM, French JM, O’Connell JE, Alberti KG, Gray CS. Glucose potassium insulin infusions in the treatment of acute stroke patients with mild to moderate hyperglycemia: the Glucose Insulin in Stroke Trial (GIST). Stroke. 1999; 30: 793–799.
3. Scott JF, Robinson GM, French JM, O’Connell JE, Alberti KG, Gray CS. Blood pressure response to glucose potassium insulin therapy in patients with acute stroke with mild to moderate hyperglycaemia. J Neurol Neurosurg Psychiatry. 2001; 70: 401–404.
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||