Article Text

Download PDFPDF

Blood pressure response to glucose potassium insulin therapy in patients with acute stroke with mild to moderate hyperglycaemia
  1. J F Scotta,
  2. G M Robinsonc,
  3. J M Frenchb,
  4. J E O'Connella,c,
  5. K G M M Albertia,
  6. C S Graya
  1. aSchool of Clinical Medical Sciences, University of Newcastle-upon-Tyne, UK, bDepartment of Statistics, cCity Hospitals Sunderland NHS Trust, Sunderland, UK
  1. Professor C S Gray, University Department of Medicine for the Elderly, F Floor, Sunderland Royal Hospital, Kayll Road, Sunderland SR4 7TP, UKchris.gray{at}


Insulin is neuroprotective in animal stroke models but its effects in acute stroke in humans are unknown. The Glucose Insulin in Stroke Trial (GIST-UK) is a randomised controlled trial investigating the benefits of maintaining euglycaemia in hyperglycaemic patients with acute stroke. Data are reported from a GIST-UK substudy which sought to determine the influence of glucose potassium insulin (GKI) infusion on blood pressure in acute stroke. All adult patients admitted to hospital with acute stroke with hyperglycaemia (plasma glucose 6.1–17 mmol/l) were potentially eligible. Randomised patients received either a GKI infusion (500 ml 10% glucose, 20 mmol potassium chloride, 16 units of insulin) or control therapy with 154 mmol/l (0.9%) saline at 100 ml/h for 24 hours. BM test strip glucose monitoring was performed 2 hourly, blood pressure monitoring 4 hourly, and plasma glucose sampling 8 hourly. Insulin concentration in the GKI infusate was altered according to test strip values to maintain test strip values between 4–7 mmol/l in the GKI group. Neurological impairment was determined using the European stroke scale (ESS). 145 patients were studied (73 GKI, 72 control). Mean systolic blood pressure was significantly lower during GKI infusion between 4 hours and 24 hours except at 8 hours. Median total ESS scores improved significantly between admission and day 7 in the GKI group (p<0.001) although there was no significant difference in total ESS score between groups at day 7. The significant reduction of systolic blood pressure in acute stroke associated with GKI therapy was not associated with neurological deterioration and may have been beneficial.

  • stroke
  • hyperglycaemia
  • blood pressure

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

There is still no simple, safe, effective treatment for most patients with acute stroke.1 2 Metabolic factors that may influence outcome such as blood pressure, body temperature, and plasma glucose concentration have also been studied but few data exist on how these variables should be managed. Hyperglycaemia in acute stroke is of particular interest because it is common, strongly associated with an adverse prognosis, and is easily modified by simple intervention.3 4

About 75% of patients with acute stroke are hypertensive on admission to hospital.5 Although hypertension resolves in most, around 30% of patients remain hypertensive at 1 week.6 Initial blood pressure is related to outcome in that severe hypertension and hypotension carry a relatively poor prognosis, but it is uncertain whether modulation of blood pressure to optimal levels in the acute phase alters outcome.7 The Glucose Insulin in Stroke Trial (GIST-UK) is the first ever randomised controlled trial to investigate the potential benefits of euglycaemia in hyperglycaemic patients with acute stroke.8 The trial aims to recruit 1200 patients with acute stroke from active centres in the north of England and Scotland by summer 2002. Primary end points comprise mortality from all causes at 3 months and a combined end point of death and severe disability (modified Rankin score>3). Secondary end points consist of neurological impairment (European stroke scale score) and function (Barthel index). Measurement of the blood pressure response to glucose potassium insulin (GKI) infusion in patients with acute stroke has allowed investigation of the haemodynamic effects of exogenous insulin under euglycaemic conditions.


This substudy received approval from Sunderland local research ethics committee. All patients older than 18 years admitted to Sunderland Royal Hospital with a clinical diagnosis of acute stroke (established by CT postrandomisation) within 24 hours of onset were screened to determine admission plasma glucose. Patients presenting after 24 hours or with heart failure (New York Heart Association grades 3/4), renal failure (creatinine ⩾200 μmol/l or urea⩾20 mmol/l), anaemia (Hb<9 g/dl), coma (Glasgow coma scale motor subscore<4), previous disabling stroke (modified Rankin score>3), dementia, isolated posterior circulation syndromes without physical disability, pure language disorders, insulin treated diabetes mellitus (type 1 or 2), or subarachnoid haemorrhage were excluded. All remaining patients with plasma glucose concentrations between 6.1 and 17 mmol/l who gave informed consent (or informed assent from the relative or care giver) were randomised. Trial treatment consisted of a combined infusate of 500 ml 10% dextrose with 16 U human soluble insulin (Actrapid®, Novo Nordisk, Denmark) and 20 mmol potassium chloride (GKI); control treatment was 500 ml 154 mmol/l saline. Treatment was administered through a peripheral vein in the non-paretic arm at a fixed rate of 100 ml/h. Monitoring was identical for each infusion. Blood pressure and pulse were taken at 4 hourly intervals in the paretic arm with an electronic sphygmomanometer and plasma glucose samples were obtained at 8 hourly intervals. Monitoring with BM Glycaemie® test strips (Boehringer Mannheim, Germany) was 2 hourly, with the aim of keeping test strip values between 4 and 7 mmol/l in the GKI group by altering insulin concentration in the GKI accordingly. No action was taken on the test strip values in the control group unless the value rose to >17 mmol/l, in which case intervention with GKI was instituted. The trial infusate ran for 24 hours or until 2400 ml was administered. Venous blood samples for electrolytes, urea, creatinine, and glucose were taken at randomisation, 24 hours, and 48 hours. All antihypertensive therapy was discontinued on admission. Clinical assessments were undertaken at randomisation, 24 hours, 48 hours, and 7 days.

Baseline laboratory and demographic data in the two treatment groups

Mean of systolic, diastolic, and mean (SEM) arterial blood pressure profiles in GKI and control treated patients with acute stroke during the 24 hour trial infusion period. p<0.05 in mean blood pressure between groups (two tailed Student's t test).


Over a 18 month period 631 consecutive patients with acute stroke were screened. Four hundred and eighty patients (76%) were excluded due to admission plasma glucose<6.1 mmol/l (n=196, 31%), presentation>24 hours (n=167, 26%), coma (n=64, 10%), dementia (n=45, 7%), previous disabling stroke (n=30, 5%), pure language disorders (n=26, 4%), previously diagnosed insulin treated type 1 or type 2 diabetes mellitus (n=18, 3%), plasma glucose>17 mmol/l (n=17, 3%), renal failure (n=16, 3%), posterior circulation event without peripheral motor disability (n=13, 2%), rapidly improving neurological deficit (n=12, 2%), cardiac failure (n=7, 1%), and anaemia (n=7, 1%). Of the 480 patients excluded, 138 (29%) satisfied multiple exclusion criteria. Informed consent or assent was gained from 151 (98%) of 154 eligible patients or their relatives or care givers. Seventy eight patients were randomised to GKI therapy and 73 to control treatment. Six patients were withdrawn from the study because of pneumonia (n=1), subdural haematoma (n=1), cerebral tumour (n=2), admission glucose>17mol/l (n=1), and withdrawal of consent (n=1). One hundred and forty five patients were therefore included in this analysis (table).There were no significant differences between the two groups at baseline (table). Therapy with GKI was associated with a significant reduction in mean systolic blood pressure at each 4 hourly measurement point between 4 and 24 hours (p<0.05, two tailed Student's t test) except at 8 hours (p=0.12, figure). Mean arterial blood pressure (diastolic pressure+(systolic pressure −diastolic pressure)/3) was significantly lower in the GKI group at 12 and 16 hours. Mean diastolic blood pressure and heart rate were lower in the GKI group although the differences were not statistically significant. There was no significant difference in median total ESS score between treatment groups on admission or at 24 hours, 48 hours, and 7 days. However, the increase in median total ESS score between admission and 7 days was significant within the GKI group (44.5 to 57; p<0.001, Wilcoxon rank sum) but not within the control group (49.5 to 58; p=0.15). There was also a significant increase in mean plasma potassium concentration over the first 24 hours in the GKI group compared with controls (table).


There is increasing evidence to suggest that control of hyperglycaemia in acute stroke may have beneficial effects on prognosis.4 There is also evidence suggesting that an admission systolic blood pressure of between 140 and 180 mm Hg is associated with a better outcome than a lower or higher admission blood pressure.7 However, it is unknown whether modulation of blood pressure or glucose influences outcome after stroke or whether maintenance of euglycaemia affects the haemodynamic response to cerebral ischaemia.

Potential causes for poststroke hypertension include the stress of hospital admission, raised intracranial pressure, loss of cerebral autoregulation, damage to hypothalamic centres responsible for blood pressure control, and the direct consequences of focal cerebral hypoxia and oligaemia. The final common pathway through which blood pressure is increased in these situations is likely to be increased catabolic hormone release. Euglycaemic insulin infusion is known to influence blood pressure in healthy normotensive humans although the effect is inconsistent in that both hypotensive and hypertensive responses have been found.9 10 Insulin infusion also increases plasma renin activity and angiotensin II concentrations and decreases aldosterone concentrations in healthy humans, but these effects might be expected to raise rather than lower blood pressure.11There is therefore little evidence to support a peripheral (non-cerebral) site of action for the hypotensive effect of insulin infusion, suggesting that our findings may be a consequence of the central effects of insulin on ischaemic cerebral tissue.

Insulin is neuroprotective when administered in animal models of transient focal cerebral ischaemia. Insulin plays a neuroinhibitory part in most brain areas and may protect against ischaemia initiated excitatory damage by increasing cerebral γ-aminobutyric acid (GABA) concentrations.12 Insulin reduces glucose utilisation in many brain areas and may inhibit glucose phosphorylation and thus reduce local lactic acid concentration.13 Insulin also stimulates Na+K+-ATPase, which reduces extracellular cerebral potassium concentrations, decreases brain glucose utilisation, and reduces the risk of cerebral oedema.14 Insulin enhances cerebral noradrenaline (norepinephrine) release and inhibits reuptake thus increasing the extracellular concentration, which may attenuate ischaemic brain damage.15 Any of these mechanisms may limit the extent of the ischaemic damage and as the metabolic stress response to stroke is proportional to the extent of cerebral damage, the likely consequence is attenuation of catabolic hormone release and reduction of systolic blood pressure.16

Current opinion suggests that blood pressure should not be actively lowered in acute stroke to avoid reduction of cerebral blood flow to the ischaemic area where autoregulation is impaired. However, our study suggests that the systolic blood pressure reduction associated with GKI therapy does not cause early neurological decline. An alternative explanation for the hypotensive effect of GKI therapy is that the saline load incurred in the control group augmented blood pressure. We therefore propose to study the blood pressure response in both treatment groups further with automated electronic 48 hour ambulatory blood pressure monitoring and also in an additional control group of patients with acute stroke who will receive no saline therapy.

In summary, GKI therapy in hyperglycaemic patients with acute stroke is associated with a significant hypotensive effect to levels of systolic blood pressure that are currently thought to be associated with optimal outcome. This haemodynamic response may be a consequence of the effects of insulin on ischaemic cerebral tissue rather than insulin's peripheral hypoglycaemic effect.


JFS, JEO'C, CSG, and KGMMA planned the study. JFS and GMR randomised the participants and performed the clinical assessments. JEO'C and CSG supervised the inpatient care of study participants. JFS, CSG and JMF undertook the statistical analysis and all authors contributed to the writing of the manuscript. This study was supported by a grant from the Stroke Association of England and Wales and a Northern and Yorkshire National Health Service Regional Training Fellowship. We thank Dr Basu and colleagues of Sunderland Royal Hospital for their help with patient recruitment and the nursing staff of the admissions unit and stroke unit for their help in implementing the trial protocols.