High blood sugar (hyperglycaemia) affects 40% of acute stroke patients and has a major adverse effect on survival and recovery. Increased production of lactic acid in brain tissue that has a poor blood supply is postulated to be the mechanism by which high blood sugar may worsen brain injury after stroke. Treatment with insulin infusions is proposed as a neuroprotective strategy, and a clinical trial is ongoing to test this hypothesis. However, the biological basis for insulin treatment has not been established, and there is uncertainty about the duration of insulin infusion that may be required to limit damage.

Magnetic resonance spectroscopy (MRS) is a brain scanning technique that allows measurement of brain lactic acid. When performed in conjunction with conventional MRI scanning, the relationship of lactate accumulation to stroke expansion can be established. SELESTIAL is a randomised, placebo-controlled trial of insulin infusions of 24 or 72 hours (h) duration in acute stroke patients with hyperglycaemia, to establish whether insulin prevents lactate accumulation over the initial 72h after stroke, how this relates to stroke evolution, and the effect of treatment on stroke size and clinical outcomes at 1 week.

Hyperglycaemia is present in 40-60% of patients with acute ischaemic stroke and adversely affects survival and outcome. This effect is independent of stroke severity or pathology, and is most striking in patients without recognised diabetes, in whom the odds of death are increased threefold, and the chance of poor functional outcome by 40%. Hyperglycaemia remains a powerful independent predictor of outcome even in the face of thrombolytic drug therapy.

The adverse effect of hyperglycaemia is hypothesised to be consequent to increased provision of substrate to hypoperfused tissue that is metabolising anaerobically, with resultant tissue accumulation of neurotoxic lactic acid. In animal models of stroke, hyperglycaemia causes increased tissue lactic acidosis and increased recruitment of ischaemic tissue in the peri-infarct region into the final infarct. Infarct volumes are higher in hyperglycaemic animals, and conversely, reducing blood glucose reduces infarct volume. Although clinical observational studies suggest protocols that incorporate blood glucose monitoring and control to be beneficial, and trials are ongoing to define the impact of routine treatment to maintain euglycaemia, the basic pathophysiology of stroke in relation to blood glucose has not been well defined in man. Preliminary studies confirm a relationship between blood glucose and lactate concentration in hypoperfused brain tissue but it is unknown whether brain lactate is reduced by control of blood glucose, and whether doing so will impact on stroke evolution. It has also been found that infarct volume increases more in hyperglycaemic patients treated with recombinant tissue plasminogen activator (rtPA).

MRI permits non-invasive and serial study of acute stroke pathophysiology. In addition to brain structure, MRI can define tissue viability (cytotoxic oedema seen on diffusion-weighted imaging, DWI), brain perfusion (bolus-tracking perfusion imaging, PI), vascular integrity (MR angiography, MRA) and tissue metabolism (1H MR spectroscopy, MRS). In acute middle cerebral artery (MCA) occlusion, evolution of cerebral damage has been defined with these techniques. The volume of hypoperfused tissue on PI initially exceeds the DWI lesion, and, over time, the DWI lesion expands to finally incorporate the majority of the PI lesion. The region of tissue with normal DWI but abnormal PI is thought to correspond to the "ischaemic penumbra", the region where hypoperfusion causes electrical failure of neurones with progression to infarction over time due to adverse metabolic and neurochemical events. The fate of the penumbra may be determined by treatment - e.g. it is salvaged by thrombolysis - and it is this penumbral region that is vulnerable to hyperglycaemia-related lactic acidosis.

Glucose lowering with insulin is an inexpensive, and widely applicable treatment. However, current clinical trials are compromised by uncertainty over the ability of treatment to influence pathophysiology, and have necessarily relied upon a "best guess" for treatment duration. Definition of the biological basis for insulin treatment by MRI criteria and comparative data for different treatment durations would strengthen and inform any positive effect from clinical trials, or prevent premature abandonment of this therapeutic modality should trials be neutral.

Inclusion Criteria:

• Acute stroke within 24h of onset
• Lesion on acute diffusion weighted MRI
• >18 years age

Exclusion Criteria:

• Coma
• Major concomitant illness limiting survival