Hyperglycaemia in acute intracerebral haemorrhage

Background and rationale Acute intracerebral haemorrhage (ICH) results from rupture of cerebral vasculature leading to bleeding into the cerebral parenchyma. ICH represents 10-50% of stroke, depending on the population studied, is associated with significant morbidity and mortality, and has limited treatment options. The INTERACT2 trial was designed to assess the role of blood pressure lowering therapy in intracerebral haemorrhage. Guideline therapy (target systolic blood pressure [SBP] 6.5). Baseline characteristics were summarised as mean (standard deviation [SD]) or median (interquartile range [IQR]) for continuous variables, and as number (%) for categorical variables. Collinearity and interactions between variables were checked. Independent associations between baseline characteristics and level of blood glucose (normoglycaemia and hyperglycaemia) and history of diabetes mellitus, were examined in multivariable logistic regression models. Adjustment was made for all significant baseline variables to determine independent predictors of hyperglycaemia. 90-day clinical outcomes studied were: death alone, major disability lone and death or major disability. Multivariable logistic regression models adjusted by all the significant and clinically important baseline variables as well as significant interactions were used to determine associations of baseline level of blood glucose, both as continuous and categorical (fourths) variables, and clinical outcomes, as well as the association between a history of diabetes mellitus and clinical outcomes. Blood glucose data was also categorised based on diagnostics thresholds (7.0 mmol/L) and sensitivity analyses were performed to further examine associations with primary outcomes. Univariate analysis was also performed to determine associations with 24 hour neurological deterioration, non-fatal adverse events and causes of death. The associations of hyperglycaemia on absolute increase in haematoma and perihaematomal oedema volumes over 24 hours were assessed by an analysis of covariance (ANCOVA) including the same adjusted variables above. Results Of the 2829 ICH patients, 176 were excluded because of missing baseline blood glucose measurements. Of the remaining 2653 patients, 1348 (51%) presented with hyperglycaemia (>6.5 mmol/L) and 292 (11%) had a history of diabetes mellitus. Baseline characteristics of INTERACT2 patients were compared between normoglycaemic and hyperglycaemic patients. All characteristics with significant difference on univariate analysis were included in the multivariate model. On multivariate analyses, the independent predictors of hyperglycaemia at admission were: female gender, patients outside of China, stroke severity (by National Institutes of Health stroke scale (NIHSS) score), systolic blood pressure (SBP), history of diabetes, cortical location of the haematoma, intraventricular haemorrhage (IVH) extension, and haematoma volume (ICH only). In non-diabetic patients, independent predictors were female sex, recruitment outside of China, high NIHSS score (an indicator of more severe neurological impairment), cortical location of ICH, large volume haematoma, and IVH extension. Primary and secondary outcomes were compared by admission blood glucose. There was a strong and near continuous relationship between baseline blood glucose level and death or major disability (odds ratio [OR] 1.29, 95% confidence interval [CI] 1.19-1.40; P<0.0001) and death (OR 1.23, 95%CI 1.12-1.37; P<0.0001) at 90 days, and these variables remained significant when adjusted for other confounders and significant interactions: adjusted odds ratio [aOR] 1.11, 95%CI 1.00-1.24; P<0.0001, aOR 1.16 95% CI (1.01-1.33); P=0.043 for death or major disability and death, respectively. Quartile analysis of admission blood glucose showed significant association with poorer outcomes in the highest quartile (aOR 1.35, 95%CI 1.01-1.80, P=0.015). For secondary outcomes: early neurological deterioration, non-fatal serious adverse events and fatal serious adverse events were all significantly greater in hyperglycaemic patients. Hyperglycaemic patients had significantly higher baseline haematoma volumes, with (14.6 vs 11.6 mL, P <0.0001) and without intraventricular haemorrhage (11.6 vs 10.2 ml, P <0.01), and less deep (78.8% vs. 86.5%, P <0.001) and more cerebellar (5.9% vs. 1.1%, P <0.01) haematomas. Diabetic patients had significantly lower haematoma volume in comparison with non-diabetics (DM 9.2 mL vs. NDM 11.2 mL; p < 0.01) whilst there was no association with haematoma location or IVH extension. Hyperglycaemic patients had no difference in haematoma growth (mean adjusted, 5.1% vs 6.8%, P =0.13) and perihaematomal oedema (mean adjusted 86.0% vs 94.1%, P =0.46) in the first 24 hours. Discussion and conclusions Admission hyperglycaemia and history of diabetes mellitus was shown to be an independent predictor of poor outcome in patients with predominantly mild to moderate severity of ICH (INTERACT2 cohort). Significantly greater adverse events (fatal and non-fatal) and early neurological deterioration was also found in the hyperglycaemic patients. The underlying mechanism of the hyperglycaemia may be a combination of diabetic pathophysiology and the physiological stress response. This was reflected in the determinants of hyperglycaemia which included history of diabetes mellitus, haematoma characteristics (ICH volume, superficial haematoma location, IVH extension), ICH severity by NIHSS score and demographic factors (females, recruitment outside China). Interestingly, diabetic patients presented with significantly lower haematoma volumes and no association with haematoma location or IVH extension, suggesting distinct pathophysiology between diabetic hyperglycaemic and non-diabetic hyperglycaemic patients. Whilst animal models suggest that hyperglycaemia exacerbates haematoma expansion and perihaematomal oedema, these associations were not observed in INTERACT2 analyses. However, hyperglycaemic patients did present with significantly greater IVH extension, baseline haematoma volumes, and of cerebellar and cortical location of the haematomas. More detailed haematoma location analysis is required to determine if specific cerebral regions responsible for glucose homeostasis are affected in hyperglycaemic patients as this will reveal more about the underlying cause for hyperglycaemia. Significantly higher baseline SBP was found in hyperglycaemic patients and further analyses is required to identify the benefit of intensive blood pressure lowering in patients presenting with admission hyperglycaemia. There are limited data and guidelines regarding the management of hyperglycaemia in ICH. Intensive insulin therapy creates the risk of hypoglycaemic events which threatens to worsen outcome. However, the results of this investigation illustrate significant association between hyperglycaemia and adverse outcomes. Based on these findings, there is justification in proceeding with a clinical trial to evaluate benefits and harms of therapy to control hyperglycaemia in ICH