Potential link between post-acute ischemic stroke exposure to hypoglycemia and hemorrhagic transformation

Abstract

Hemorrhagic transformation (HT) is a severe complication of acute ischemic stroke (AIS) owing to its limited treatment options and poor prognosis. In the last decade, the rates of HT incidence have been associated with blood glucose levels. In particular, hyperglycemia at the time of admission has been associated with increased rates of HT in AIS patients. Recent pilot clinical trials have attempted to use intensive insulin therapy during stroke treatment to reduce the severity of cerebral infarction and possibly alleviate the risk of HT. However, the results of these studies have shown no clear clinical benefit. In addition, intensive insulin therapy has increased rates of hypoglycemia which may be associated with larger infarct growth. We hypothesize that hypoglycemia, similarly to hyperglycemia, is a risk factor for worse outcomes in AIS by promoting HT. This review serves to call attention to patterns present within intensive insulin therapy trials and shed light into the pathophysiological effects of hypoglycemia. It is critical that efforts be directed toward the prevention of HT by optimizing insulin therapy during the treatment of AIS.

Introduction

Hemorrhagic transformation (HT) is a term that encompasses all types of post-ischemic hemorrhages. It is observed in approximately 8.7% of reported acute ischemic stroke (AIS) cases (1) and commonly occurs in the presence of thrombolytic therapy (2-4). As a result, a considerable amount of research has been dedicated to minimizing the risk of HT in order to improve patient outcomes. Acute hyperglycemia has been found to be an independent risk factor for HT, specifically for symptomatic intracranial hemorrhage (SICH) (5) and parenchymal hematoma (PH) (6, 7). When compared to normoglycemia, the risk of HT triples when blood glucose rises above 150 mg/dL (6). In addition, multiple studies have shown acute hyperglycemia to be associated with higher mortality rates and worsened functional outcomes (8). This association has also been observed in animal models (9, 10).

The identification of hyperglycemia as an independent risk factor for HT has led to the development of eight randomized pilot clinical trials seeking to correct admission hyperglycemia by utilizing intensive insulin therapy (IIT) (11-18). Within these studies, the primary goal was to establish an effective and safe IIT methodology, as well as to begin testing clinical outcomes following the correction of admission hyperglycemia. However, a meta-analysis of these trials has concluded that IIT does not improve mortality or functional outcomes. Therefore, IIT in the setting of AIS cannot be generally recommended (19). Moreover, implementing an IIT regimen has shown to increase the risk for hypoglycemia (20, 21). Acute episodes of hypoglycemia have been associated with increasing cerebral infarct size in animal stroke models (22, 23). The recurrent exposure to hypoglycemia also appears to increase cerebral ischemic damage (24). The rate of HT in these animal models was not reported.

In summary, hyperglycemia has been shown to increase the rate of HT following AIS. Clinical trials attempting to alleviate hyperglycemia through the use of IIT have not been shown to improve clinical outcomes and inadvertently expose patients to hypoglycemia. The effects of hypoglycemia on the rate of HT is currently unknown. In this review article we present evidence and arguments that supports the hypothesis that hypoglycemia in AIS patients may increase the risk for HT.

Pathophysiological Effects of Hypoglycemia

The pathophysiology of hypoglycemia most notably affects the cardiovascular and nervous systems. Considering the nature of these effects, it is plausible that these effects may influence the severity of stroke and HT following thrombolysis treatment. Longer durations of type 1 diabetes mellitus are known to have more severe effects; however, acute hypoglycemia may have additional adverse effects that have yet to be studied (25). Nonetheless, a variety of mechanisms, both acute and chronic, are thought to be involved.

Inflammatory Cytokines

Acute responses to hypoglycemia, both in patients with diabetes mellitus and those without, include the release of inflammatory markers that can be measured in venous blood. These markers include P-selectin, CD40 expression, IL-6, TNF-alpha, endothelin-1 and C-reactive protein, all of which have been shown to increase during acute episodes of hypoglycemia (26-29). The release of these inflammatory markers is thought to aggravate chronic vasculopathy, and contributes to the onset of macrovascular events such as ischemic stroke. Inflammatory markers, such as endothelin, promote vasoconstriction which have been linked to worsening severity of strokes (28). In addition, inflammatory cytokines, including IL-1, have been shown to increase the severity of hypoglycemia during each episode (30). Insulin, on the other hand, is known to reduce inflammatory markers when euglycemia is maintained, providing evidence of its anti-inflammatory properties (31).

Beyond the initial impact of worsening the severity of AIS, the inflammatory response also leads to the disruption of the blood-brain barrier (BBB). The release of pro-inflammatory cytokines including IL-1 and TNF-α lead to the recruitment and extravasation of leukocytes, compromising the endothelial lining (32). Matrix metalloprotease (MMP) dysregulation, most notably MMP-9, has been tested in experimental models as a possible mechanism for HT following AIS (33). The mechanism of MMP-9 is by degradation of the extracellular matrix, a component of the BBB. These results have been translated to human studies, evaluating MMP-9 serum concentration in AIS patients. MMP-9 was found to be an independent predictor of HT in all stroke subtypes (34). In humans, MMP-9 serum concentration has been shown to increase during acute hypoglycemia (35). Putting it together, the disruption of the BBB then leads to the onset of HT (32).

Acute Hypertensive Response

In response to hypoglycemia, counter-regulatory changes occur in order to raise blood glucose levels (e.g., freeing glucose from glycogen storage and promoting hepatic production of glucose). In addition, the autonomic nervous system induces activation of the sympathoadrenal response resulting in the diffuse release of epinephrine and norepinephrine into the blood circulation. Hemodynamic effects of catecholamine release include tachycardia, increased systolic blood pressure, increased myocardial contractility and decreased central venous pressure (36, 37). In addition to hypoglycemia, hyperinsulinemia has also been associated with an enhanced sympathoadrenal response (38). It is possible, therefore, that the overcompensation of hyperglycemia during AIS by exogenous insulin, evidenced by hypoglycemia, leads to this elevated response.

The acute elevation of blood pressure in the presence of AIS, known as acute hypertensive response (AHR), is independently associated with poor outcomes (39). AHR has been shown to occur in over 60% of AIS patients (40). Animal models have shown AHR increases the rate of HT (41) and reducing blood pressure via labetalol treatment, decreases the rate of HT significantly (42). So, hypoglycemia is thought to promote HT via AHR secondary to an enhanced sympathoadrenal response to AIS. Previous clinical studies have already shown hypertension to be an independent risk factor for HT (43, 44).

Platelet Activation

Platelet activation has been discovered to occur during acute episodes of hypoglycemia by measuring several factors involved in the coagulation cascade. Serum fibrinogen and Factor VIII have been shown to increase and total platelet count typically declines (45). Activated platelet markers, including soluble P-selectin and platelet-monocyte aggregation have also been shown to increase during episodes of hypoglycemia (26). Putting this together, acute platelet activation leads to the progression of an endothelial pro-inflammatory, pro-thrombotic environment that may worsen AIS initially, and leads to an environment that is prone to hemorrhage owing to both vascular weakness and dysregulation. Studies have shown that AIS patients who undergo HT have lower levels of a prothrombotic subset of platelets, known as coated platelets, when compared to AIS patients without HT (46). Unfortunately, very few studies have evaluated the effects of hypoglycemia and platelet function in relation to HT so this mechanism remains with speculation.

Lessons from Intensive Insulin Therapy in Hospital Settings

Clinical trials outside the realm of stroke have implemented IIT attempting to improve patient outcomes. Herein, we discuss the limitations of glucose monitoring and the risk of hypoglycemia observed in these trials.

A clinical trial by Van den Berghe et al was the first to show that the administration of IIT in patients admitted to the surgical intensive care unit significantly reduced in-hospital mortality (47). However, the strict inclusion criteria resulted in a highly-specific patient population. All patients were mechanically ventilated. Calorie intake was standardized by total parenteral nutrition or total enteral feedings, allowing insulin administration to be accurately titrated. Even still, 39 patients receiving IIT had at least one episode of severe hypoglycemia (< 40 mg/dL). However, the results from this study cannot be extrapolated to patients admitted to the hospital with other severe illnesses, such as AIS. Subsequent trials failed to show a mortality benefit in critically ill patients (48).

Normoglycemia in Intensive Care Evaluation—Survival Using Glucose Algorithm Regulation (NICE-SUGAR) trial was conducted to test if IIT reduces mortality in critically ill patients (49). Patients admitted to intensive care units (both surgical and medical) requiring three or more days of treatment were included in the trial. Patients were randomized to receive either IIT or conventional treatment. More than 20% of patients in each group suffered severe sepsis at randomization and 15% in each group suffered from a traumatic injury. The number of patients with stroke was not reported. Episodes of severe hypoglycemia (≤40 mg/dl) occurred in 6.8% of patients receiving IIT and 0.5% of patients receiving conventional treatment (49). However, the incidence of mild to moderate hypoglycemia (40 – 70 mg/dl) was not reported. Interestingly, the mortality rate at 90 days was significantly increased in the IIT group compared to the control group (27.5% vs 24.9%). Reasons for this are poorly understood, but hypoglycemia secondary to IIT could have played a role.

Due to the increased risk of severe hypoglycemia reported in the previous clinical trials, the optimal range of blood glucose during insulin therapy has since been reevaluated and new established guidelines are more lenient (50). Multiple meta-analyses have come to similar conclusions (20, 51, 52). The review by Ellahham (53) expands on this idea and revisits the strategies proven to be effective in critically ill patients, encompassing all patients presenting with hyperglycemia.

The aforementioned clinical trials reveal limitations of IIT, specifically the risk of hypoglycemia and the lack of a significant decrease in patient mortality. A major flaw in the assessment of hypoglycemia events is basing the quantification on the frequency of symptomatic hypoglycemia. The well-known concept of hypoglycemia-associated autonomic failure makes this assessment difficult. Chronic exposure to episodes of hypoglycemia lead to defective glucose counter-regulation and hypoglycemia unawareness by shifting glycemic thresholds for the sympathoadrenal response (54). This leads to the vicious cycle of recurrent hypoglycemia and worsening glycemic control. Therefore, routine blood sampling should be the primary source of evidence when determining episodes of hypoglycemia in the hospitalized patient.

Effect of Glucose-Lowering Therapies in AIS patients

Several studies have been dedicated to evaluate the safety and feasibility of hyperglycemia correction strategies in AIS patients. The UK Glucose Insulin in Stroke Trial (GIST-UK) was the first clinical trial to study the efficacy of insulin therapy in patients suffering from acute stroke (12, 55). The aim was to determine whether treatment with IIT immediately after AIS reduces mortality at 90 days. The intervention arm received variable-dose-insulin by glucose-potassium insulin (GKI) infusion. Episodes of hypoglycemia (<72 mg/dL over a period >30 minutes) occurred in 73 patients. An additional 187 patients were noted to have plasma glucose <72 mg/dL but did not meet the temporal requirement (55). The results of the study revealed no significant reduction in mortality with GKI infusion. Being the first clinical trial for insulin treatment in acute stroke, this study had multiple limitations. It was terminated prematurely due to insufficient enrollment. This resulted in an underpowered statistical analysis. HT was not a measured outcome. However, 7 patients receiving GKI had a recurrent stroke within 72 hours after the initiation of treatment compared to 3 patients in the control arm. Because of the small numbers, no clear conclusion can be drawn.

Spectroscopic Evaluation of Lesion Evolution in Stroke: Trial of Insulin for Acute Lactic Acidosis (SELESTIAL) was developed following the completion of the GIST-UK and utilized a similar IIT protocol (14). 76% of patients receiving GKI therapy became hypoglycemic at least once with a total of 42 episodes recorded. There was no significant difference in brain infarct growth between the two groups. HT and 90-day mortality were not measured outcomes in this study.

Several other trials, including The Glucose Regulation in Acute Stroke Patients (GRASP), The Treatment of Hyperglycemia in Ischemic Stroke (THIS), Staszewski, Kreisel and Walters applied various IIT methods in AIS patients. The GRASP trial noted hypoglycemic episodes (<55mg/dL) in 30% of patients in the IIT group (17). Kreisel et al observed 25 incidents of hypoglycemia (<60 mg/dL) occurring in 8 patients, 7 of which were in the IIT group (13). Walters noted only one episode of hypoglycemia in the IIT group (18). The THIS trial noted 35% of patients receiving IIT had at least one episode of hypoglycemia (<60mg/dL), 64% of which were asymptomatic (11). Staszewski noted 8% of patients in the IIT group experienced symptomatic hypoglycemia (<60 mg/dL) (16). Of note, the Staszewski trial was the first to include only patients without diabetes during enrollment.

Evidence from these trials suggests that, undoubtedly, the risk of hypoglycemia increases with the use of IIT. Also, because each trial used their own unique IIT protocol, the rates of hypoglycemia varied depending on their definition of hypoglycemia, the frequency of blood glucose measurement and the intensity of insulin therapy. The INSULINFARCT trial used a unique IIT protocol with the addition of brain MRI analysis for the assessment of infarct growth. Episodes of asymptomatic hypoglycemia (<54 mg/dL) occurred in 5.7% of the IIT group and 0% in the control group. None of the patients experienced symptomatic hypoglycemia. Additional analysis showed that when hypoglycemia was defined as <64 mg/dL instead, 34.5% of the IIT group and 1.1% of the control group had an episode of hypoglycemia. Infarct growth was significantly larger in in the IIT group. Of note, symptomatic extracranial HT occurred in 12.5% who received IIT and 3% who received SIT (15). This was the first trial to use imaging analysis with an IIT protocol in AIS patients. The increased infarct growth and higher rates of extracranial HT in the IIT group may have been related to the increased rates of hypoglycemia observed during treatment.

With the recent advent of endovascular mechanical thrombectomy (MT), the association between blood glucose and outcomes in patients treated with MT has gained clinical attention. Due to its superior ability to achieve complete reperfusion compared to intravenous thrombolytics, MT is thought to more clearly analyze the relationship of hyperglycemia and complete reperfusion in affecting the outcomes of AIS. The multicenter randomized trial by Kim et al discovered patients undergoing MT for AIS who presented with hyperglycemia had lower rates of excellent outcomes when compared to patients without hyperglycemia (13% vs 34%) (56). Of note, the rates of SICH and any ICH occurred more often in patients with lower blood glucose levels. However, when comparing blood glucose levels with patients who did not have SICH or any ICH, the results were not statistically significant (56).

After reviewing each clinical trial in detail, several common themes have come to light. First, there appears to be a lack of consensus on the definition of hypoglycemia. As of 2009, the American Association of Clinical Endocrinologists and American Diabetes Association came together to define hypoglycemia as plasma glucose less than 70 mg/dL (50). In the trials that defined hypoglycemia at a lower value, episodes of milder hypoglycemia were most likely overlooked. In addition, there seems to be a lack of uniformity when it comes to IIT methodology. Even though the pilot studies were primarily implemented to test various IIT methods to optimize efficacy and safety, there remains no clear conclusions for optimum therapy. Finally, the clinical evidence suggesting hypoglycemia leads to HT is currently limited. Most studies involving IIT and AIS did not include rates of HT in their results nor did they associate periods of hypoglycemia with HT.

Conclusion

Studies have shown that patients presenting with AIS and hyperglycemia have worse outcomes; however, as of 2017, trials have failed to improve outcomes through the utilization of IIT. The greatest concern when utilizing IIT is the risk of hypoglycemia. Therefore, an insulin protocol which corrects hyperglycemia while minimizing the duration and severity of hypoglycemia has been highly sought after. The pilot trials mentioned in this review have given great insight into the optimum implementation of IIT. Future studies must adapt similar protocols and use appropriate follow-up screening tools in order to overcome previous limitations. The literature also indicates that hypoglycemia, by multiple potential mechanisms, may contribute to HT in AIS patients. It is paramount that more attention be directed towards optimizing insulin therapy in the setting of AIS, while appropriately assessing for adverse effects, especially in regards to HT and episodes of hypoglycemia. Until larger randomized controlled studies are completed, it is wise to maintain the blood glucose in a range of 140 to 180 mg/dL in all patients with AIS according to the current guidelines of the American Stroke Association (57).