1 Introduction
Stroke is a major global health issue, being the second leading cause of death and third leading cause of disability worldwide (1). In 2019, there were 12.2 million new stroke cases, resulting in 6.55 million deaths and a loss of 143 million disability-adjusted life years (1). Acute ischemic stroke (AIS), which occurs from blocked cerebral arteries, represents about 87% of all strokes (2). The financial impact of stroke treatment and aftercare is also significant. A meta-analysis showed that the worldwide cost of stroke, covering both direct and indirect healthcare expenses, varies between $1,809.51 and $325,108.84, placing a heavy burden on healthcare systems (3).
In AIS patients, hyperglycemia can be triggered by an acute stress response from the hypothalamic-pituitary-adrenal axis and the sympathetic nervous system (4). This hyperglycemia is linked to larger ischemic areas, worsened functional outcomes, and higher mortality (5). Surprisingly, about one-third of AIS patients show stress-induced hyperglycemia even without prior diabetes (6, 7). Current research emphasizes the importance of admission glucose levels, considering them a key modifiable factor for AIS outcomes (8–10). While some studies associate stress-induced hyperglycemia with prognosis, their clinical application can be limited by factors like diabetes control and diet (11–13). To overcome this, Roberts et al. (13) proposed the the concept of stress-hyperglycemia ratio (SHR). In the following studies, SHR was proved to be a potential prognostic indicator in critically ill, acute myocardial infarction, and AIS patients (14–17). This ratio compares stress-induced blood glucose to measures like HbA1c or estimated average glucose (EAG). Multiple formulas exist for calculating SHR, such as fasting glucose (mmol/L)/HbA1c (%) (18). The EAG formula, introduced by Nathan et al., is [(1.59 × HbA1c) – 2.59] (19). Another related metric, the glucose-to-HbA1c ratio (GAR), uses FBG (mg/dL)/HbA1c (%) (17). Regardless of these variations, SHRs provide insight into stress-induced hyperglycemia in AIS, potentially offering a better gauge of its clinical impact.
Our prior research has unequivocally demonstrated that a higher SHR is significantly associated with an elevated risk of adverse outcomes, mortality, early neurological deterioration (END), hemorrhagic transformation (HT), and infectious complications, regardless of diabetes history and treatment methods (20). Nonetheless, further exploration into the influence of dynamic changes in stress hyperglycemia on AIS outcomes is essential. Thus, it is imperative to investigate the dose-response relationship between stress hyperglycemia and the risk of adverse outcomes in AIS patients. Considering the substantial burden that stroke places on healthcare systems and the economy, our study seeks to provide invaluable insights for the primary prevention of stroke.
2 Methods
The current study adhered to the guidelines outlined by the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) (21). Additionally, it was prospectively registered at the International Prospective Register of Systematic Reviews (PROSPERO) with the identifier CRD42023424852 (22). The study's compliance with the PRISMA 2020 checklist can be found in Supplementary Table 1. The objective of this systematic review and meta-analysis was to investigate the relationship between various levels of stress hyperglycemia and the risk of adverse outcomes in patients diagnosed with AIS.
2.1 Search strategy
A systematic search of eligible studies was conducted in several databases, including PubMed, Embase, the Cochrane Library, the China National Knowledge Infrastructure (CNKI), Wan Fang, and the Chinese VIP Information (VIP) database. The search encompassed the period from the inception of the databases up until the end of July 2022, with language restrictions set to English and Chinese. It should be noted that, based on our previous study (20) and the data extraction characteristics in dose-response meta-analysis, a re-evaluation of the full-text articles from the previous 16 included studies was performed, resulting in the identification of 7 studies that met the inclusion criteria for the dose-response meta-analysis. Subsequently, the same search strategy was manually applied to search PubMed and Chinese databases for relevant studies published after August 1, 2022. We searched the Chinese databases manually using the search term “应激性高血糖比值” 或 “应激性高血糖比率” 或 “stress hyperglycemia ratio” 或 “SHR”, and then screened the literature one by one. Additionally, a thorough review of topic-related editorials, perspectives, methodologies, and comments was conducted to identify potentially useful information. Supplementary literature was also manually investigated for other relevant studies. To ensure comprehensive coverage, the reference lists of retrieved articles were checked manually to avoid any omissions. Detailed information regarding the systematic search process can be found in Supplementary Table 2.
2.2 Study selection
Inclusion and exclusion criteria were applied to select the appropriate studies for analysis. The following criteria were used for inclusion: (a) studies had to be published in a peer-reviewed journal; (b) publication had to be in either English or Chinese languages; (c) the exposure of interest had to be different stress hyperglycemia levels assessed by SHR; (d) the study had to investigate adverse outcomes related to stroke; (e) the availability of odd ratios (ORs) or hazard ratios (HRs) with corresponding 95% confidence intervals (CIs) for different endpoints; (f) the reported results had to be adjusted for relevant covariates. On the other hand, studies were excluded if they were in the form of reviews, letters, commentaries, conference abstracts, or case reports, as they did not meet the criteria for inclusion in the analysis.
2.3 Data extraction and quality assessment
Two reviewers conducted the data extraction process independently, using a predefined form that included the following information: first author's name, publication year, country, age, sex, number of participants and cases for different endpoints, SHR levels, endpoints, duration of follow-up, and covariates adjusted in the multivariable analysis. If a study reported multiple adjustment models, only the model with the highest number of adjusted covariates was included.
Furthermore, the quality assessment of the included studies was performed using the Newcastle-Ottawa Scale (NOS) (23). The NOS evaluates the studies based on three aspects: selection, comparability, and outcome. Studies that received more than six stars were considered to be of high quality. Any discrepancies during the data extraction and quality assessment process were resolved through consultation with a third author to ensure consensus.
2.4 The definition of adverse endpoints
The 3-month poor functional outcome was assessed by modified Rankin Scale (mRS), and the standard of the scale can be found in the Supplementary Table 3. According to the European Cooperative Acute Stroke Study II criteria, early neurological deterioration (END) is defined as an increase of ≥4 points in the National Institutes of Health Stroke Scale (NIHSS) score within 24 h after vascular recanalization therapy. Symptomatic intracerebral hemorrhage (sICH) is defined as the presence of brain parenchymal bleeding on head CT scans within 7 days after intravascular treatment, accompanied by an increase of ≥4 points in the NIHSS score compared to the baseline. Based on the study of Merlino et al. (24), intracranial hemorrhage (ICH) and sICH were regarded as hemorrhagic transformation (HT). Stroke recurrence was defined as a new neurological deficit (new ischemic stroke, hemorrhagic stroke, and transient ischemic attack) or re-hospitalization. The primary outcome was 3-month poor functional outcome and secondary outcomes were mortality at 3 months and 1 year, END, HT (ICH and sICH), and stroke recurrence.
2.5 Statistical analysis
In this meta-analysis, the effect sizes were represented by ORs and 95% CIs. ORs and HRs were considered equivalent. If the reference category for the SHR was not the first tertile, the method described by Orsini was applied to convert the reference category to the first tertile (25). When studies reported separate effect sizes for diabetes and non-diabetes subgroups, each subgroup was treated as an independent study. To ensure consistency, SHR levels reported in mg/dl were converted to mmol/L by dividing the values by 18. For each included study, we extracted the SHR values for each category. If the highest category was open-ended, we estimated the mean value by multiplying the upper bound of the category by 1.5. Similarly, if the lowest category was open-ended, we estimated the mean value by dividing the lower bound of the category by 1.5. Studies that provided fewer than three categories of SHR levels or did not report the numbers of cases or participants in each category were excluded from the dose-response meta-analysis. To pool the ORs of the highest vs. lowest categories of SHR from different studies, we employed a two-stage random-effects model (26). Publication bias was assessed using Egger's test (27). We also utilized restricted cubic spline models with three knots to examine potential non-linear trends between SHR and the risk of adverse outcomes (25, 28), calculating non-linear p-values (29). Cochran's Q-test was used to assess heterogeneity among studies, with significance set at P = 0.1 or I2 > 50% (30, 31). All statistical analyses were performed using R version 4.1.2 and the “meta” R package. Statistical significance was defined as P < 0.05.
4 Discussion
We first performed a dose-response meta-analysis on SHR's effects on AIS adverse outcomes in all individuals and specifically non-diabetics. RCS models further indicated a “J-shaped” non-linear relationship between SHR and adverse outcomes like poor functional outcomes at 3 months, mortality at 3 months and 1 year, END, HT, and stroke recurrence. However, this trend was not observed for HT and stroke recurrence. Notably, the worst outcomes were most linked to the highest SHR group, as studies often categorized SHR into tertiles or quartiles.
Shi et al. (52) found in their meta-analysis that blood glucose is crucial for stroke risk, with a non-linear relationship between FBG and stroke risk. As seen in Figure 3, the risk of 3-month mortality for diabetics is on par or higher than for non-diabetics, as shown by Duan et al. (48) and Merlino et al. (43). Yet, the 1-year mortality risk is lower for diabetics, as found by Li et al. (35). These mixed findings underline the complex relationship between diabetes and post-AIS outcomes. The initial 90 days post-AIS are vital for diabetic patients, aligning with previous studies (35, 43, 47, 48) that they may face more severe AIS effects or complications during this period. Hence, immediate optimal glycemic control post-stroke is essential. However, our results also suggest a lower 1-year mortality risk for diabetics, possibly due to long-term interventions like continuous blood glucose control and lifestyle changes. Further research is needed to understand these observations better.
Various equations, such as FGB (mmol/L)/HbA1c (%), have been commonly used to calculate the SHR to assess the relationship between glycemic control and stroke outcomes. However, it is crucial to acknowledge the potential limitations associated with this equation, particularly concerning comorbidities known to affect HbA1c levels, such as hemoglobinopathies or anemia. Comorbid conditions like hemoglobinopathies or anemia can significantly impact HbA1c measurements, rendering the use of the FBG/HbA1c equation less reliable in certain cases. These conditions can lead to variations in HbA1c levels that may not accurately reflect long-term glycemic control. Therefore, caution is necessary when interpreting SHR values in individuals with such comorbidities, as they may result in an overestimation or underestimation of the true association between glycemic control and stroke outcomes. Researchers and clinicians must be aware of these potential confounding factors and consider them when utilizing SHR as a metric to evaluate the relationship between diabetes management and stroke risk. Alternative methods or adjustments may be needed in specific patient populations to ensure a more accurate assessment of glycemic control in relation to stroke outcomes.
SHR provides a more precise measure of stress-induced hyperglycemia by adjusting for prior diabetes history and baseline blood glucose levels, thus better indicating disease prognosis (10, 53). Earlier studies have linked elevated GAR with worse outcomes in acute ischemic stroke patients receiving thrombolytic therapy (24). However, for broader understanding and application, Nathan proposed using EAG over HbA1c or glycated albumin (8, 34, 35). Although SHR offers a more logical and scientifically sound choice, research on its clinical relevance in AIS is limited, especially for those undergoing IVT and EVT (18, 33, 39).
The pathophysiological mechanisms underlying the association between SHR and stroke severity as well as prognosis in patients with acute ischemic stroke are still not fully elucidated. A particular study employed correlation analysis to shed light on this relationship, revealing that higher SHR levels are predominantly linked to increased stroke severity and heightened inflammatory responses upon admission (46). These findings align with a potential mechanistic framework through which SHR may impact stroke prognosis and potentially contribute to a self-perpetuating cycle. On one hand, the occurrence of a stroke represents a stress event that triggers the activation of the hypothalamic-pituitary-adrenal axis and the sympathetic nervous system (4, 54). Dysregulation of neuroendocrine hormones promotes glucagon secretion (55, 56), hinders insulin release, induces insulin resistance, stimulates liver glycogenolysis and gluconeogenesis (57), ultimately leading to elevated blood glucose levels. Moreover, as stroke severity escalates, so does the intensity of the stress response, consequently contributing to higher SHR levels. On the other hand, hyperglycemia has the potential to induce inflammation and the release of vasoconstrictive agents (57). Animal experiments have demonstrated that hyperglycemia can cause cerebral vasospasm through the NO pathway (58), induce platelet aggregation through superoxide production, activate the coagulation system, inhibit the fibrinolytic system, generate free radicals, and make patients more susceptible to ischemic injury (59). Hyperglycemia may also damage the blood-brain barrier, exacerbate cerebral edema, and cause more severe ischemic brain damage (60). Additionally, high blood glucose levels may lead to endothelial dysfunction and oxidative stress in stroke patients, resulting in reduced cerebral blood flow and impaired cerebral blood flow autoregulation, exacerbating ischemic injury (61–63). Finally, hyperglycemia may have direct neurotoxic effects on the ischemic penumbra, leading to neuronal damage (17, 58, 64) and worsening the stroke's severity. This cascade of reactions may culminate in a detrimental feedback loop that ultimately influences the clinical prognosis of stroke patients.
China, due to its population and rising risk factors, is witnessing a surge in AIS incidence. Multiple studies from China have detailed the relationship between stress hyperglycemia and AIS outcomes (32, 34, 35, 39, 44, 46–51). Stress hyperglycemia is consistently associated with increased mortality, functional impairment, and complications following AIS (10, 24, 34, 35, 39, 43, 44, 48, 49, 51). Zhang et al. (47) linked high SHR with poor 90-day outcomes in non-diabetic AIS patients. Interestingly, the impact of stress hyperglycemia varies based on the presence of diabetes. Non-diabetic patients with severe stress hyperglycemia face higher risks, while the same doesn't hold for diabetic patients, potentially due to chronic exposure to hyperglycemia (48). Regardless of diabetes status, SHR predicts stroke recurrence in AIS patients with intracranial atherosclerotic stenosis (50). The findings emphasize the need for personalized glucose control and diligent glycemic management after a stroke, especially considering its effects on treatment outcomes.
While Perez-Vega et al. (65) indeed demonstrated the prognostic significance of glucose levels at admission for patients undergoing MT for large-vessel occlusion, it's essential to consider the broader context of glucose monitoring and its relation to stress hyperglycemia. Blood glucose levels upon admission, though clinically simpler, offer a snapshot of the patient's current metabolic state. However, they might not provide a complete picture of the patient's physiological response to acute stress, which is where the SHR becomes particularly relevant. The SHR offers a more comprehensive insight into the patient's acute stress response. Stress hyperglycemia, or the acute elevation of blood glucose in response to physiological stressors, is different from chronic hyperglycemia. By using SHR, we can better differentiate between patients who have high glucose levels primarily due to their chronic metabolic state (e.g., uncontrolled diabetes) and those who are experiencing a significant stress response. Furthermore, while the timely adjustment of SHR by HbA1c level might be a concern due to the lag in HbA1c measurement, it's important to note that HbA1c represents an average glucose level over ~2–3 months. Hence, a recent HbA1c measurement (within the last month or so) would be sufficiently relevant for calculating SHR in the acute setting.
To the best of our current knowledge, this dose-response meta-analysis represents a pioneering effort in exploring the impact of dynamic fluctuations in stress hyperglycemia on susceptibility to adverse outcomes among individuals with AIS. Our findings unequivocally establish a positive association between the Stress-Hyperglycemia Ratio (SHR) and the risk of adverse outcomes in AIS patients. Furthermore, we have observed a distinct “J-shaped” non-linear dose-response pattern, indicating that elevated SHR is significantly correlated with an increased likelihood of 3-month poor functional outcomes, 3-month and 1-year mortality, as well as neurological deficits. These conclusions are in alignment with prior research, thereby deepening our understanding of the influence of dynamic changes in stress hyperglycemia on the risk of adverse outcomes in AIS patients. In summary, these investigations provide invaluable insights into the intricate relationship between stress hyperglycemia and AIS patient outcomes, underscoring the critical importance of implementing effective glycemic management strategies within this specific population. Further research endeavors are warranted to delve into the complex interplay between glucose control approaches and prognosis among stroke patients originating from diverse ethnic backgrounds and populations.