The Association between Impaired Glucose Regulation and Prognosis of Chinese Patients with Intracerebral Hemorrhage

This study aimed at observing the influence of impaired glucose regulation (IGR) on 1-year outcomes in patients with intracerebral hemorrhage (ICH). Patients hospitalized for ICH from 2008 to 2009 were recruited consecutively at 35 centres across China. A standard oral glucose tolerance test at day 14 ± 3 after stroke onset or before discharge was performed to identify IGR. The outcomes were death (modified Rankin scale [mRS] score of 6), dependency (mRS score of 2 to 5) and poor outcome (mRS score of 2 to 6) at 1 year. Cox proportion hazard model for death and logistic regression model for dependency and poor outcome were performed to investigate the influence of IGR on 1-year outcomes. A total of 288 non-diabetic ICH patients were included in this analysis, among which 150 (52.1%) were IGR. IGR was associated with 1-year dependency (adjusted odds ratio [OR] 2.18, 95% confidence interval [CI], 1.19–3.99; P = 0.01) and poor outcome (adjusted OR 2.17; 95% CI, 1.24–3.80; P = 0.007) of patients with ICH. However, IGR showed no significant association with 1-year death (adjusted hazard ratio 1.49, 95% CI, 0.60–3.67; P = 0.39). IGR was independently associated with 1-year poor outcome of ICH in Chinese patients, with more important influence on dependency than death.

included in this analysis (Fig. 1). The clinical characteristics and proportion of IGR (41.9% versus 52.1%; P = 0.09) were well balanced between patients included and those lost to follow-up ( Table 1).

Association between IGR and outcomes of ICH. Univariable analyses of risk factors for 1-year out-
comes of ICH are presented in Table 3. Outcomes of ICH were associated with body mass index, hematoma location, National Institutes of Health Stroke Scale (NIHSS) score on admission and Glasgow Coma Scale (GCS) score on admission. The 1-year outcomes of patients with IGR or NGR were compared and presented in Table 4. Patients with IGR had higher proportion of dependency (43.9% versus 25.6%, P = 0.002) and poor outcome (50.0% versus 29.9%, P < 0.001), comparing with patients with NGR. After adjusting for potential covariates, IGR was associated with 1-year dependency (adjusted odds ratio [OR] 2.18; 95% confidence interval [CI], 1.19-3.99; P = 0.01) and poor outcome (adjusted OR 2.17; 95% CI, 1.24-3.80; P = 0.007). However, there was no significant association between IGR and death (10.7% versus 5.8%; adjusted Hazard ratio [HR] 1.49; 95% CI, 0.60-3.67; P = 0.39).

Discussion
Our study indicated that IGR after ICH was associated with 1-year poor outcome and dependency of Chinese patients with ICH. No clear association was observed between IGR and 1-year mortality.
Most studies about IGR focused on the prevalence of IGR in patients with stroke 7,8 and the prognosis value of IGR in outcomes of ischemic stroke 4,9 . For ICH patients, there was only 1 previous study on IGR and outcomes in a mixed stroke population including 168 patients with ICH 6 , showing the adverse influence of IGR on short-term unfavourable outcome. However, the definition of IGR was based on fasting glucose levels measured on day 2-4 after admission in that study. In our study, a standard OGTT was performed on day 14 ± 3 after onset or before discharge, and patients in acute stage of ICH were not performed with the test. Therefore, the influence of stress response on IGR was relatively minor, which emphasized the direct effect of abnormal glucose regulation on outcomes.
Isolated IGT accounted for a large proportion of IGR in our study, which was consistent with the characteristic that isolated IGT was common among Chinese patients with IGR 10 . Considering the distinct pathophysiological mechanisms between IGT and IFG 11 , the prognosis utility of IGT or IFG might be different. Unfortunately, our study failed to invest this point due to limited sample size.
The association of IGR and poor outcome in ICH could be explained by several pathophysiological disturbances. First, insulin resistance, as a major feature of IGR, is a state of decreased responsiveness to normal circulating levels of insulin, which could block the insulin signaling in neurons 12 . Previous studies have documented the increased neuronal damage with blocked insulin signaling 13,14 . Second, insulin resistance may frequently be accompanied by hyperinsulinemia. Exposure to hyperinsulinemia rather than hyperglycemia, has been proven to decrease the responsiveness of Akt activation 15 . Akt is a serine/threonine protein kinase involving in many signaling pathways to regulate cellular nutrient metabolism, cell growth, apoptosis and survival, the decreased activation of which may increase neuronal damage and thereby adversely influence functional recovery [16][17][18][19] . This may explain why IGR could influence functional outcome though the glucose was not remarkably increased. Third, insulin-like growth factor I (IGF-I), an important endogenous neurotrophic factors expressed in adult brain, was proved to reduce cell loss and improve long-term neurological function in animal models 20,21 . Previous study has reported that the stroke outcome could be impacted by IGF-I serum levels 22 . IGF-I could be suppressed by hyperinsulinemia and hence to influence the recovery and regenerative of damaged brain tissue 23 .
Our study shows inconsistent influence of IGR after ICH on dependency and death. The possible explanations are listed as follows. First, the majority of death after ICH occurs in the first month after ICH [24][25][26] . The harmful influence of IGR on brain, such as increasing neuronal damage and inhibiting neuronal repair and regeneration, may not be able to induce a statistically mortality differences in such a short time. Second, most early death after ICH results from the direct neurologic consequences of hemorrhage. It may be unable to statistically evaluate the relatively "minor" influence of IGR on death because its potential contributors to death is to some extent obscured by the hemorrhage severity-associated predictive factors such as the hematoma volume, location and neurologic function level 27 . However, for those survivors, it may be possible to distinguish the independent contribution of "minor" prognostic factors and hence the association between IGR and dependency was observed.

Variables
Included (n = 288) Lost to follow up (n = 93) P Value Age (year), mean (SD) 57. In our study, no clear association was observed between IGR and 1-year mortality. However, previous result from ACROSS-China indicated that IGR predicted 1-year mortality of Chinese patients with ischemic stroke. The possible explanation for the contrary death prognosis between ischemic and hemorrhagic stroke is the timing of death. As we mentioned above, it may take time for IGR to show its contribution to death. Comparing with ischemic stroke, the majority of death after ICH occurs in the earlier stage [24][25][26]28 and mainly depends on factors such as the hematoma volume and location 27 . Under the obscuration of early death associated factors and short duration, it may be difficult for IGR to present a significant contribution to death in ICH. While for ischemic stroke, with relatively low early mortality, the contribution of IGR to death may be able to present statistically in a relatively long time.
The prevalence of IGR was high (26.3-32.8%) in patients with ICH 29,30 . Therefore, the association between IGR and clinical outcomes of ICH patients deserved investigation. To our knowledge, this is the first study reported the adverse influence of IGR on long-term clinical outcomes in solely ICH patients. The glucose management for patients with ICH in the current guidelines focused on the control of hyperglycemia, which didn't mention the detection and intervention of IGR 31 . Considering its adverse effect on poor outcome, IGR deserves clinical concern in patients with ICH. Previous trial evaluating IGR intervention in non-diabetic patients with ischemic stroke and insulin resistance has showed a benefit effect on stroke recurrence 32 . It is likely, though unproven, that medical intervention, which were typically used in patients with established diabetes, may improve the functional outcome of patients with ICH and IGR. We recommend the routine screening of IGR with standard OGTT in ICH patients and hope to get more information about the effectiveness of IGR intervention on ICH outcomes in future randomized controlled trials.
Several limitations must be acknowledged. First, though it was a nationwide cohort, the study sample of ICH was relatively small and confined to Chinese population. The applicability of our conclusion merits   further confirmation. Second, information on 1-year follow-up was available in 75.6% of patients with NGR or IGR in this study. However, baseline characteristics of patients with and without follow-up information were well-balanced. Third, it was unclear that IGR was ICH-induced or resulted from undiagnosed abnormal glucose regulation before ICH. However, the significant higher HbA1c level in IGR group than NGR group might indicate the abnormal glucose regulation existed before ICH onset. Finally, it was uncertain that IGR after ICH was persistent or transient. Whether the influence of IGR on outcomes could be altered by transformation of glucose metabolism status needed to be clarified in subsequent studies.
In conclusion, our study indicates that IGR was independently associated with 1-year poor outcome of ICH in Chinese patients, with more important influence on dependency than death.

Materials and Methods
Study Population. The study population was derived from the study of ACROSS-China, the baseline characteristics and recruiting criteria of which have been described in detail elsewhere 29 . In brief, ACROSS-China was a nationwide, multicenter and prospective cohort study, aimed at investigating the prevalence and influence of abnormal glucose regulation among patients hospitalized with acute stroke. Patients with ischemic stroke, ICH, or subarachnoid hemorrhage (SAH) within 14 days after onset were recruited consecutively from 35 participating hospitals across China from August 2008 to October 2009. A standard OGTT was performed in all the participants without previously known DM at the day 14 ± 3 after stroke onset or before discharge according to the Age (year), mean (SD) 57.   Table 4. Association between IGR and 1-year outcomes. * Patients with NGR was referent group. OR for the outcome of dependency and poor outcome and HR for the outcome of death. † Adjusted for gender, age, body mass index, and National Institute of Health Stroke Scale score on admission. ‡ Adjusted for gender, age, hematoma location, National Institute of Health Stroke Scale score on admission, and Glasgow Coma Scale score on admission. § Adjusted for gender, age, National Institute of Health Stroke Scale score on admission, and Glasgow Coma Scale score on admission. NGR indicates normal glucose regulation; IGR, impaired glucose regulation; OR, odds ratio; HR, hazard ratio and CI, confidence interval.
World Health Organization (WHO) criteria 33 . ACROSS-China was performed in accordance with the guidelines of the Helsinki Declaration. The protocol and data collection were approved by the Ethics Committees at Beijing Tiantan Hospital and all participating hospitals. Written informed consent was obtained from each participant or his/her designated relatives. Acute stroke was diagnosed according to WHO criteria 34 combined with brain CT or MRI confirmation. For the current study, patients with ischemic stroke and SAH were excluded. We also excluded patients with previously known DM, no OGTT performed, NDM based on OGTT results and patients lost to follow-up.

Data Collection and Variable Definition.
Demographic characteristics, clinical information, status on admission including stroke severity and blood pressure values, hematoma locations, laboratory values on admission, treatment during hospitalization and OGTT results were collected from the database. Stroke severity was measured using the initial NIHSS score and GCS score.
Outcome Measures. The outcomes were all-cause death (modified Rankin scale [mRS] score of 6), dependency (mRS score of 2 to 5) and poor outcome defined as mRS score of 2 to 6 at 1 year 36 . The 1-year follow-up of participants were conducted by trained research personnel at Beijing Tiantan Hospital. The mRS score at 1 year after onset of ICH were recorded through telephone interview. A death certificate from the local citizen registry or the attended hospital was used to confirm the case fatality.
Statistical Analysis. Baseline demographic and clinical characteristics were expressed as mean (standard deviation) or median (interquartile range) for continuous variables and as percentages for categorical variables. The chi-square test for categorical variables and Mann-Whitney test for continuous variables were used as needed. Cox proportion hazard model for death and logistic regression model for dependency and poor outcome were performed both in univariable and multivariable analyses to show the association between IGR and outcomes. HRs and ORs with 95%CI were calculated. Age, gender and all significant baseline variables in the univariable analysis were included in the multivariable analysis. A 2-sided P value < 0.05 was set as the level for statistical significance. All analyses were performed with SAS software version 9.4 (SAS Institute Inc, Cary, NC, USA).