Association of Glomerular Filtration Rate With Outcomes of Acute Stroke in Type 2 Diabetic Patients: Results From the China National Stroke Registry

Chronic kidney disease (CKD) is a worldwide public health problem with an estimated prevalence of 10.8% in Chinese adults (1–3). Growing evidence has demonstrated an association between impaired kidney function, defined as estimated glomerular filtration rate (eGFR) <60 mL/min/1.73 m², and all-cause mortality and other adverse cardiovascular outcomes in the general population (4–6). A recent meta-analysis showed that low eGFR was independently associated with incidence of stroke across a variety of participants, including Chinese stroke patients from Taiwan (7). However, the impact of low eGFR on stroke-related outcomes remains controversial (8,9). As the prevalence of stroke keeps increasing in China in recent years (10), it is very important to know the impact of reduced eGFR on adverse outcomes of stroke in Chinese patients.

Diabetes is another important public health problem in China, with an age-standardized prevalence of 9.7% (11). It is not only demonstrated as an independent risk factor for the incidence of stroke but is also proven to be a predictor for stroke outcomes among different ethnic groups (7,12). However, the effects of reduced eGFR on all-cause mortality or stroke recurrence and disability after acute stroke among diabetic patients have not been clarified (12,13). In this study, we systemically investigated whether a link exists between different levels of eGFR, calculated with the Chinese modification of Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) equation, and poor outcomes of acute stroke in a nationally representative cohort of acute stroke patients with type 2 diabetes.

The study population was from the China National Stroke Registry (CNSR) and substudy of CNSR (Abnormal Glucose Regulation in Patients with Acute Stroke Across China [ACROSS]). CNSR was a national, hospital-based, prospective cohort study designed to evaluate the quality of care for hospitalized stroke patients and measure the clinical and functional outcomes at 12 months after disease onset. The design of CNSR has been described previously (14). ACROSS was a prospective study aimed at investigating the prevalence and distribution of abnormal glucose regulation among hospitalized patients with ischemic and hemorrhagic stroke (15). Of 25,666 patients from CNSR from 2007 to 2008 (22,216) and ACROSS from 2008 to 2009 (3,450), 6,261 patient with cerebrovascular events and diabetes were enrolled in the current study. Patients (19,405; 75.6%) were excluded due to undetermined diagnoses, incomplete information at baseline, hospital transfers, and absence of diabetes (Supplementary Fig. 1). Written informed consent was signed by patients or their legally authorized representatives.

Trained research coordinators of CNSR and ACROSS at each institute collected baseline information, including patient demographics, vascular risk factors, stroke severity (National Institutes of Health Stroke Scale [NIHSS]), medication use, imaging studies, diagnosis, and complications. Vascular risk factors included history of stroke, hypertension, dyslipidemia, diabetes, atrial fibrillation, coronary heart disease, current or previous smoking, and moderate or heavy alcohol consumption. Hypertension was defined as systolic blood pressure ≥140 mmHg or diastolic blood pressure ≥90 mmHg, any use of antihypertensive drug, or self-reported history of hypertension. Diabetes was defined as fasting glucose level ≥7.0 mmol/L, nonfasting glucose concentration ≥11.1 mmol/L, any use of glucose-lowering drugs, or any self-reported history of diabetes. Dyslipidemia was defined as serum triglyceride ≥150 mg/dL, LDL cholesterol ≥140 mg/dL, HDL cholesterol ≤40 mg/dL, any use of lipid-lowering drugs, or any self-reported history of dyslipidemia. Atrial fibrillation was defined as history of atrial fibrillation confirmed by at least one electrocardiogram or presence of the arrhythmia during hospitalization. BMI was calculated by dividing measured weight in kilograms by the square of measured height in meters.

The serum creatinine level was measured on admission using the Jaffe method and standardized to isotype dilution mass spectrometry values according to each center’s protocol. The following is the CKD-EPI equation with adjusted coefficient of 1.1 for the Asian population (16,17): eGFR = 141 × min (SCr/κ,1)^α × max (SCr/κ,1)^−1.209 × 0.993^Age × 1.018 (if female), where SCr is serum creatinine, κ is 0.7 for females and 0.9 for males, α is −0.329 for females and −0.411 for males, min is the minimum of SCr/κ or 1, and max indicates the maximum of SCr/κ or 1.

Follow-up was performed by telephone interview. Data collection was performed by trained research personnel who were blinded to patient baseline clinical status. Patients were asked the standardized follow-up questions at 12 months after stroke onset. Outcome data collected included all-cause mortality, stroke recurrence, and stroke disability. Recurrent cerebrovascular events included ischemic stroke, intracranial hemorrhage (ICH), and subarachnoid hemorrhage (SAH). Recurrent stroke end points included fatal stroke and nonfatal stroke. We also evaluated the combined end point of death or recurrent stroke in the prognostic analysis. Stroke disability was divided into three categories based on a modified Rankin Scale (mRS): 0–1 (mild), 2–3 (moderate), 4–6 (severe).

We explored the association between eGFR distribution and outcomes so as to divide eGFR into clinically relevant groups spanning 5 mL/min/1.73 m²: <45, 45–49, 50–54, 55–59, 60–64, 65–69, 70–74, 75–79, 80–84, 85–89, 90–119, and ≥120 mL/min/1.73 m². Additionally, eGFR values were also categorized into five categories: <45, 45–59, 60–89, 90–119, and ≥120 mL/min/1.73 m², according to the new classification for CKD assessment and management but with minor modification due to the small sample of eGFR of <45 mL/min/1.73 m² (1).

We compared baseline and clinical characteristics of diabetic patients with cerebrovascular events according to eGFR levels. Continuous variables are expressed as means with SD or median with interquartile range (IQR), as appropriate. Categorical data are presented as proportions. The differences between eGFR categories were tested for continuous variables with normal distribution using one-way ANOVA and continuous variables with skewed distribution using Kruskal-Wallis test. The χ² or Fisher exact test was used for categorical variables.

We evaluated the association between eGFR and clinical outcomes among all patients, and stratified patients by different stroke subtypes using a logistic regression model (ordinal logistic regression for stroke disability) constructed with eGFR = 90–119 mL/min/1.73 m² as the reference group, adjusting for covariates of age, sex, history of stroke, hypertension, dyslipidemia, atrial fibrillation, coronary heart disease, diabetes, smoking, alcohol, BMI at admission, baseline NIHSS, lipid-lowering drug at discharge, antihypertensive drug at discharge, pneumonia, and urethral infection.

All analyses were conducted with SAS version 9.2 software (SAS Institute Inc., Cary, NC). Two-tailed P values <0.05 were considered to be statistically significant.

Of 6,261 stroke patients with diabetes, 4,836 were included in the final analysis after 1,425 were excluded due to lack of serum creatinine value on admission, missing 1-year follow-up information, or baseline age or sex. Patients excluded from the analysis were more likely to be older and to have a lower NIHSS score and higher BMI at baseline. They also had a higher prevalence of dyslipidemia and previous stroke and a lower proportion of infection and heavy alcohol consumption, as compared with those entered into the final analysis. Other factors such as sex, hypertension, and atrial fibrillation did not differ significantly between the two groups (Supplementary Fig. 1 and Supplementary Table 1).

Demographic and clinical characteristics of patients are shown in Table 1. Patients with reduced eGFR (<90 mL/min/1.73 m²) were older; were more likely to be female; had a higher NIHSS score; and were more likely to have a history of stroke, hypertension, atrial fibrillation, coronary heart disease, and infection than those with the normal or high eGFR (≥90 mL/min/1.73 m²). There was no significant statistical difference in dyslipidemia and BMI among eGFR categories. The median value for eGFR in all patients was 92.6 mL/min/1.73 m², and the distribution of eGFR values based on the Kernel density estimation is included in Supplementary Fig. 2.

Incidences of clinical outcomes across eGFR categories are shown in Table 2. In general, lower eGFR was associated with higher rates of adverse outcomes; however, adverse outcomes in the eGFR ≥120 mL/min/1.73 m² group were slightly higher than those in eGFR = 90–119 mL/min/1.73 m².

Adjusted odds ratios (ORs) for clinical outcomes according to eGFR categories among patients stratified by the stroke subtype are presented in Table 3. We found that low eGFR (<45 mL/min/1.73 m²) was independently associated with risks of all clinical outcomes, including all-cause death, recurrent stroke, combined end point (stroke or death), and stroke disability in stroke/transient ischemic attack (TIA) patients or patients with ischemic events, but not in patients with hemorrhagic stroke. In addition, high eGFR (≥120 mL/min/1.73 m²) was positively associated with increased risk of outcomes in all stroke subtypes, including hemorrhagic stroke.

In addition, the association between clinical outcomes and eGFR categories stratified by 5 mL/min/1.73 m² showed that low eGFR (<45 and 45–49 mL/min/1.73 m²) and high eGFR (≥120 mL/min/1.73 m²) were independently associated with the risk of stroke outcomes, except for stroke disability (45–50 mL/min/1.73 m²) (Fig. 1).

To our knowledge, this is the first nationwide investigation to examine the association between decreased eGFR and poststroke outcomes in type 2 diabetic patients in China. Our study demonstrated that reduced eGFR was independently associated with all-cause mortality and other poststroke outcomes in type 2 diabetic patients; stroke subtype analysis in our cohort showed that this association was only evident in ischemic stroke and TIA (Table 3). We also observed a U-shaped relationship between variation of eGFR and poststroke outcomes, that is, increased ORs were observed among those with low and high levels of eGFR. The cutoff points of eGFR associated with poor outcomes of stroke were eGFR <45 and ≥120 mL/min/1.73 m², respectively (Table 3 and Fig. 1).

Data on the effect of reduced eGFR on poststroke outcomes have controversies. Some studies suggested that reduced eGFR is associated with a poor outcome in ischemic stroke patients (18,19), whereas other studies do not (20,21). Mittleman and colleagues (19) recently demonstrated that a reduced or highly elevated eGFR (<60 or >125 mL/min/1.73 m²) was associated with a higher mortality rate compared with patients with moderate levels of eGFR in 1,175 acute ischemic stroke patients. In the current study, we demonstrated the detrimental effects of decreased or highly increased eGFR (<45 or ≥120 mL/min/1.73 m²) on adverse outcomes of stroke in a much larger diabetic population. Mechanisms by which reduced or elevated eGFR increases stroke outcomes are still under investigation. The increased risk of adverse poststroke outcomes with high or low eGFR might be partly explained by factors associated with decline of renal function, like anemia, oxidative stress, and inflammation, which could accelerate atherosclerosis and endothelial dysfunction (22). Our previous work has also indicated diabetes to be an independent predictor for adverse outcomes of acute stroke (23). The combination of diabetes and CKD might increase the risk of adverse stroke outcomes further.

By applying the Chinese modification of the CKD-EPI equation, which had been validated to be less biased at a higher eGFR value than the Modification of Diet in Renal Disease study equation (24), we obtained new cutoff points associated with poor outcomes in our cohort. These two cutoff points formed a U-shape relationship between eGFR and adverse stroke outcomes in our diabetic patients (Fig. 1). The lower cutoff point in our study (eGFR <45 mL/min/1.73 m²) coincides with the new classification of stage 3b CKD (eGFR range from 30 to ∼44 mL/min/1.73 m²), which has more serious adverse outcomes than stage 3a CKD (eGFR range from 45 to ∼59 mL/min/1.73 m²) (1). The other cutoff point in our analysis is eGFR ≥120 mL/min/1.73 m², which is also linked with poor poststroke outcomes (Table 3). In corroboration with our findings, Shlipak et al. (25) showed that the association between quintiles of creatinine and all-cause mortality appeared to be J shaped among 4,637 participants in the Cardiovascular Health Study. Given these findings, eGFR ≥120 mL/min/1.73 m² may not reflect proper kidney function, which is often falsely elevated in patients with grossly abnormal muscle mass like amputation, paralysis, and muscular disease and even in the early stage of diabetic nephropathy (26).

The clinical implications of our findings are important as they provide evidence that prevention and management of CKD is a very important measure for type 2 diabetic patients with ischemic stroke or TIA, especially among those whose eGFR is <45 or ≥120 mL/min/1.73 m².

Our study, however, still has some limitations. First, because our cohort was comprised of Chinese adult stroke patients with a relatively short follow-up period, the results may not be generalized to other ethnic stroke populations. Second, since patients with missing baseline serum creatinine or lost follow-up within 1 year were not included in the study, a selection bias is possible, such as discrepancy in sex and dyslipidemia (Supplementary Table 1). Third, we only adjusted for the control of risk factors at discharge but not during the follow-up period in multivariate analysis of stroke outcomes. Last, albuminuria has proved to be a very important independent risk factor for many adverse clinical outcomes and was not included in our study; thus, we were unable to evaluate eGFR and albuminuria simultaneously along with other factors in our patients.

In conclusion, our data demonstrate that eGFR is an independent predictor of all-cause mortality and other poor outcomes after acute stroke. The cutoff points of eGFR associated with adverse stroke outcomes in this group of patients are <45 and ≥120 mL/min/1.73 m², respectively. These findings emphasize the importance of eGFR evaluation in clinical practice in type 2 diabetic patients with acute stroke.

The authors thank all participating hospitals, colleagues, nurses, and imaging and laboratory technicians.

Funding. This study was supported by a grant from the Ministry of Science and Technology of the People’s Republic of China (2008ZX09312-008, 2011BAI08B02, 2012ZX09303, and 200902004).

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. Y.L. interpreted analysis of the data and prepared the manuscript. X.W. interpreted analysis of the data, prepared the manuscript, and conducted the statistical analysis. Yi.W. conceived, designed, and coordinated the study; interpreted analysis of the data; prepared the manuscript; oversaw subject recruitment; gathered clinical data; and performed follow-up of patients. C.W. conceived, designed, and coordinated the study. H.W. and D.W. contributed to comments on the draft manuscript and revised the manuscript. L.L. oversaw subject recruitment. Q.J. gathered clinical data and performed follow-up of patients. G.L. conducted the statistical analysis. X.Z. and Yo.W. conceived, designed, and coordinated the study; contributed to comments on the draft manuscript; and revised the manuscript. Yo.W. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.