Predicting the One-Year Prognosis and Mortality of Patients with Acute Ischemic Stroke Using Red Blood Cell Distribution Width Before Intravenous Thrombolysis.

PURPOSE: Red blood cell (RBC) distribution width (RDW) is known to reflect the heterogeneity of RBC volume, which may be associated with cardiovascular events or mortality after myocardial infarction. However, the association between RDW and stroke, especially regarding endpoints such as death, remains ambiguous. This study aimed to explore the prognostic value of RDW and its effect on mortality among patients with acute ischemic stroke (AIS) undergoing intravenous thrombolysis (IVT) after one year. PATIENTS AND METHODS: We retrospectively reviewed patients with AIS treated with IVT between January 2016 and March 2018. We grouped the patients according to modified ranking scale (MRS) scores as follows:0-2, favorable functional outcome group; and 3-6, unfavorable functional outcome. Predictors were determined using multivariate logistic regression (MVLR). The area under receiver-operating characteristic curve (AUC) was used to evaluate the predictive capability of variables. Furthermore, the Cox proportional hazard model was used to assess the contribution of risk factors to the outcome of death at one year later.
RESULTS: MVLR analysis showed that RDW (odds ratio [OR], 1.179; 95% confidence interval [CI], 0.900-1.545; p = 0.232) was not an independent predictor of unfavorable functional outcome, but it (OR 1.371; 95% CI 1.109-1.696; p = 0.004) was an independent biomarker for all-cause mortality. The optimal RDW cut-off value to predict mortality was 14.65% (sensitivity: 42%, specificity: 88.3%, AUC: 0.649, p < 0.001). Furthermore, higher RDW (hazard ratio, 2.860; 95% CI, 1.724-4.745; p < 0.001) indicated a greater risk of death.
CONCLUSION: The baseline RDW is a potential predictor of mortality in patients with AIS undergoing IVT, but RDW might not be associated with worse survival function among stroke survivors, which will help us to improve treatments and the management of patients with AIS.

Introduction

Red blood cell distribution width (RDW) is the most reliable index for differentiating between iron deficiency anemia and thalassemia trait.1 It has recently been shown to predict mortality in patients with severe sepsis or septics hock,2 acute kidneyinjury,3 acute heart failure, and acute myocardialinfarction4,5 as well as in critically ill and elderly patients.6,7 In fact, RDW is associated with an increased risk of perioperative stroke-related death in patients with valvular heart disease undergoing elective valvular surgery.8 A previous study has demonstrated a marked increase in the incidence of new-onset heart failure and subsequent death after discharge from hospital in patients with increased RDW.4 In addition, a recent report has hypothesized that RDW is a dynamic marker of risk in patients with sepsis based on the finding that an increase in RDW from baseline showed a correlation with prolonged hospital stay.2 Another epidemiologic cohort study has speculated that RDW may be a reliable marker of death not only in chronic conditions, but also in acute diseases.3

Increased RDW may comprehensively represent a variety of harmful biological mechanisms. Elevated RDW values have been shown to reflect malnutrition, liver failure, and renal dysfunction9 and to be a possible marker for accelerated red blood cell (RBC) destruction or ineffective RBC production.10 Emerging literature points to the potential association of RDW with inflammation.11 Abnormal increases in RDW can lead to carotid intima thickening, which accelerates atherosclerosis—a major risk factor for cerebral infarction.12,13

Stroke has been known to be the second leading cause of death and the third leading cause of disability worldwide,14 however, previous studies have failed to identify the prognostic role of RDW in patients with stroke, especially regarding endpoints such as death, remains ambiguous. In laboratories, RDW is a routinely used index of whole blood count. This parameter can be dynamically monitored with ease because the procedure involved in RDW determination is rapid and inexpensive. If the increased RDW can be used to detect major complications in time to strengthen comprehensive treatment and reduce mortality, the validation of RDW as an effective parameter for predicting outcomes of clinical conditions could have far-reaching implications in clinical diagnostics. The present study was designed to assess the prognostic role of RDW on 1-year mortality in patients with acute ischemic stroke (AIS).

Materials and Methods

Study Design and Population

We retrospectively analyzed the clinical and relevant imaging data of 570 patients with AIS who were treated consecutively with intravenous thrombolysis (IVT) in the First Affiliated Hospital of Wenzhou Medical University between January 2016 and March 2018. The diagnosis of AIS was performed by a neurologist specializing in stroke.

The main inclusion criteria were as follows: age, ≥18 years; onset-to-treatment time, <4.5 h; symptoms of neurological impairment caused by acute cerebral infarction; and informed consent signed by the patient or family member.

Exclusion criteria were as follows: thrombectomy after thrombolysis, active internal hemorrhage, intracranial tumor, cerebral aneurysm or recent cerebral hemorrhage, evidence of active bleeding or trauma (such as fracture) on physical examination, and major surgery in the past two weeks. In addition, cases with incomplete follow-up data and clinical data were excluded.

Finally, 480 patients were included in the current study after excluding 90 patients (20, beyond the therapeutic window; 13, recent cerebral hemorrhage; 6, intracranial tumor; 21, incomplete follow-up data; and 30, incomplete clinical data). The study was approved by the ethics committee of our institution.

The NIH Stroke Scale (NIHSS) score, which has a score range of 0–42, was used to assess the degree of neurological deficit in patients with stroke; the higher the score, the greater the severity of neurological damage. We recorded the patients’ NIHSS scores on admission and at discharge. Trial of ORG 10172 in acute stroke treatment classification was performed in all patients.

We grouped the patients according to the modified ranking scale (MRS) scores as follows: 0–2, favorable functional outcome; and 3–6, unfavorable functional outcome.

Patients underwent cranial baseline brain computed tomography (CT)/magnetic resonance (MR) imaging or cerebral vascular examinations (for example, CT angiography, CT perfusion, and MR angiography).

Primary Predictor Variable

Emergency blood samples were collected on admission using an ethylene diamine tetra acetic acid catheter. Within one hour after the sample was collected, the whole blood cell count was performed using an automatic hematology analyzer (Sysmex Company, XE-2100, Japan). We also documented existing comorbidities, such as hypertension, diabetes, hyperlipidemia, and heart disease, and blood pressure and blood sugar at the time of admission, which are risk factors for stroke. The expected value of RDW was 0.0% to 15.0%.

Follow-Up and Study Endpoints

In this study, the recovery of neurological function of all patients after one year was recorded. We classified the MRS scores of 0–2 as a favorable functional outcome and 3–6 as an unfavorable functional outcome; one-year all-cause mortality was the primary endpoint. The follow-updata were collected during outpatient visits or using standardized telephone questionnaires.

Statistical Analysis

Continuous variables that followed a normal distribution were expressed as means ± standard deviation (SD) and analyzed by one-way analysis of variance, while continuous variables with a non-normal distribution were represented as median (interquartile range [IQR]) and analyzed by Mann–Whitney U-test. Categorical variables are expressed as percentage numbers and analyzed by χ2 test or Fischer’s exact test. Variables that differed significantly with p values of <0.05 were selected as covariates for univariate and multivariate logistic regression analysis. Odds ratios (ORs) and their 95% confidence intervals were calculated. The optimal cut-off values were calculated based on the area under the receiver-operating characteristic (ROC) curve (AUC). The risk of a future clinical event in an individual patient was assessed by Cox multivariate proportional hazards regression analysis. All statistical analyses with p values of <0.05 were considered statistically significant. We conducted all statistical analyses using SPSS version 25.0 (SPSS Inc., Chicago, Illinois, USA).

Results

Clinical Characteristics

In this study, medical records of 480 patients were reviewed and included. The baseline features of all the included patients are shown in Table 1. Briefly, the median age of the studied patients was 71 (IQR, 16) years, and 37.5% of the patients were women. The median NIHSS score on admission was 7 (IQR, 10) and the mean RDW score was 13.76 (SD, 1.21). Furthermore, 159 subjects were smokers and 144 subjects consumed alcohol.

Table 1

Detailed Baseline and Clinical Features of the Patients

	Total (n = 480)	Favorable Functional Outcome (n = 238)	Unfavorable Functional outcome (n = 242)	p
Clinical
Age (median, IQR)	71(16)	75(15)	68(15)	<0.001
Females, n(%)	180(37.5)	104(43.7)	76(31.4)	0.005
Smoking history, n (%)	159(33.1)	63(26.5)	96(39.7)	0.002
Alcohol consumption, n (%)	141(29.4)	59(24.8)	82(33.9)	0.029
History of hypertension, n (%)	366(76.3)	192(80.7)	174(71.9)	0.024
History of diabetes, n (%)	134(27.9)	81(34.0)	53 (21.9)	0.003
History of dyslipidemia, n (%)	181(37.7)	96(40.3)	85(35.1)	0.239
Cardiac disease, n (%)	126(26.3)	67(28.2)	59(24.4)	0.348
Previous stroke or TIA, n (%)	80(16.7)	50(21.0)	30 (12.4)	0.011
Systolic blood pressure, (mean, SD)	152.19±25.67	152.18± 25.61	152.19 ±25.78	1.000
Diastolic blood pressure, (mean, SD)	84.66±15.23	84.54 ±15.69	84.77± 14.80	0.871
Glucose levels (mean, SD)	6.70±3.16	7.20 ±3.39	6.20± 2.84	<0.001
Laboratory data
Red blood cell (mean, SD)	4.48±0.58	4.39±0.64	4.57±0.49	<0.001
RDW (mean, SD)	13.76±1.21	13.95 ±1.37	13.57± 0.99	0.001
PLT (mean, SD)	196.40±58.02	191.04 ±57.78	201.68± 57.90	0.045
Leukocytes (mean, SD)	8.14±2.96	8.54 ±3.29	201.68 ±57.90	0.003
Clinical variables
NIHSS at admission (median, IQR)	7(10)	13(11)	4 (5)	<0.001
OTT time (min) (mean ± SD)	210.61±19.97	210.38 ±20.19	211.12 ±19.70	0.684
NIHSS at discharge (median, IQR)	5(9)	10(9)	2(3)	<0.001
TOAST, n (%)				0.052
Large-artery atherosclerosis, n (%)	270(56.3)	136(57.1)	134(55.4)
Small-vessel disease, n (%)	36(7.5)	10(4.2)	26(10.7)
Cardioembolic, n (%)	116(24.2)	62(26.1)	54 (22.3)
Other or unknown cause, n (%)	58(12.1)	30(12.6)	28 (11.6)
HT, n (%)	133(27.7)	90(37.8)	43(17.8)	<0.001

Abbreviations: RDW, red blood cell distribution width; PLT, platelet; OTT, onset to treatment time; NIHSS, The NIH Stroke Scale; IQR, Interquartile Rang; SD, Standard Deviation; TOAST, Trial of org 10172 in acute stroke treatment; HT, hemorrhagic transformation.

Association Between RDW Values and Clinical Prognosis

Among the 480 patients included, 242 (MRS,0–2) had a good prognosis and 238 (MRS,3–6) had a poor prognosis. Detailed baseline and clinical characteristics of the patients are summarized in Table 1. By comparison (favorable functional outcome group vs unfavorable functional outcome group), NIHSS score at admission (median: 13 [IQR, 11] vs 4 [IQR, 5]; p < 0.001), NIHSS score at discharge (median: 10 [IQR, 9] vs 2 [IQR, 3]; p < 0.001), age (median: 75 [IQR, 15] vs 68 [IQR, 15]; p < 0.001), RDW values (13.95 ± 1.37 vs 13.57 ± 0.99; p = 0.001), and sex (women: 43.7% vs 31.4%; p = 0.005) were significantly different between the two groups.

Then we screened out the factors that predicted the outcome of poor function, according to a logistic regression analysis of variables that differed significantly (p < 0.05), only three factors were screened out, which were NIHSS score at discharge (multivariate analysis OR, 1.400; 95% CI, 1.268–1.546; p < 0.001), age (multivariate analysis OR, 1.058; 95% CI, 1.028–1.088; p < 0.001) and previous stroke or transient ischemic attack (multivariate analysis OR, 2.533; 95% CI, 1.279–5.016; p = 0.008) (Table 2). RDW did not show a significant correlation with clinical outcome after one year (multivariate analysis OR, 1.179; 95% CI, 0.900–1.545; p = 0.232).

Table 2

The Logistic Regression Analyses of Predictors to Unfavorable Functional Outcome in 1 Year

	Univariate Analysis	P	Multivariate Analysis	P
	Odds Ratio(95% CI)		Odds Ratio(95% CI)
Age	1.058 (1.039–1.077)	<0.001	1.058 (1.028–1.088)	<0.001
Male gender	0.590 (0.406–0.857)	0.006
History of diabetes	1.840 (1.226–2.761)	0.003
History of hypertension	1.631 (1.065–2.499)	0.025
Smoking history	0.548 (0.372–0.806)	0.002
Alcohol consumption	0.643 (0.433–0.956)	0.029
Previous stroke or TIA	1.879 (1.147–3.078)	0.012	2.533 (1.279–5.016)	0.008
Glucose levels	1.115(1.046–1.189)	0.001
RDW	1.331(1.123–1.579)	0.001	1.179 (0.900–1.545)	0.232
Red blood cell	0.556(0.400–0.772)	<0.001
PLT	0.997 (0.994–1.000)	0.046
Leukocytes	1.098(1.030–1.170)	0.004
NIHSS at admission	1.259 (1.205–1.315)	<0.001
NIHSS at discharge	1.438 (1.344–1.538)	<0.001	1.400(1.268–1.546)	<0.001
HT	2.814 (1.847–4.288)	<0.001

Abbreviations: RDW, Red blood cell distribution width; PLT, Platelet; HT, Hemorrhagic transformation; TIA, transient ischemic attack; NIHSS, The NIH Stroke Scale.

Association Between RDW Values and Mortality

The primary endpoint (death) was observed in 55 patients. NIHSS scores at admission (p < 0.001) and discharge (p < 0.001), age (p < 0.001), diabetes (p = 0.026), bleeding transformation (p = 0.011), RDW (p < 0.001), and RBC (p = 0.001) were significantly different, as shown by the univariate analysis. Multivariate analysis showed that NIHSS score at discharge (OR, 1.211; 95% CI, 1.129–1.299; p < 0.001) and age (OR, 1.087; 95% CI, 1.048–1.128; p < 0.001) remained important prognostic factors of mortality. Besides, individuals with higher RDW (OR, 1.371; 95% CI, 1.109–1.696; p = 0.004) had a higher risk of all-causemortality. The final multivariable model is shown in Table 3.

Table 3

The Logistic Regression Analyses of Predictors to Mortality in 1 Year

	Univariate Analysis	P	Multivariate Analysis	P
	Odds Ratio(95% CI)		Odds Ratio(95%C)
Age	1.083 (1.051–1.116)	<0.001	1.087 (1.048–1.128)	<0.001
History of diabetes	1.826 (1.074–3.107)	0.026
Red blood cell	0.483 (0.310–0.754)	0.001
RDW	1.969 (1.418–2.736)	<0.001	1.371 (1.109–1.696)	0.004
NIHSS at admission	1.110 (1.074–1.146)	<0.001
NIHSS at discharge	1.155 (1.114–1.196)	<0.001	1.211 (1.129–1.299)	<0.001
HT	1.990 (1.172–3.379)	0.011

Abbreviations: HT, Hemorrhagic transformation; RDW, Red blood cell distribution width; NIHSS, The NIH Stroke Scale.

Age was highly significant in death assessment. For further analysis, we divided the patients into two groups depending on their age (high: age > 70; low: age ≤ 70) according to the median age. Then, we evaluated the relationship between RDW and death outcome in both the age-basedgroups. It is interesting to know that after adjusting various confounding variables, RDW value was significantly different in both the groups (age > 70 group: multivariate OR, 1.342; 95% CI, 1.008–1.788; p = 0.044; age ≤ 70 group: multivariate OR, 1.445; 95% CI, 1.034–2.021; p = 0.031).

Based on ROC analysis, the best cut-off RDW value that predicts mortality was 14.65% (AUC = 0.649; 95% CI, 0.569–0.730; p < 0.001). At this threshold, a sensitivity of 42.0% and specificity of 88.3% were observed (Table 4).

Table 4

The Baseline and Procedural Characteristics According to the Biomarker

	AUC(95% CI)	Sensitivity(%)	Specificity(%)	Youden Index(%)	p
Basic NIHSS + Age	0.790 (0.739–0.841)	92.8	53.5	46.3	<0.001
Basic NIHSS + Age+RDW	0.813 (0.767–0.859)	87.0	65.0	52.0	<0.001

Abbreviations: RDW, red blood cell distribution width; NIHSS, The NIH Stroke Scale.

We drew the ROC curve based on the traditional markers of risk (basic NIHSS score + age) and another RDW-richmodel (Figure 1). After adding new risk markers, there were higher significant correlations with unfavorable clinical outcome (AUC = 0.813; 95% CI, 0.767–0.859; p < 0.001).

Figure 1

Receiver-operating characteristic (ROC) curve displayed of multivariate model features The receiver-operating characteristic (ROC) curve based on the classic risk factors (NIHSS+Age) and multivariable model enriched with RDW. After adding RDW, there were higher significant correlations with mortality (AUC =0.813; 95% CI 0.767–0.859; p<0.001).

Abbreviations: RDW, red blood cell distribution width; NIHSS, The NIH Stroke Scale; AUC, area under the ROC curve.

Multivariate Cox regression proportional hazard model analyses were further performed after adjusting the possible confounding effect. As predicted, there was a positive correlation between RDW (HR, 2.860; 95% CI, 1.724–4.745; p < 0.001) and mortality risk (Table 5, Figure 2).

Table 5

The Cox Regression Model Analyses Assessment the Death Risk

	Univariate Cox Regression Model		Multivariate Cox Regression Model
	HR(95% CI)	P	HR(95% CI)	p
Basic NIHSS	1.091 (1.065–1.119)	<0.001	1.084 (1.056–1.114)	<0.001
RDW	4.294 (2.661–6.929)	<0.001	2.860 (1.724–4.745)	<0.001
Age	1.075 (1.047–1.104)	<0.001	1.050 (1.022–1.078)	<0.001

Abbreviations: RDW, red blood cell distribution width; NIHSS, The NIH Stroke Scale.

Figure 2

Death risk assessment using multivariate cox regression model analyses. Using cox survival curves after adjusting confounding effect. As shown in the figure, higher RDW indicated greater risk of death (HR 2.860; 95% CI 1.724–4.745; p <0.001).

Abbreviation: RDW, Red blood cell distribution width.

Discussion

The study suggests that RDW assessment at admission in emergency department (ED) may be an independent biomarker of death in patients with AIS who are treated with IVT.

A study involving 15,852 community-dwellingadults showed that higher RDW was strongly associated with the risk of all-cause mortality.15 Two previous studies have shown that an increase in RDW values may increase the incidence rate of a cardiovascular event in people with a coronary disease or heart failure.9,16 Furthermore, evidence suggests that RDW could be a novel prognostic biomarker in esophageal and hepatocellular cancer.17,18 In addition, two recent reports have demonstrated the effectiveness of RDW in predicting postoperative damage to the central nervoussystem.19,20 Moreover, other previous studies have reported that RDW may provide prognostic information for the functional outcome of stroke patients.21,22

The cause and mechanism underlying the association between RDW and mortality risk are not clearly known yet. Salvagno et al recently summarized the clinical significance of RDW in several biological and metabolic abnormalities, including shorter telomere lengths, increased erythrocyte fragmentation, and release of iron contained in the hemoglobin molecules.10 Felker et al provided a variety of novel insights into the relationship between RDW and outcome under conditions, such as nutritional deficiencies, renal dysfunction, hepatic congestion, and inflammatory stress.9 If the patient has a high RDW initially, parenteral nutrition should be actively provided, and the daily energy supply should be guaranteed after admission, which may be helpful in improving the one-yearsurvival rate. In addition, it was reported that oxidative stress might contribute to anisocytosis.23

A number of indicators, including anemia and RBC levels, have been proposed as strong predictors of cardiovascular disease and mortality in different populations.24–26 However, the association between RDW and death was found to be independent of baseline hemoglobin level and mean corpuscularvolume.27 In addition, in individuals with asymptomatic heart failure, the risk of all-cause mortality from the highest quartile of baseline RDW is nearly 80% higher than that of the lowest quartile, similar to C-reactive protein (CRP). Unlike the detection of new cardiovascular risk factors, RDW detection is cost-effective.16 RDW was observed to be a better prognostic marker than NT-pro-BNPin patients with pulmonary hypertension and right-sidedheart failure.27 In particular, RDW was found to be an effective and independent predictor of death in pulmonary embolism, which usually requires accurate and rapid identification and timely risk stratification under emergencyconditions.28 Therefore, in patients with high RDW before thrombolysis, other cardiac indicators may require special attention, and this information can be used to help actively treat the primary disease and identify the cause of stroke. Among the proposed functional outcome measures, higher neutrophil-lymphocyte ratio and platelet distribution width are known to be predictive of clinical outcomes and 3-month mortality in patients with AIS.29–31 These indicators, along with RDW values, can be easily measured by conventional whole blood analysis. Therefore, using a prognostic approach that combines these outcome measures to predict long- or short-term mortality in patients with AIS who are treated with IVT would be a more effective, convenient, and innovative alternative to existing clinical diagnostic methods.

Nonetheless, the exact physiological mechanism of the relationship between RDW and mortality from ischemic stroke after IVT is unclear. RDW has recently been related to adverse outcomes in patients with atherosclerosis.12,13 The RDW value is positively correlated with cholesterol content of the erythrocyte membrane,32 which increases the volume of the necrotic lipid core and leads to the rupture of atherosclerotic plaque.33 This information suggests that hyperlipidemia can be observed as early as possible before the vascular condition is evaluated in patients with high RDW, which may be beneficial to their prognosis. Inflammation is a critical issue in atherosclerosis, ischemia, and ischemic stroke.34,35 RDW increases the levels of plasma inflammatory biomarkers such as interferon γ and colony-forming unit erythroid cells to reduce endothelial nitric oxide production.35 Recent studies have shown that RDW levels are correlated with CRP levels.23,36,37 However, because CRP is not a routinely used index for diagnosis in emergencies, the relationship between RDW and CRP was not discussed. This also reminds us that the CRP level should be included in emergency routine evaluation, lung CT scan should be completed as soon as possible, patients’ state of inflammatory stress should be acutely observed, and anti-inflammatory treatment should be actively administered. Interestingly, in this study, there was no apparent difference in the level of leukocytes between the two RDW groups. This indicates that the mechanism of association between RDW and mortality is inflammation; to confirm this, further investigation is necessary. Meanwhile, recent studies reported the response of patients with AIS to IVT,38,39 higher RBC fraction group showed higher IVT responsiveness. RBC can affect clot stabilization and tPA-inducedfibrinolysis.40 However, there are very few studies on the effect of RDW on IVT reactivity and lysis of blood clots; thus, further experiments are needed to confirm these findings.

We acknowledge that our study has some limitations. First, because of differences in treatment procedures of discharged patients, unknown factors might have affected the results. Second, we did not discuss the relationship between RDW and age in depth, which limited the generalizability of the study’s results. Last, we did not explore the role of RDW in inflammation. In future experiments, we will address the above-mentioned limitations to obtain results that are more reliable.

Conclusions

Our study demonstrated that RDW before thrombolysis is an independent predictor of one-year mortality in patients with AIS, rather than a prognostic factor for the severity of stroke-based clinical outcomes. Thus, RDW before thrombolysis may be one of the future areas of development that will have potential implications on the stroke mechanism and therapeutic strategy after IVT for patients who are identified as having a high risk of mortality within the first year through RDW testing.