PDF icon Download PDF
Open Access
CC BY-NC-ND 4.0 · Thromb Haemost
DOI: 10.1055/a-2666-6277
New Technologies, Diagnostic Tools and Drugs

ANGWC Nomogram for Predicting Poor 3-month Outcome after Intravenous Thrombolysis in Young and Middle-aged Patients with First-ever Stroke

Shuang-Yin Lei   *
1   Department of Neurology, Stroke Center, the First Hospital of Jilin University, Changchun, China
,
Yu-Qian Yang   *
1   Department of Neurology, Stroke Center, the First Hospital of Jilin University, Changchun, China
,
Yang Qu
1   Department of Neurology, Stroke Center, the First Hospital of Jilin University, Changchun, China
,
Reziya Abuduxukuer
1   Department of Neurology, Stroke Center, the First Hospital of Jilin University, Changchun, China
,
Zi-Han Song
1   Department of Neurology, Stroke Center, the First Hospital of Jilin University, Changchun, China
,
Jia-Cheng Liu
1   Department of Neurology, Stroke Center, the First Hospital of Jilin University, Changchun, China
,
Jia-Xin Ren
1   Department of Neurology, Stroke Center, the First Hospital of Jilin University, Changchun, China
,
Peng Zhang
1   Department of Neurology, Stroke Center, the First Hospital of Jilin University, Changchun, China
,
Hang Jin
1   Department of Neurology, Stroke Center, the First Hospital of Jilin University, Changchun, China
,
Yi Yang 
1   Department of Neurology, Stroke Center, the First Hospital of Jilin University, Changchun, China
,
Zhen-Ni Guo
1   Department of Neurology, Stroke Center, the First Hospital of Jilin University, Changchun, China
2   Department of Neurology, Neuroscience Research Center, the First Hospital of Jilin University, Changchun, China
› Author Affiliations

Funding This work was supported by the National Natural Science Foundation of China (U24A20686), Science and Technology Department of Jilin Province (YDZJ202302CXJD061, 20220303002SF), and Jilin Provincial Key Laboratory (YDZJ202302CXJD017) to Y.Y., the Talent Reserve Program of the First Hospital of Jilin University (JDYYCB-2023002) to Z.G., and the Graduate Innovation Fund of Jilin University (2025CX292) to S.L.
› Further Information
 
 PDF Download Permissions and Reprints


Abstract

Background

The rising prevalence of ischemic stroke among younger individuals is concerning. Despite the significant benefits of intravenous thrombolytic therapy for young and middle-aged patients, the disease burden in this demographic is often overshadowed by that of older patient groups. This study aims to identify the risk factors for poor 3-month outcomes following intravenous thrombolysis in young and middle-aged patients with first-ever stroke and to develop a novel nomogram.

Methods

This prospective study included patients aged 18 to 59 years with a first ischemic stroke from 16 hospitals. Patients from one of the hospitals were designated as the training cohort. General information and clinical data were compared between the favorable outcome (modified Rankin Scale [mRS] score < 2) and the poor outcome groups (mRS score ≥ 2). Logistic regression analysis was used to develop the nomogram. An additional 113 patients from the other 15 hospitals were recruited as the validation cohort.

Results

A total of 217 patients were enrolled in the training cohort, with 47% experiencing poor 3-month outcomes. Five variables were selected to construct the ANGWC (Age, NIHSS score, blood Glucose, White blood cell count, and total Cholesterol) nomogram. The AUC of the nomogram was 0.802 in the training cohort and 0.747 in the validation cohort. The Hosmer–Lemeshow goodness-of-fit test showed a P-value > 0.05, and calibration curve slopes were close to 1. Decision curve analysis indicated good clinical utility.

Conclusion

The ANGWC nomogram provides a reliable prediction of poor 3-month outcomes after intravenous thrombolysis in young and middle-aged patients with first-ever stroke.


 

Keywords

ischemic stroke - intravenous thrombolysis - young and middle-aged - nomogram - predictive model

Introduction

Stroke ranks as the third leading cause of death worldwide, as reported in the latest Global Burden of Disease (GBD) Study 2021, with ischemic stroke accounting for 65.3% of all new stroke cases.[1] While the incidence of ischemic stroke is growing in the older population (a trend linked to the global aging demographic), there is also an alarming pattern of it increasingly affecting younger individuals. Notably, the prevalence and incidence of stroke among younger patients have increased significantly, even as the proportion of older patients with stroke in the overall incidence declines.[2] [3] Globally, over 2 million young individuals experience a stroke each year, and in the past decade, the incidence of stroke in those aged 18 to 50 years has surged by approximately 40%.[4] Young and middle-aged adults, often considered the backbone of social and economic development, face profound individual, familial, and societal consequences when confronted with life-threatening or disabling ischemic stroke. Surveys reveal that the hospitalization duration and costs for young and middle-aged patients with ischemic stroke (<65 years) exceed those of older adults (≥65 years), underscoring the substantial economic and medical burdens associated with this demographic.[5] Given these trends, optimizing diagnostic, therapeutic, and management strategies for ischemic stroke in young and middle-aged patients has become a critical focus in clinical practice.

Intravenous thrombolysis remains the preferred treatment for acute ischemic stroke, offering significant benefits to patients within the therapeutic time window (<4.5 hours). Evidence suggests that younger patients generally achieve greater benefits and better outcomes from thrombolysis compared with older patients, although they often face delays in assessment and treatment.[6] Despite advancements in the use of intravenous thrombolysis, dual antiplatelet therapy, and statins in this population, functional outcomes at 3 months and 1 year still fail to meet expectations.[7] The growing prevalence of traditional risk factors such as hypertension, diabetes, and dyslipidemia in younger populations, combined with inadequate self-management behaviors, exacerbates their vulnerability. Young and middle-aged patients with stroke are more likely to miss the optimal recovery period, increasing their risk of recurrent cardiovascular and cerebrovascular events, as well as mortality.[8] In light of these challenges, clinicians must predict outcomes following thrombolysis in young and middle-aged patients with stroke using readily available indicators. Such predictions can inform pre-thrombolysis decision-making and enable early post-thrombolysis interventions, including medication and psychological counseling.

Numerous models have been reported in the literature to predict poor outcomes following intravenous thrombolysis in patients with stroke.[9] [10] [11] Among these, nomograms have emerged as practical statistical tools for estimating the risk of specific outcomes in individual patients. Compared with auxiliary methods such as artificial neural networks, risk stratification, and decision tree analyses, nomograms offer superior predictive performance. Moreover, they enable busy clinicians to make evidence-based, personalized decisions more efficiently.[12] The START model, which includes factors such as baseline NIHSS score, age, pre-stroke modified Rankin Scale (mRS), and onset-to-treatment time,[13] and the STARTING-SICH model, which includes variables such as systolic blood pressure, age, onset-to-treatment time, NIHSS score, glucose levels, aspirin alone, aspirin plus clopidogrel, anticoagulant with international normalized ratio ≤1.7, current infarction sign, and hyperdense artery sign,[14] are well-known nomograms for predicting poor outcomes after thrombolysis. Although previous research has developed models with commendable stability and clinical applicability, limited attention has been paid to the unique needs of young and middle-aged individuals experiencing their first ischemic stroke. However, few predictive tools have been specifically developed for this demographic.

This study addresses this gap by exploring two key aspects in young and middle-aged patients with stroke (aged 18–59 years) following intravenous thrombolysis: (1) identifying easily accessible, clinically relevant independent predictors associated with poor 3-month functional outcomes and (2) developing a clinically applicable nomogram-based predictive model.


 

Methods

Study Population

This prospective cohort study included patients aged 18 to 59 years with a confirmed diagnosis of ischemic stroke who met specific inclusion criteria. Patients for the training cohort were enrolled at the First Hospital of Jilin University between September 2016 and April 2023, while those for the external validation cohort were recruited from 15 other hospitals in Northeast China between September 2021 and January 2022 ([Supplementary Table S1], available in the online version). The study received approval from the Ethics Committee of the First Hospital of Jilin University (Approval No. 2015–156). All participants provided written informed consent and retained the right to withdraw from the study at any time. The inclusion criteria were as follows: (1) Diagnosis of first-ever stroke confirmed by imaging methods such as computed tomography (CT) or magnetic resonance imaging (MRI). (2) Pre-stroke mRS score ≤1. (3) Administration of intravenous thrombolysis with 0.9 mg/kg recombinant tissue plasminogen activator (rt-PA).


 

Data Collection

Baseline characteristics collected included demographic information (age and sex), vascular risk factors (cigarette smoking, alcohol consumption, hypertension, diabetes mellitus, coronary heart disease, hyperhomocysteinemia, and dyslipidemia), thrombolysis-related parameters (baseline NIHSS score, onset-to-alteplase needle time, systolic blood pressure, and diastolic blood pressure), and concomitant treatments (antiplatelet therapy, anti-hypertension therapy, anti-diabetic therapy, and lipid-lowering therapy). Peripheral blood samples were obtained from all patients upon admission. The laboratory data collected included baseline blood glucose, glycated hemoglobin A1c, white blood cell count, neutrophil count, lymphocyte count, high-sensitivity C-reactive protein, triglycerides, total cholesterol, low-density lipoprotein cholesterol (LDL-C), high-density lipoprotein cholesterol (HDL-C), etc. Additionally, the analysis incorporated infarct location and Trial of Org 10172 in Acute Stroke Treatment (TOAST) classification.


 

Follow-up and Outcome Evaluation

After 3 months of intravenous thrombolysis treatment, patients were assessed by specialized researchers from participating hospitals. The mRS score was assessed by trained personnel through structured telephone interviews or review of outpatient records. Prognosis was categorized into six levels based on the mRS score: no symptoms, symptoms without significant disability, mild disability, moderate disability, severe disability, and death, with scores ranging from 0 to 6. For the purpose of this study, patients were divided into two groups: good outcome (mRS score 0–1) and poor outcome (mRS score 2–6).[15]


 

Statistical Analysis

All data analysis was conducted using IBM Statistical Package for the Social Sciences (SPSS) Statistics 26.0 software (Armonk, NY, USA). Continuous variables were first assessed for distribution using the Kolmogorov-Smirnov test. Normally distributed data are expressed as mean and standard deviation (SD), and after homogeneity of variance test, the independent samples t-test was used for comparison. Skewed data are reported as the median and interquartile range (IQR) and analyzed using the Mann–Whitney U test. Categorical variables were expressed as frequencies and percentages and compared using the Chi-square test or Fisher's exact test, as appropriate. Univariate logistic regression analysis was performed, and variables with a P-value of <0.1 were included in multivariate logistic regression to identify independent risk factors for poor prognosis. Variance inflation factors (VIF) were calculated to assess multicollinearity among variables, with a threshold of VIF > 5 indicating significant multicollinearity.

Model development involved variable selection using a backward stepwise approach guided by Akaike's information criterion (AIC). A visual nomogram was created (including only variables with P-values < 0.05) using Stata 16.0 (Stata Corporation, College Station, TX, USA). Model validation was conducted in three areas:

  1. Discriminative performance: Assessed using the area under the receiver operating characteristic curve (ROC-AUC).

  2. Calibration: Evaluated using the Hosmer–Lemeshow test and calibration curves to compare predicted versus actual outcomes.

  3. Clinical validity: Analyzed using decision curve analysis (DCA) to determine the range of threshold probabilities and the extent of clinical benefit.

Internal validation of model stability was performed using 10-fold cross-validation across all patients. The comparison method of prediction performance between models can be found in the [Supplementary Method] (available in the online version). Visual figures were prepared using GraphPad Prism 10.1.2 (GraphPad Software Inc., La Jolla, CA, USA). All statistical tests were two-sided, with P-values <0.05 considered statistically significant.


 

 

Results

Baseline Characteristics of the Patients

The study enrolled 330 patients, with 191 (57.9%) achieving a favorable 3-month functional outcome. The screening procedure is outlined in [Fig. 1], and the baseline clinical characteristics of the entire cohort are provided in [Supplementary Table S2] (available in the online version). The training and test cohorts included 217 (poor 3-month outcome: 47.0%; male: 78.3%) and 113 (poor 3-month outcome: 32.7%; male: 80.5%) patients, respectively. The comparison of baseline data between the training cohort and the validation cohort is summarized in [Supplementary Table S3] (available in the online version).

Zoom
Fig. 1 Flowchart of patient inclusion in this study. Patients who met the inclusion criteria were recruited from a single center, the First Hospital of Jilin University, to serve as the training cohort. Of these, 290 patients with ischemic stroke aged between 18 and 59 years who underwent intravenous thrombolytic therapy (IVT) were identified. Based on the study protocol, 73 patients were excluded, consisting 28 who had experienced recurrent stroke and 45 who did not have complete clinical data. In total, 217 patients were included in the nomogram. Based on 3-month modified Rankin Scale (mRS) scores, these patients were further categorized into a good prognosis group (mRS score <2, n = 115) and a poor prognosis group (mRS score ≥2, n = 102). Additionally, a validation cohort comprising 113 patients from 15 other centers was used to assess the model's performance.

In the training cohort, patients with poor outcomes were generally older (p = 0.007). They showed higher baseline NIHSS scores (p < 0.001), blood glucose (p < 0.001), glycated hemoglobin levels (p = 0.010), white blood cell count (p = 0.002), neutrophil count (p = 0.026), total cholesterol levels (p = 0.013), and HDL-C (p = 0.004). Additionally, these patients had a higher likelihood of having large-artery atherosclerosis (p = 0.018). No statistically significant differences were observed between the two groups for the remaining variables (p > 0.05, [Table 1]).

Table 1

Baseline characteristics of patients according to favorable outcome and poor outcome in the training cohort

Variables

Favorable outcome (n = 115)

Poor outcome (n = 102)

Test value

P-value

Demographic characteristics

Age, years (IQR)

52 (47–55)

53 (50–57)

−2.675

0.007

Male, n (%)

93 (80.9)

77 (75.5)

0.922

0.337

Vascular risk factors

Cigarette smoking, n (%)

78 (67.8)

62 (60.8)

1.171

0.279

Alcohol consumption, n (%)

55 (47.8)

53 (52.0)

0.370

0.543

Hypertension, n (%)

39 (33.9)

34 (33.3)

0.008

0.928

Diabetes mellitus, n (%)

6 (5.2)

10 (9.8)

1.665

0.197

Coronary heart disease, n (%)

22 (19.1)

20 (19.6)

0.008

0.929

Hhcy, n (%)

45 (39.1)

47 (46.1)

1.069

0.301

Dyslipidemia, n (%)

78 (67.8)

72 (70.6)

0.193

0.660

Thrombolysis-related data

Baseline NIHSS score (IQR)

6 (4–9)

9 (5–12)

−4.535

<0.001

ONT, min (IQR)

186 (147–237)

188 (146–237)

−0.142

0.887

SBP, mmHg (IQR)

151 (128–173)

157.5 (141–172)

−1.620

0.105

DBP, mmHg (IQR)

91 (80–105)

99 (88–105)

−1.703

0.089

Concomitant treatments

Anti-platelet therapy, n (%)

105 (91.3)

92 (90.2)

0.079

0.778

Anti-hypertension therapy, n (%)

33 (28.7)

38 (37.3)

1.799

0.180

Anti-diabetic therapy, n (%)

6 (5.2)

10 (9.8)

1.665

0.197

Lipid-lowering therapy, n (%)

112 (97.4)

98 (96.1)

0.026

0.872

Laboratory data

Blood glucose, mmol/L (IQR)

5.24 (4.70–5.96)

5.70 (5.15–6.99)

−3.527

<0.001

HbA1c, % (IQR)

5.5 (5.3–5.9)

5.7 (5.4–6.4)

−2.581

0.010

WBC, 109/L (SD)

7.59 (2.26)

8.61 (2.44)

−3.194

0.002

NE, 109/L (IQR)

4.71 (3.59–6.03)

5.12 (4.18–6.81)

−2.225

0.026

LY, 109/L (IQR)

2.02 (1.65–2.49)

1.97 (1.45–2.46)

−1.648

0.099

hs-CRP, mg/L (IQR)

3.16 (3.03–4.58)

3.23 (3.09–6.18)

−1.413

0.158

TG, mmol/L (IQR)

1.56 (1.07–2.27)

1.48 (1.22–2.19)

−0.344

0.731

TC, mmol/L (IQR)

4.84 (4.05–5.49)

5.10 (4.51–5.91)

−2.475

0.013

LDL-C, mmol/L (SD)

2.93 (0.75)

3.10 (0.77)

−1.653

0.100

HDL-C, mmol/L (IQR)

1.17 (1.04–1.34)

1.29 (1.09–1.53)

−2.851

0.004

AST, U/L (IQR)

22.3 (18.90–26.50)

22.85 (19.60–26.85)

−0.889

0.374

ALT, U/L (IQR)

18.5 (13.80–28.00)

20.15 (14.85–28.60)

−1.059

0.289

Hcy, μmol/L (IQR)

13.7 (10.5–19.9)

14.2 (10.9–25.7)

−1.369

0.171

UA, μmol/L (SD)

343 (82.37)

326 (80.96)

1.520

0.130

Lesion information

Infarct location

0.618

0.734

Cortex, n (%)

10 (8.70)

8 (7.84)

Subcortex, n (%)

65 (56.50)

63 (61.76)

Others, n (%)

40 (34.80)

31 (30.40)

TOAST classification

8.000

0.018

LAA, n (%)

24 (20.87)

39 (38.23)

SAO, n (%)

67 (58.26)

45 (44.12)

Others, n (%)

24 (20.87)

18 (17.65)

Abbreviations: ALT, alanine aminotransferase; AST, aspartate aminotransferase; DBP, diastolic blood pressure; HbA1c, glycated hemoglobin A1c; Hcy, homocysteine; HDL-C, high-density lipoprotein cholesterol; Hhcy, hyperhomocysteinemia; hs-CRP, high-sensitivity C-reactive protein; LAA, large artery atherosclerosis; LDL-C, low-density lipoprotein cholesterol; LY, lymphocyte count; NE, neutrophil count; NIHSS, National Institutes of Health Stroke Scale; ONT, onset-to-alteplase needle time; SAO, small artery occlusion; SBP, systolic blood pressure; TC, total cholesterol; TG, triglycerides; TOAST, the Trial of Org 10172 in Acute Stroke Treatment; UA, uric acid; WBC, white blood cell count.



 

Variable Analysis and Screening

Subsequent logistic regression analysis was performed on all the included variables in the training cohort. The univariate logistic regression analysis revealed that age, baseline NIHSS score, blood glucose, hemoglobin A1c, white blood cell count, neutrophil count, lymphocyte count, total cholesterol, HDL-C, and TOAST classification were associated with poor prognosis (p < 0.1, [Table 2]). These variables were then included in the multivariate logistic regression analysis. The backward stepwise selection process identified age (OR, 1.077; 95% CI, 1.025–1.131; p = 0.003), baseline NIHSS score (OR, 1.175; 95% CI, 1.081–1.277; p < 0.001), blood glucose (OR, 1.364; 95% CI, 1.103–1.688; p = 0.004), white blood cell count (OR, 1.204; 95% CI, 1.055–1.137; p = 0.006), and total cholesterol (OR, 1.545; 95% CI, 1.110–2.151; p = 0.01) as independent risk factors for poor 3-month outcomes following intravenous thrombolysis in young and middle-aged patients with first-ever ischemic stroke ([Fig. 2A]). The VIF values for all screened variables were less than 5, indicating no severe multicollinearity issues among the variables ([Fig. 2B]).

Table 2

Univariate and multivariate logistic analysis of poor 3-month outcome in the training cohort

Variables

Univariate logistic analysis

Multivariate logistic analysis

OR (95% CI)

P-value

OR (95% CI)

P-value

Age

1.066 (1.021–1.112)

0.003

1.077 (1.025–1.131)

0.003

Baseline NIHSS score

1.151 (1.069–1.240)

<0.001

1.175 (1.081–1.277)

<0.001

Blood glucose

1.376 (1.143–1.657)

0.001

1.364 (1.103–1.688)

0.004

HbA1c

1.618 (1.184–2.211)

0.003

WBC

1.205 (1.069–1.359)

0.002

1.204 (1.055–1.373)

0.006

NE

1.188 (1.039–1.358)

0.012

LY

0.689 (0.456–1.042)

0.078

TC

1.454 (1.114–1.898)

0.006

1.545 (1.110–2.151)

0.010

HDL-C

4.243 (1.650–10.912)

0.003

TOAST: SAO

0.413 (0.219–0.779)

0.006[a]

TOAST: Others

0.462 (0.208–1.022)

0.057[a]

Abbreviations: CI, confidence intervals; HbA1c, glycated hemoglobin A1c; HDL-C, high-density lipoprotein cholesterol; LY, lymphocyte count; NE, neutrophil count; NIHSS, National Institutes of Health Stroke Scale; OR, odds ratio; SAO, small artery occlusion; TC, total cholesterol; TOAST, the Trial of Org 10172 in Acute Stroke Treatment; WBC, white blood cell count.


a Compare with large artery atherosclerosis as a reference.


Zoom
Fig. 2 Logistic regression analysis and variance inflation factor (VIF) analysis for prognostic variables. (A) Multivariate logistic regression. Based on univariate logistic analysis, a stepwise multivariate logistic regression analysis was conducted to adjust for confounding factors, ultimately including age, baseline National Institutes of Health Stroke Scale (NIHSS) score, blood glucose, white blood cell count, and total cholesterol in the final model. (B) VIF analysis for the unadjusted (left) and adjusted (right) variables. CI, confidence interval; HbA1c, hemoglobin A1c; HDL-C, high-density lipoprotein cholesterol; OR, odds ratio; TOAST, Trial of Org 10,172 in Acute Stroke Treatment classification. *P-value < 0.05, **P-value < 0.01, ***P-value < 0.001.

 

Construction of an Individualized Nomogram

Based on the results of the logistic regression analysis described above, the ANGWC (Age, baseline NIHSS score, blood Glucose, White blood cell count, and total Cholesterol) nomogram was constructed ([Fig. 3]). The score can be calculated using individualized patient indicators, and the total score can subsequently be used to predict the risk of adverse outcomes following thrombolysis.

Zoom
Fig. 3 ANGWC nomogram for predicting poor 3-month outcomes after intravenous thrombolysis in young and middle-aged patients with first-ever stroke. The nomogram comprises five predictors, each with an initial score between 0 and 11. The total score corresponds to the predicted probability of the outcome and is calculated by summing the individual scores of the five predictors. NIHSS, National Institutes of Health Stroke Scale.

 

Evaluation and Validation of the ANGWC Nomogram

The performance of the ANGWC nomogram was evaluated from three key perspectives: (1) The receiver operating characteristic (ROC) curve was plotted, resulting in area under the curve (AUC) values of 0.802 (95% CI: 0.743–0.861, p < 0.001) for the training cohort and 0.747 (95% CI: 0.652–0.842, p < 0.001) for the validation cohort ([Fig. 4A, B]). The five independent risk factors included in the model were age (AUC = 0.605), NIHSS score (AUC = 0.678), glucose level (AUC = 0.639), white blood cell count (AUC = 0.630), and total cholesterol level (AUC = 0.597) ([Supplementary Fig. S1], available in the online version). These results indicate that the ANGWC nomogram can effectively predict adverse outcomes. (2) The Hosmer–Lemeshow test indicated excellent calibration in both the training cohort (p = 0.575, χ2 = 6.644) and the validation cohort (p = 0.692, χ2 = 5.598). The calibration curves closely approximated the ideal curve, confirming the model's good predictive accuracy ([Fig. 4C, D]). (3) Decision curve analysis (DCA) revealed that the net benefit was positive and significantly higher than that of the combination of age and NIHSS score when the high-risk threshold ranged between 0 and 0.8 in the training cohort. The DCA results from the validation cohort further supported the model's clinical utility in decision-making ([Fig. 4E, F]). Moreover, the 10-fold internal cross-validation of the ROC curve in the training cohort (ANGWC nomogram, AUC = 0.802; 10-fold cross-test, AUC = 0.786) confirmed the model's favorable generalizability ([Supplementary Fig. S1], available in the online version).

Zoom
Fig. 4 Validation of the ANGWC nomogram performance. Assessment was conducted separately for the training (left column, blue) and validation cohorts (right column, green). (A, B) Receiver operating characteristic (ROC) curves. Area under the curve (AUC) values for the two groups were 0.802 and 0.747, respectively, indicating a good model prediction accuracy. (C, D) Calibration curves. The slopes were close to 1, indicating a good match between the predicted probability (x-axis) and observed frequency (y-axis). (E, F) Decision curve analysis (DCA) curves. In both the training and validation cohorts, the model showed a higher net benefit value when the high-risk threshold ranged from 0 to 0.8. X-axis denotes the threshold probability, and the Y-axis denotes the net benefit. The gray line represents a scenario in which all patients were untreated or predicted negative, resulting in a net benefit of zero. The black line represents a situation in which all patients were treated with intravenous thrombolysis and predicted to experience poor 3-month functional outcomes. CI, confidence interval; NIHSS, National Institutes of Health Stroke Scale.

Sensitivity analysis was conducted in the validation cohort and included two aspects: (1) subgroup analysis based on the distribution of the 15 centers across cities revealed high discriminative performance in 4 cities (AUC = 0.786–0.924) and a good fit ([Supplementary Table S4], available in the online version); (2) further analysis was performed by dividing the 4 cities into two groups based on gross domestic product (GDP) ranking. Robust model performance was observed across all three performance metrics in both higher-resourced (7 centers, n = 46; AUC = 0.760) and lower-resourced (8 centers, n = 67; AUC = 0.797) cohorts; results are shown in [Supplementary Fig. S2] (available in the online version).

The discriminative performance of the ANGWC nomogram and the START model is further evaluated in [Table 3]. In the training cohort, the ANGWC model exhibited significantly superior predictive performance compared with the START model, with an AUC improvement of 0.099 (0.802 versus 0.703, p = 0.001), a net reclassification improvement (NRI) of 0.654 (p < 0.001), and an integrated discrimination improvement (IDI) of 0.167 (p < 0.001). In the validation cohort, the ANGWC model had a higher AUC value than the START model (0.747 versus 0.730, p = 0.691), although the difference was not statistically significant. However, it showed significant improvements in NRI (0.639, p = 0.001) and IDI (0.089, p = 0.001).

Table 3

Comparison of the ANGWC nomogram and START model in predicting poor 3-month outcomes

Training cohort

Validation cohort

START model

ANGWC model

P-value

START model

ANGWC model

P-value

AUC (95% CI)

0.703 (0.634–0.773)

0.802 (0.743–0.861)

0.001[a]

0.730 (0.632–0.828)

0.747 (0.652–0.842

0.691[a]

NRI (95% CI)

Reference

0.654 (0.416–0.891)

<0.001

Reference

0.639 (0.266–1.013)

0.001

IDI (95% CI)

Reference

0.167 (0.119–0.214)

<0.001

Reference

0.089 (0.036–0.142)

0.001

Abbreviations: AUC, area under the curve; CI, confidence intervals; IDI, integrated discrimination improvement; NRI, net reclassification improvement.


a Compared with ANGWC model by Delong test.



 

 

Discussion

To the best of our knowledge, this is the first clinical prediction model dedicated to assessing functional outcomes in patients with stroke following intravenous thrombolysis through an age-focused perspective. Using readily available clinical indicators, we identified age, NIHSS score, blood glucose level, white blood cell count, and total cholesterol level as independent risk factors for poor 3-month functional outcomes after thrombolysis.

In clinical practice, age and NIHSS score are among the primary data points gathered by emergency physicians and healthcare providers within the stroke care green channel. Advanced age and greater stroke severity frequently coexist, as older individuals often experience a higher burden of comorbidities and more compromised physical function.[16] The aging process inherently affects the primary cellular components of the neurovascular unit, leading to structural and functional anomalies. These anomalies include the early activation of neuronal apoptotic pathways, a shift of glial cells toward a neurotoxic phenotype, and reduced vascular endothelial cell competence in preserving blood–brain barrier stability.[17] These irreversible cytotoxic processes are significant intrinsic mechanisms contributing to the failure of reperfusion therapy, reflecting a more complex hypoxic–ischemic environment, larger infarct sizes, and more severe clinical manifestations (e.g., higher NIHSS scores). Although this study included only patients aged <60 years, the results reaffirm advanced age and NIHSS score as independent predictors of poor prognosis following thrombolysis, highlighting their continued prognostic relevance, even in younger patients. Although older age is associated with a greater risk of poor prognosis, it is crucial to recognize that age alone is no longer considered a contraindication for intravenous thrombolysis, as affirmed by international guidelines.[16] [18] From a different perspective, thrombolytic therapy should not be prematurely dismissed in young patients with mild strokes, particularly given the potential for ischemic stroke mimics. Adopting a personalized approach that thoroughly considers both indications and contraindications while striving to implement early therapeutic strategies to promote neural recovery remains a key area for further effort.

Beyond age and NIHSS scores, equal emphasis must be placed on laboratory test results. Understanding the roles, mechanisms, and dynamics of these parameters is crucial, as ischemic stroke not only affects the nervous system but also silently disrupts multiple organ systems.[19] Our analysis revealed that white blood cell count, glucose level, and total cholesterol level serve as independent risk factors predicting unfavorable 3-month outcomes post-thrombolysis. Elevated white blood cell count, typically indicative of heightened inflammatory responses and immune dysfunction, has been linked to poor outcomes in patients with stroke, whether measured before or after thrombolysis.[20] [21] [22] Following ischemic stroke, both circulating blood-borne white blood cells and resident brain-infiltrating cells rapidly activate, disrupt the blood–brain barrier, and migrate to the lesion site, releasing substantial amounts of neurotoxic substances.[23] Simultaneously, ischemic stroke triggers inflammatory responses in peripheral organs, such as the spleen, gastrointestinal tract, and adrenal glands, leading to complications that significantly affect prognosis.[19] Hyperglycemia is another well-established predictor of adverse outcomes following thrombolysis, including increased risks of intracranial hemorrhage, poor long-term prognosis, and mortality.[24] [25] Notably, hyperglycemia occurs in up to 40% of patients with stroke, regardless of prior diabetes history.[26] Pathologically, hyperglycemia is thought to promote lactate accumulation, disrupt energy metabolism in ischemic brain tissue, increase the permeability of the blood–brain barrier, and exacerbate reperfusion injury.[27] A recent multicenter cohort study of 4,181 patients with acute ischemic stroke demonstrated that the combination of hyperglycemia and elevated white blood cell count at admission is a strong predictor of poor outcomes following thrombolysis, surpassing the predictive value of either factor alone.[28] This finding aligns with our results, underscoring the synergistic contribution of inflammatory markers and blood glucose levels to the risk of unfavorable stroke outcomes.

The relationship between lipid metabolism disorders and stroke prognosis has been extensively studied, but the “cholesterol paradox” persists, with conflicting findings on cholesterol's role in stroke outcomes. Targeted research in patients undergoing vascular reperfusion remains limited. Some studies suggest that low levels of total cholesterol[29] and LDL-C[30] may protect against early adverse events, such as cerebral edema and symptomatic intracranial hemorrhage, yet their impact on long-term functional outcomes appears negligible. Previously, our N2H3 nomogram identified a high HDL-C/LDL-C ratio as an independent risk factor for poor 3-month outcomes following thrombolytic therapy in patients with stroke.[31] In line with recently published studies,[32] [33] our findings show that elevated total cholesterol levels are detrimental to 3-month functional outcomes, likely associated with less-than-ideal recanalization rates in large artery atherosclerosis subtype linked to high cholesterol levels. In the young and middle-aged population, elevated total cholesterol levels may indicate dysregulation of overall lipid metabolism in patients, while individual measurements of LDL-C and HDL-C may not sufficiently capture this process. Interestingly, a recent study has demonstrated that high cholesterol levels may lead to more severe consequences in the young individuals compared with the elderly.[34] Epidemiological studies among young adults indicate that elevated LDL-C and reduced HDL-C levels are not directly linked to stroke incidence, despite being established risk factors, while increasing total cholesterol levels are associated with a higher risk of adverse clinical events, including stroke, myocardial infarction, and death.[35] [36] Our findings further support the hypothesis that total cholesterol is the primary lipid component contributing to poor outcomes in younger and middle-aged stroke patients. Additionally, the cardiovascular risks associated with LDL-C and HDL-C may be more significant in the elderly population, with cumulative effects. Compensatory mechanisms and physiological reserves in young and middle-aged patients with stroke may cause laboratory parameters to remain within normal ranges during the early stages of the disease, masking underlying changes. This could be another factor contributing to discrepancies observed between this study and others. Consequently, dynamic monitoring of these parameters should not be underestimated. In our study, concomitant treatments (including anti-platelet, anti-hypertension, anti-diabetic, and lipid-lowering therapies) showed no significant inter-group differences, suggesting that these interventions did not introduce confounding effects. However, the study may have been underpowered to detect subtle effects of these treatments, and further research in larger cohorts is warranted to confirm these observations.

Recently, Yuan et al[37] designed a specialized nomogram to predict stroke recurrence risk in individuals aged 18 to 49 following ischemic stroke. Similarly, Mbarek et al[38] compared the performance of four distinct machine learning models in forecasting poor outcomes in patients aged 18 to 50 with acute ischemic stroke. However, these studies encompassed diverse treatment modalities without specifying whether patients underwent reperfusion therapy or other targeted interventions. Existing predictive tools for outcomes following intravenous thrombolysis rarely focus on young and middle-aged adult populations.[14] [31] [39] [40] Our study addresses this critical gap by pioneering a targeted approach. Previously, predictive tools such as DRAGON and ASTRAL scores have proven valuable for predicting poor 3-month outcomes after intravenous thrombolysis. Both the tools incorporated age and NIHSS score as categorical variables.[41] In contrast, our study modeled these variables as continuous parameters, reducing potential discriminatory errors inherent to categorical variables. Moreover, the model addresses a specific limitation of the SPAN index (calculated as the sum of age and NIHSS score with a cutoff ≥100),[42] enhancing its applicability to young patients and those with mild symptoms. Incorporating three additional laboratory parameters—white blood cell count, blood glucose, and total cholesterol—into the ANGWC nomogram underscores the importance of interpreting laboratory results comprehensively and considering systemic and metabolic interactions in clinical decision-making. Our model demonstrated excellent performances during the establishment and validation processes, showcasing strong discriminatory power, calibration, and a notable positive clinical net benefit. Considering the heterogeneity of different clinical settings in the real world (as evidenced by the differences in the baseline characteristics of the two cohorts and the DCA results), we conducted subgroup sensitivity analyses in the validation cohort, and still achieved positive results. Additionally, the ANGWC nomogram demonstrated superior prediction performance compared with the START model. These data demonstrate that we have developed a robust and efficient clinical prediction model for young and middle-aged patients with ischemic stroke. In the study, we fully recognize the potential differences brought by regional and hospital-level variations and have implemented corresponding measures to minimize bias. However, despite these efforts, due to the limited sample size of the 15 multi-center validation cohorts, potential biases may still exist.

Given the increasing prevalence of ischemic stroke among younger populations, it is imperative for both patients and clinicians to adopt proactive measures. Young and middle-aged patients should prioritize self-management practices, such as regular monitoring of blood glucose and lipid levels and the active prevention of cardiovascular and cerebrovascular diseases. In the context of the “youthification” of traditional risk factors, clinicians must exercise caution in managing young and middle-aged patients, ensuring long-term outcomes are not underestimated due to their age. Subgroup analyses and the development of predictive tools tailored to specific age groups are highly recommended. Recent advances in biomarkers derived from serum, imaging, and other fields, particularly cardiology, offer promising predictive benefits for optimizing acute ischemic stroke management. The development of innovative, non-invasive predictive biomarkers is strongly encouraged to enhance diagnostic precision and therapeutic strategies.


 

Limitations

When interpreting the results of this study, some limitations must be acknowledged. First, this study was conducted in Northeast China, potentially limiting the generalizability of its findings to other regions and populations. Previous studies have demonstrated significant ethnic and regional disparities in stroke outcomes, with genetic predisposition, environmental factors, and healthcare accessibility playing key roles in shaping these differences. Future research should explore the application value of the model across different settings and demographics. Second, there is no universally consistent age range for defining stroke in young and middle-aged adults, with cut-offs varying from 15 to 65 years.[43] [44] [45] To ensure standardization and enhance comprehension, our study adopted the World Health Organization (WHO) criteria, categorizing young and middle-aged patients as those aged 18 to 59 years. However, this definition may limit the applicability and performance of the model in populations outside this specific range. Third, additional potential factors influencing outcomes have not been adequately explored. Compared with older patients, younger individuals with ischemic stroke experience a broader spectrum of risk factors and face unique challenges, including increased stigma, reduced rehabilitation engagement, and significant financial toxicity, all of which may adversely affect their physical and mental recovery trajectories.[46] [47] Although our study indicates that routine standardized treatment during hospitalization does not affect 3-month outcomes, factors such as anxiety and depression scores, psychotherapy, and rehabilitation therapy were not assessed. Finally, no follow-up data beyond 1 year (including stroke recurrence, adverse cardiovascular events, and mortality) was collected; thus, the model's ability to predict long-term outcomes remains uncertain and warrants further investigation in future studies.


 

Conclusion

The ANGWC nomogram represents a novel and practical tool for predicting poor 3-month outcomes following intravenous thrombolysis in young and middle-aged patients with ischemic stroke. This simple risk prediction model shows significant potential for informing individualized clinical decision-making and improving early risk stratification; however, further comprehensive validation and calibration in diverse populations are essential to confirm its generalization capability and robustness.


 

What is known about this topic?

  • The incidence of ischemic stroke among younger individuals is surging, and the potential burden thereby imposed surpasses that of elderly patients.

  • The development of novel diagnostic and therapeutic strategies for stroke continues to advance dynamically across diverse academic domains.

  • However, there are no specific clinical prediction tools for young and middle-aged ischemic stroke patients who received intravenous thrombolysis.

What does this paper add?

  • This study developed the first nomogram for young and middle-aged stroke patients, integrating neurological severity, metabolic dysregulation, and subclinical inflammation to predict 3-month functional outcomes after intravenous thrombolysis.

  • By offering multicenter validation for clinical generalizability and decision curve analysis, the clinical value of ANGWC nomogram was confirmed in real-world treatment decisions.


 

Conflict of Interest

None declared.

Acknowledgment

We thank all the research members of the 16 centers who participated in the data collection and collation.

Data Availability Statement

Data will be made available from the corresponding author upon reasonable request.


Authors' Contribution

Z.G., Y.Y., H.J., and S.L. conceived this study. Y.Q. and R.A. collected data. J.R. and P.Z. provided research materials and methods. S.L. and Y.Q. analyzed the data. S.L., Y.-Q.Y., Z.S., and J.L. drafted the manuscript. All authors of the manuscript have reviewed and approved the final version of the manuscript.


* These authors contributed equally to this work.


Supplementary Material


Address for correspondence

Hang Jin, MD, PhD
Department of Neurology, Stroke Center, the First Hospital of Jilin University
Changchun
China   

Yi Yang, MD, PhD
Department of Neurology, Stroke Center, the First Hospital of Jilin University
Changchun
China   

Zhen-Ni Guo, MD, PhD
Department of Neurology, Stroke Center, the First Hospital of Jilin University
Changchun
China   

Publication History

Received: 07 February 2025

Accepted: 18 July 2025

Article published online:
11 August 2025

© 2025. The Author(s). This is an open access article published by Thieme under the terms of the Creative Commons Attribution-NonDerivative-NonCommercial License, permitting copying and reproduction so long as the original work is given appropriate credit. Contents may not be used for commercial purposes, or adapted, remixed, transformed or built upon. (https://creativecommons.org/licenses/by-nc-nd/4.0/)

Georg Thieme Verlag KG
Oswald-Hesse-Straße 50, 70469 Stuttgart, Germany


  Permissions and Reprints
Zoom
Fig. 1 Flowchart of patient inclusion in this study. Patients who met the inclusion criteria were recruited from a single center, the First Hospital of Jilin University, to serve as the training cohort. Of these, 290 patients with ischemic stroke aged between 18 and 59 years who underwent intravenous thrombolytic therapy (IVT) were identified. Based on the study protocol, 73 patients were excluded, consisting 28 who had experienced recurrent stroke and 45 who did not have complete clinical data. In total, 217 patients were included in the nomogram. Based on 3-month modified Rankin Scale (mRS) scores, these patients were further categorized into a good prognosis group (mRS score <2, n = 115) and a poor prognosis group (mRS score ≥2, n = 102). Additionally, a validation cohort comprising 113 patients from 15 other centers was used to assess the model's performance.
Zoom
Fig. 2 Logistic regression analysis and variance inflation factor (VIF) analysis for prognostic variables. (A) Multivariate logistic regression. Based on univariate logistic analysis, a stepwise multivariate logistic regression analysis was conducted to adjust for confounding factors, ultimately including age, baseline National Institutes of Health Stroke Scale (NIHSS) score, blood glucose, white blood cell count, and total cholesterol in the final model. (B) VIF analysis for the unadjusted (left) and adjusted (right) variables. CI, confidence interval; HbA1c, hemoglobin A1c; HDL-C, high-density lipoprotein cholesterol; OR, odds ratio; TOAST, Trial of Org 10,172 in Acute Stroke Treatment classification. *P-value < 0.05, **P-value < 0.01, ***P-value < 0.001.
Zoom
Fig. 3 ANGWC nomogram for predicting poor 3-month outcomes after intravenous thrombolysis in young and middle-aged patients with first-ever stroke. The nomogram comprises five predictors, each with an initial score between 0 and 11. The total score corresponds to the predicted probability of the outcome and is calculated by summing the individual scores of the five predictors. NIHSS, National Institutes of Health Stroke Scale.
Zoom
Fig. 4 Validation of the ANGWC nomogram performance. Assessment was conducted separately for the training (left column, blue) and validation cohorts (right column, green). (A, B) Receiver operating characteristic (ROC) curves. Area under the curve (AUC) values for the two groups were 0.802 and 0.747, respectively, indicating a good model prediction accuracy. (C, D) Calibration curves. The slopes were close to 1, indicating a good match between the predicted probability (x-axis) and observed frequency (y-axis). (E, F) Decision curve analysis (DCA) curves. In both the training and validation cohorts, the model showed a higher net benefit value when the high-risk threshold ranged from 0 to 0.8. X-axis denotes the threshold probability, and the Y-axis denotes the net benefit. The gray line represents a scenario in which all patients were untreated or predicted negative, resulting in a net benefit of zero. The black line represents a situation in which all patients were treated with intravenous thrombolysis and predicted to experience poor 3-month functional outcomes. CI, confidence interval; NIHSS, National Institutes of Health Stroke Scale.