Intravenous thrombolysis versus early antiplatelet therapy in acute ischaemic stroke with small artery occlusion

Abstract

Background

The efficacy of intravenous thrombolysis (IVT) versus early antiplatelet therapy (APT) in small artery occlusion (SAO) stroke remains debated.

Methods

Ischaemic stroke (IS) patients with SAO who received IVT or early APT without IVT≤4.5 hours from stroke onset were screened from a prospective multicentre IS registry study from 1 January to 1 June 2021. The primary outcome was unfavourable functional outcome (FO) at 3 months. The secondary outcome was early neurological deterioration (END). The safety outcome was symptomatic intracerebral haemorrhage (sICH).

Results

There were 1125 SAO patients with 394 receiving IVT. 411 patients (36.5%) exhibited unfavourable FO, and sICH occurred in 3 cases (0.27%), all in IVT group, at the follow-up. END was observed in 213 patients (18.9%). After propensity score matching and multivariable adjustment, IVT significantly reduced the likelihood of unfavourable FO at 3 months (aOR 0.447, 95% CI 0.305 to 0.656), but no significant difference was found in END (aOR 0.867, 95% CI 0.569 to 1.321). Clustering analysis identified two distinct phenotypes: phenotype 0 (characterised by traditional cardiovascular risk factors) and phenotype 1 (marked by prominent inflammatory markers). A significant treatment-by-phenotype interaction was observed (p=0.002), with a comparable magnitude of benefit in phenotype 0 (aOR 0.405, 95% CI 0.244 to 0.673) compared with phenotype 1 (aOR 0.414, 95% CI 0.218 to 0.783).

Conclusion

IVT significantly reduced the likelihood of unfavourable FO at 3 months in SAO patients but did not significantly reduce END. Patients with traditional risk factors may benefit more from IVT than those with elevated inflammatory markers.

Trial registration number

ChiCTR2100045258.

WHAT IS ALREADY KNOWN ON THIS TOPIC

• The efficacy and safety of intravenous thrombolysis (IVT) on long-term functional outcome (FO) in stroke with small artery occlusion (SAO) remain controversial. Furthermore, early neurological deterioration (END) is common in SAO patients, the role of IVT in preventing END is not well established.

WHAT THIS STUDY ADDS

• This multicentre registry study provides real-world evidence that IVT significantly reduced the likelihood of unfavourable FO at 3 months versus early antiplatelet therapy in SAO patients. Patients with traditional risk factors at admission may benefit more from IVT compared with those with elevated inflammatory markers. However, IVT did not significantly reduce the END in SAO patients during hospitalisation.

HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY

• These results support the application of IVT in SAO patients to improve FO, highlighting the importance of close neurological monitoring for END following thrombolysis. Subsequent multicentre randomised controlled trials will be necessary to confirm these results and investigate phenotype-specific management approaches.

Introduction

Intravenous thrombolysis (IVT) has been recommended by clinical guidelines across the world as the preferred treatment for acute ischaemic stroke (IS) without specific recommendations on subtypes.^{1 2} The effect of IVT in patients with small artery occlusion (SAO) subtype, however, has been a topic of debate.³,⁵ Most prior studies either failed to clearly define SAO patients or lacked specialised analyses, resulting in limited data. Although some clinical studies suggested therapeutic benefits of IVT in SAO, these findings are constrained by methodological limitations, such as small sample sizes and retrospective designs.^{6 7} Additionally, in clinical practice, approximately 16.9–30.0% of SAO patients experience early neurological deterioration (END) during the acute onset period, which adversely affects their long-term functional outcome (FO).^{8 9} The potential of IVT to mitigate END in SAO patients remains a critical yet underexplored area. To address these concerns, this study seeks to investigate two clinical questions by using observational data in a prospective, multicentre, web-based IS registry: (1) whether IVT, compared with early antiplatelet therapy (APT) without IVT, improves FO at 3 months and (2) whether IVT is linked to a decreased END during the acute onset period.

Methods

Study population and patient selection

Patients were enrolled from a prospective, multicentre, web-based IS registry (Chinese Clinical Trial Registry, ChiCTR2100045258) between January and June 2021.¹⁰ Consecutive acute IS patients admitted to regional stroke centres ≤7 days of symptom onset were included. Comprehensive data, including demographics, medical history, prior therapies, in-hospital treatments and FO, were obtained via standardised case report forms as per the registry’s prospective protocol. Dedicated quality control personnel, appointed by the study lead, conducted monthly data integrity checks throughout the study period. Stroke aetiology was classified using the Trial of Org 10 172 in Acute Stroke Treatment (TOAST) criteria. FO was assessed at 3 months post stroke using the modified Rankin Scale (mRS). For patients unable to attend in-person follow-up visits, standardised telephone interviews were conducted to ensure complete data collection. Stroke classification, clinical evaluations and outcome assessments were performed exclusively by certified neurologists.

Eligible patients were selected from the registry according to the criteria: (1) ≥18 years of age; (2) SAO subtype according to TOAST criteria; (3) hospital arrival ≤4.5 hours from stoke onset. Exclusion criteria: (1) receipt of urokinase thrombolysis; (2) suffering from severe haematological disorders, hepatic or renal disease; (3) substantial missing data; and (4) loss to 3-month follow-up.

Data collection and outcome definition

Data, including demographic information (previous medical history, smoking, drinking, etc), clinical information (National Institutes of Health Stroke Scale (NIHSS) Score at admission, mRS post stroke at 3 months), laboratory test (routine blood, lipid, glucose, liver and kidney function, etc) and treatment information (before admission and during hospitalisation), were collected. Patients received either IVT or early APT without IVT, according to clinical treatment guidance and the physician’s discretion. The primary outcome was defined as unfavourable FO at 3 months (ie, the mRS of 2–6 points at 3 months).¹¹ The secondary outcome was END (ie, either an increase of ≥2 points in total NIHSS Score or an elevation of ≥1 point in the motor subscore within 7 days of admission).^{10 12} The safety outcome was symptomatic intracranial haemorrhage (sICH) based on the European Cooperative Acute Stroke Study 3 criteria.¹³

Statistical analysis

We presented continuous variables as either mean±SD or median (IQR), and analysed them via the t-test, Mann-Whitney U test or Kruskal-Wallis H test, according to their distribution. Categorical variables were characterised via frequencies (percentages) and compared via the χ² test. Univariate analyses were performed to identify baseline characteristics with significant imbalances between the IVT and early APT groups. Variables with p<0.05 were selected as covariates for the subsequent propensity score matching (PSM). We estimated propensity scores via a logistic regression model, where IVT served as the dependent variable and the previously identified unbalanced covariates functioned as independent variables. Patients were subsequently matched in a 1:1 ratio via the nearest-neighbour approach (calliper width: 0.1). Multivariable logistic regression models were constructed to evaluate associations between IVT and two outcomes: (1) unfavourable FO at 3 months and (2) END.

For the selection of variables influencing stroke outcome, the Light Gradient Boosting Machine (LightGBM) algorithm was employed due to its robust handling of missing values and ability to rank variable importance based on outcome impact. SHapley Additive exPlanations (SHAP) values were computed to quantify each variable’s contribution to the model’s predictions. The top 20 variables, ranked by SHAP values, were identified as the most predictive of FO. K-means clustering was performed within the matched cohort to explore phenotypic and clinical patterns. The optimal cluster count was identified using the Calinski-Harabasz (CH) Score and Silhouette Coefficient (SC). Subgroup analyses were conducted to assess outcome differences across the identified clusters based on their phenotypic and clinical profiles. Treatment-by-subgroup interaction was examined using the likelihood ratio test within logistic regression models. R (V.4.1.2) and Python V.3 were used to perform all statistical analyses and visualisations.

Results

Study population

In this study, 13 017 acute IS patients were consecutively enrolled from 18 participating hospitals. Of these, 4083 were classified as the SAO subtype. Among these, 2547 patients presented >4.5 hours from stroke onset, and 102 received treatment of urokinase. After excluding 84 patients with severe haematological, hepatic or renal diseases, 133 with incomplete data and 92 with no follow-up information, the final analysis comprises 1125 patients (as shown in figure 1).

Figure 1. Patient screening flow chart. APT, antiplatelet therapy; IS, ischaemic stroke; IVT, intravenous thrombolysis; PSM, propensity score matching; SAO, small artery occlusion.

Baseline characteristics before PSM

The mean age of the final analysis patients was 66.3±11.8 years and 461 (41.0%) were women. Further detailed information can be found in online supplemental table 1. Significant baseline imbalances between IVT and early APT groups were identified (online supplemental table 2). These variables included: drinking, hypertension, history of stroke/transient ischaemic attack, atrial fibrillation, coronary heart disease (CHD), diabetes mellitus, dyslipidaemia, mRS on admission, prior anticoagulant and antihypertensive use, NIHSS at admission, neutrophil count, lymphocyte count, haematocrit, etc. All variables above were included in the subsequent PSM analysis.

Outcome and risk factors analysis before PSM

At 3-month follow-up, 411 patients (36.5%) exhibited an unfavourable FO (mRS 2–6), and sICH was observed in 3 cases (0.27%), exclusively in the IVT subgroup. Within the first 7 days after admission, 213 patients (18.9%) experienced END. Univariate analysis for unfavourable FO at 3 months identified associations with higher proportions of drinking (p=0.035), more prior anticoagulation use (p<0.001), more occurrence of END (p=0.005), higher white blood cell (WBC) count (p<0.001), neutrophil count (p<0.001) and monocyte count (p<0.001), higher creatinine levels (p=0.012) and IVT (p<0.001). Univariate analysis identified factors associated with END, including older age (p=0.034), lower proportion of female (p=0.015), shorter time from onset to admission (p<0.001), less anticoagulation before admission (p<0.001), higher NIHSS score on admission (p<0.001), higher WBC count (p<0.001), haemoglobin (p=0.011), neutrophil count (p<0.001) and monocyte count (p=0.001), higher uric acid (p=0.001), triglycerides (p=0.017) and homocysteine level (p=0.048) (online supplemental table 1). Multivariate analysis results showed that IVT significantly reduced the likelihood of unfavourable FO at 3 months (adjusted OR (aOR) 0.485, 95% CI 0.362 to 0.649), while no significant difference was found for the occurrence of END in SAO patients (aOR 0.933, 95% CI 0.666 to 1.309).

Baseline characteristics after PSM

A total of 556 patients (IVT and early APT 1:1) were included in the subsequent analysis after PSM, which significantly reduced imbalances between the IVT and early APT groups (online supplemental table 2 and online supplemental figure 1). The mean age stood at 66.2±11.6 years, with 231 patients (41.5%) being women. Detailed information can be found in online supplemental table 2.

Outcome and risk factors analysis after PSM

Of the 556 patients, 188 (33.8%) had unfavourable FO at 3 months. END occurred in 117 patients (21.0%). Factors associated with unfavourable FO at 3 months included higher proportions of CHD (p=0.043), higher WBC count (p<0.001), neutrophil count (p=0.004), monocyte count (p<0.001) and IVT (p<0.001). In univariate analysis, factors associated with END occurrence included higher age (p=0.024), shorter time from onset to admission (p=0.042), higher NIHSS score on admission (p<0.001), higher WBC count (p=0.016) and neutrophil count (p=0.005) (online supplemental table 3). Multivariate analysis showed that IVT significantly reduced the likelihood of unfavourable FO at 3 months (aOR 0.447, 95% CI 0.305 to 0.656), with no significant difference in END (aOR 0.867, 95% CI 0.569 to 1.321) (table 1). The distribution of scores on the mRS at 3 months is shown in figure 2.

Table 1. Adjusted IVT associations with unfavourable FO at 3 months and END in unmatched and matched cohorts.

Unfavourable FO at 3 months	Unmatched cohort (n=1125)		Matched cohort (n=556)
	aOR (95% CI)	P value	aOR (95% CI)	P value
Early APT	Ref	–	Ref	–
IVT	0.485 (0.362 to 0.649)*	<0.001	0.447 (0.305 to 0.656)†	<0.001
END
Early APT	Ref	–	Ref	–
IVT	0.933 (0.666 to 1.309) ‡	0.689	0.867 (0.569 to 1.321) §	0.506
sICH	3 (0.27%)	–	1 (0.17%)	–

P value<0.05 was considered meaningful.

^*Adjusted for drinking, anticoagulant therapy, END, WBC, neutrophil count, monocyte count, creatinine.

^†Adjusted for CHD, WBC, neutrophil count, monocyte count.

^‡Adjusted for age, sex, onset to admission time, anticoagulant therapy, NIHSS Score on admission, WBC, haemoglobin, neutrophil count, monocyte count, uric acid, triglycerides, homocysteine.

^§Adjusted for age, onset to admission time, NIHSS Score on admission, WBC, neutrophil count.

aOR, adjusted OR; APT, antiplatelet therapy; CHD, coronary heart disease; END, early neurological deterioration; FO, functional outcome; IVT, intravenous thrombolysis; NIHSS, National Institutes of Health Stroke Scale; WBC, white blood cell.

Figure 2. mRS distribution at 3 months in small artery occlusion patients. APT, antiplatelet therapy; IVT, intravenous thrombolysis; mRS, modified Rankin Scale.

Clustering analysis

Within the matched cohort (n=556), the top 20 variables linked to FO at 3 months were identified using LightGBM, ranked by feature importance (figure 3). K-means clustering was performed on these variables, with an optimal cluster number of k=2 determined using the CH Score and SC (online supplemental figures 2A,B). Two phenotypes (0 and 1) were identified, with the top three phenotypic characteristics being WBC count, neutrophil count and monocyte count (online supplemental figures 2C,D). Compared with phenotype 0, phenotype 1 was characterised by higher WBC count, neutrophil count, monocyte count, higher haemoglobin, mean corpuscular volume, mean corpuscular haemoglobin (MCH) and MCH concentration, lower prevalence of diabetes, lower systolic blood pressure, blood glucose levels, total cholesterol, triglyceride levels and lower rates of IVT (online supplemental table 4). There was no significant difference in the occurrence of END (p=0.099) between phenotype 0 and phenotype 1, but phenotype 1 had a significantly larger proportion of unfavourable FO at 3 months than phenotype 0 (p=0.002).

Figure 3. Light Gradient Boosting Machine-based feature selection for clustering (A,B). NIHSS, National Institutes of Health Stroke Scale; SHAP, SHapley Additive exPlanations.

Subgroup analysis of IVT on FO at 3 months

Subgroup analysis for the primary outcome (figure 4) revealed significant treatment-by-subgroup interactions for age (>60 vs ≤60 years; p for interaction=0.005), hypertension (present vs absent; p for interaction=0.028) and phenotype (phenotype 0 vs phenotype 1; P-interaction=0.002). IVT showed a directionally greater magnitude of benefit in reducing unfavourable FO among older patients, those with hypertension and patients with phenotype 0 (characterised by traditional risk factors) compared with their respective counterparts.

Figure 4. Subgroup analysis of intravenous thrombolysis on functional outcome at 3 months in small artery occlusion patients. TIA, transient ischaemic attack.

Discussion

Our study demonstrated that IVT significantly reduced the likelihood of unfavourable FO at 3 months in SAO patients, whereas it did not significantly decrease the END. Patients presenting with traditional risk factors at admission benefit more from thrombolytic therapy compared to those with elevated inflammatory response indicators.

Alteplase is a well recognised and established therapy for acute IS when administered within the approved therapeutic window.¹⁴,¹⁶ However, its efficacy may vary across stroke subtypes due to distinct underlying pathophysiological mechanisms.¹⁷,²⁰ Evidence specifically addressing its effectiveness in the SAO subtype remains relatively limited and, at times, conflicting.^{11 21}

Our finding that IVT significantly improved long-term FO in SAO patients, with a low sICH rate (0.27%), aligns with several prior studies suggesting benefit.^{6 11} Eggers et al reported a 3-month prognostic advantage for thrombolysis in both lacunar strokes (LS) and non-LS patients using the Austrian Stroke Registry.¹⁸ A meta-analysis encompassing 23 studies (11 comparative studies and 3 randomised clinical trials) showed an increase in excellent outcomes (mRS 0–1) or favourable outcomes (mRS 0–2) with intravenous recombinant tissue plasminogen activator (rtPA) compared with the placebo group in SAOs.¹¹ Our findings reinforce these observations and provide further clinical evidence supporting IVT for acute SAO management. Recently, the European Stroke Organisation released guidelines focusing on cerebral small vessel disease that further corroborate that alteplase confers a beneficial effect in LS, consistent with broader clinical trial data across stroke severities and subtypes. Consequently, the Multidisciplinary Working Group has provided an evidence-based recommendation, supported by two expert consensus statements, advocating for the management of LS in accordance with current guidelines on thrombolytic use in IS.²²

Clustering analysis revealed two distinct clinical phenotypes: phenotype 0, characterised by traditional cardiovascular risk factors and phenotype 1, characterised by prominent inflammatory markers. A significant IVT-by-phenotype interaction was observed (p for interaction=0.002), with greater reduction in unfavourable FO in phenotype 0 vs phenotype 1. Systemic inflammation is known to correlate with stroke severity and infarct volume.^{23 24} Inflammatory cells, particularly neutrophils, can release neutrophil extracellular traps (NETs), which are DNA–protein complexes that form a mesh-like structure, actively contributing to thrombus stability and propagation.²⁵ Consequently, patients with heightened inflammatory responses may make clots more resistant to fibrinolysis and experience more challenging thrombus recanalisation, potentially leading to poorer outcomes. Conversely, lower inflammatory markers may indicate a more favourable response to thrombolysis.²⁶ These findings suggest that SAO patients with high inflammatory indices at admission may require tailored therapeutic strategies, phenotype 1 patients may benefit from adjunctive immunomodulation (eg, NETs-targeted therapies), warranting further investigation. Additionally, advanced age (>60 years) and hypertension were also associated with enhanced IVT response, though their underlying biological mechanisms remain unclear and warrant dedicated investigation. These findings collectively suggest SAO patients may benefit from phenotype-targeted IVT management strategies.

END remains a significant challenge in SAO patients, contributing to poor postdischarge outcome. However, evidence on the impact of thrombolytic therapy on END in SAO patients is limited and inconsistent. Marcelinus et al observed no significant effect of alteplase on preventing END in 132 patients with single small subcortical infarcts.²⁷ Similarly, Hwang et al observed that intravenous alteplase did not reduce the END in 76 acute LS patients.²⁸ Park et al observed no difference in NIHSS Score improvement ≤24 hours between thrombolysis-treated and thrombolysis-untreated patients across various stroke subtypes, and also found no significant prevention of END with thrombolysis compared with APT in 35 patients with cerebral infarction.²⁹ In contrast, some studies suggest that IVT may be beneficial in SAO patients with branch atheromatous disease.^{30 31} In our study, IVT did not reduce the occurrence of END in either prepropensity or postpropensity score-matched cohorts. This lack of effect may be attributed to the limited collateral circulation in perforating vessels and the short half-life of alteplase, which may reduce its ability to reach small vessel occlusion sites compared with large vessel pathology. The variability in prior studies may stem from small sample sizes and heterogeneous inclusion criteria, underscoring the need for larger, well-designed clinical trials to clarify the role of IVT in preventing END in SAO patients.

These findings have direct clinical implications. They support the use of IVT in eligible SAO patients to improve long-term FO, consistent with guideline recommendations. However, clinicians should not expect IVT alone to prevent END. Consequently, prompt evaluation of potential END risks and associated risk factors is essential following thrombolytic treatment. Clinicians should adopt tailored preventive and therapeutic strategies, such as the application of non-peptide antagonists targeting the platelet glycoprotein IIb/IIIa receptor, which have garnered growing attention in recent years.

Several limitations warrant consideration. First, as an observational study, residual confounding may persist despite the use of PSM and multivariable adjustment. Second, heterogeneity in APT regimens (monotherapy vs dual APT (DAPT)) was not controlled for in the outcome analyses. However, guideline-mandated avoidance of APT agents within 24 hours post-thrombolysis inherently limited DAPT use in IVT patients. Paradoxically, this differential exposure may strengthen our primary finding that IVT improves outcome, as higher DAPT utilisation in controls would have favoured their outcomes. Third, the lack of systematic capture of major systemic bleeding events limits comprehensive safety assessment. Fourth, the absence of key inflammatory biomarkers (eg, hypersensitive C-reactive protein (hs-CRP), interleukin 6) restricted mechanistic exploration of phenotype–treatment interactions. Finally, generalisability to other ethnicities requires validation in non-Asian populations.

Conclusion

IVT was associated with a significantly reduced likelihood of unfavourable FO at 3 months in SAO patients but did not significantly reduce the END during hospitalisation. These results support using IVT in SAO patients to improve FO, while highlighting the importance of close neurological monitoring for END following thrombolysis. Future multicentre randomised controlled trials are needed to validate these results and explore phenotype-targeted management strategies.

Data availability statement

Data are available upon reasonable request.