Thrombus Perviousness Is Associated With Differential Treatment Outcomes Following IVT and EVT in MeVO Stroke

Abstract

BACKGROUND:

In acute ischemic stroke due to medium vessel occlusion, the relative effectiveness of intravenous thrombolysis (IVT) and endovascular therapy (EVT) remains controversial, and treatment benefits may vary across patients. We evaluated the interaction between thrombus perviousness and treatment modality on 90-day functional outcome in patients with medium vessel occlusion.

METHODS:

This single-center retrospective cohort study analyzed 255 patients with imaging-confirmed medium vessel occlusion treated at a stroke center in China between 2018 and 2024 (154 IVT and 101 EVT with or without prior IVT). Thrombus perviousness was quantified using the thrombus perviousness index (TPI) from noncontrast computed tomography and computed tomography angiography. Multivariable logistic regression and inverse probability of treatment weighting evaluated the TPI-by-treatment interaction for 90-day functional independence (modified Rankin Scale score ≤2). Sensitivity analyses used alternative weighting strategies and a surrogate thrombus perviousness measure.

RESULTS:

The 90-day functional independence rate was similar between the IVT and EVT groups (51.9% versus 51.5%; P=0.96), and TPI was not independently associated with outcome. In the inverse probability of treatment weighting-weighted model, a significant interaction between TPI and treatment modality was observed (interaction odds ratio, 0.37 [95% CI, 0.24–0.56]; P<0.001). Higher TPI was associated with a greater likelihood of functional independence in IVT-treated patients (odds ratio per 0.05-unit increase, 1.73 [95% CI, 1.19–2.53]; P=0.005), but lower odds in EVT-treated patients (odds ratio, 0.65 [95% CI, 0.49–0.87]; P=0.003). Exploratory analyses in the IVT subgroup revealed a continuous increase in functional independence as TPI increased, without a stable cutoff. Rates of symptomatic intracranial hemorrhage and 90-day mortality did not differ between groups, and TPI was not associated with hemorrhage risk.

CONCLUSIONS:

In medium vessel occlusion, thrombus perviousness was associated with treatment outcomes, with higher perviousness favoring IVT-related benefits, whereas the use of EVT may be relatively advantageous in patients with low-perviousness thrombi.

Medium vessel occlusion (MeVO) accounts for ≈25% to 40% of acute ischemic strokes and has distinct clinical and imaging characteristics.¹ Compared with classic large vessel occlusion (LVO), MeVO affects smaller-caliber arteries and exhibits greater anatomic variability, with more heterogeneous collateral circulation patterns.² MeVO encompasses cardioembolic, artery-to-artery, and small vessel disease-related mechanisms, reflecting substantial heterogeneity and complicating reperfusion decision-making.³

Currently, intravenous thrombolysis (IVT) and endovascular therapy (EVT) are the major reperfusion options for MeVO; however, previous comparative studies have reported inconsistent and sometimes conflicting results. Some studies have suggested that EVT may achieve higher rates of angiographic recanalization, without a clear advantage in 90-day functional independence.⁴ More recently, 2 randomized controlled trials published in early 2025—ESCAPE-MeVO (Endovascular Treatment for Medium Vessel Occlusion)⁵ and DISTAL (Distal Medium Vessel Occlusion Trial)⁶—demonstrated that EVT combined with best medical therapy did not significantly improve 90-day functional independence or reduce mortality and was associated with higher rates of symptomatic intracranial hemorrhage. These results suggest substantial heterogeneity in the treatment response to EVT within this population. Therefore, occlusion location or baseline stroke severity alone may not fully account for interindividual differences in treatment response, highlighting the need to explore additional imaging or biological markers to support more individualized reperfusion strategies in MeVO.

Thrombus-related characteristics have attracted attention as potential determinants of reperfusion effectiveness. Basic and clinical studies have shown that thrombus composition is closely linked to imaging-derived measures of perviousness and may influence individual responses to reperfusion therapies.^7–9 In recent years, noncontrast computed tomography (NCCT)- and computed tomographic angiography (CTA)-derived thrombus attenuation characteristics have been shown to reflect thrombus microstructure noninvasively and rapidly in the acute setting.¹⁰ Thrombus perviousness is commonly quantified as the difference in attenuation between CTA and NCCT, and it is often normalized using contralateral vessel measurements. This imaging metric has been shown to correlate with histopathologic residual flow, contrast penetration within the thrombus, and early recanalization after IVT.¹¹ However, most of this evidence has been derived from patients with LVO, and its applicability in MeVO remains insufficiently validated. Moreover, previous studies have largely examined thrombus perviousness within a single reperfusion strategy, leaving potential differential effects across treatment modalities largely unexplored.

Mechanistically, thrombus perviousness may influence intrathrombus thrombolytic distribution and fragmentation during mechanical manipulation, thereby affecting distal microvascular perfusion. Thus, thrombus perviousness may be differentially associated with outcomes across distinct reperfusion strategies rather than serving solely as a prognostic marker within a single treatment context; however, this has not been systematically evaluated across IVT and EVT in MeVO.

Therefore, this study aimed to determine whether thrombus perviousness is associated with differential treatment-outcome patterns across IVT and EVT in MeVO rather than functioning solely as an independent predictor. Using routinely acquired emergency NCCT and CTA, the thrombus perviousness index (TPI) was quantified and incorporated as a continuous variable into interaction models to evaluate its association with functional outcomes under distinct reperfusion strategies, thereby exploring its potential role in treatment stratification for acute ischemic stroke due to MeVO.

Methods

Data Availability Statement

Deidentified data are not publicly available but may be obtained from the corresponding author upon reasonable request and institutional approval.

Ethics Statement

This study was approved by the Ethics Committee of the First Hospital of Jilin University (Approval No. 23K177-002). The requirement for informed consent was waived by the committee owing to the retrospective nature of the study and the use of anonymized clinical data.

Patients and Study Design

This single-center retrospective cohort study included consecutive patients with acute ischemic stroke who presented to the Stroke Center of the First Hospital of Jilin University between January 2018 and December 2024 with imaging-confirmed MeVO. This study was reported in accordance with the Strengthening the Reporting of Observational Studies in Epidemiology reporting guideline for cohort studies. All patients underwent NCCT and CTA, and the intracranial culprit vessel occlusion was independently confirmed by experienced neuroradiologists.

The inclusion criteria were: (1) onset-to-admission time ≤24 hours; (2) availability of baseline NCCT and CTA with sufficient image quality for analysis; (3) CTA-confirmed intracranial culprit vessel occlusion; and (4) imaging features consistent with the definition of MeVO, as previously described.¹² The exclusion criteria were: (1) LVO involving the intracranial internal carotid artery, middle cerebral artery M1 segment, or basilar artery trunk; (2) inability to localize the thrombus on CTA; (3) missing 90-day modified Rankin Scale (mRS) assessment for functional independence; and (4) missing key baseline clinical variables, such as the baseline National Institutes of Health Stroke Scale (NIHSS) score or onset-to-treatment time (OTT).

A total of 255 patients met all eligibility criteria and were included in the final analysis (Figure 1).

Figure 1.

Flowchart of patient inclusion and exclusion. BA indicates basilar artery; CTA, computed tomographic angiography; EVT, endovascular therapy; ICA, internal carotid artery; IVT, intravenous thrombolysis; LVO, large vessel occlusion; M1, M1 segment of the middle cerebral artery; MeVO, medium vessel occlusion; mRS90, 90-day modified Rankin Scale; and NCCT, noncontrast computed tomography.

Imaging Acquisition and Registration

Imaging data were processed using a standardized workflow involving NCCT intensity correction and CTA-to-NCCT registration, as detailed in the Supplemental Methods.^13,14

Thrombus Segmentation and Quantification of Perviousness

The thrombus location was identified on CTA at the site of vessel occlusion, and manual thrombus segmentation was independently performed by 2 blinded readers on multiplanar reconstructed images. The segmented thrombus regions were subsequently mapped to the registered NCCT images for attenuation measurement, and discrepancies were resolved by consensus.

Thrombus attenuation was measured using predefined regions of interest (ROIs), and attenuation of the corresponding contralateral nonoccluded vessel was used for normalization. To enhance measurement robustness, attenuation values were sampled at 3 representative levels (proximal, middle, and distal) along the thrombus axis using small circular ROIs (1 mm²), with values averaged for analysis. Care was taken to avoid vessel walls, calcifications, and obvious imaging artifacts during ROI placement. The measurement workflow is illustrated in Figure 2.

Figure 2.

Measurement of thrombus attenuation on noncontrast computed tomography (NCCT) and computed tomographic angiography (CTA). A, Thrombus region of interest (ROI) placement on NCCT. B, Corresponding thrombus ROI on CTA. C, Contralateral reference ROI placement on NCCT in the anatomically matched, nonoccluded artery. D, Corresponding contralateral reference ROI on CTA. Thrombus perviousness was quantified using the thrombus perviousness index, defined as the CTA-NCCT attenuation difference (in Hounsfield units) within the thrombus, normalized by the corresponding CTA-NCCT attenuation difference measured in the contralateral reference artery.

Thrombus perviousness was quantified using the TPI, defined as follows, with all attenuation values expressed in Hounsfield units¹⁵:

$${\displaystyle \mathrm{T}\mathrm{P}\mathrm{I}=\frac{\mathrm{C}\mathrm{T}{\mathrm{A}}_{\mathrm{t}\mathrm{h}\mathrm{r}\mathrm{o}\mathrm{m}\mathrm{b}\mathrm{u}\mathrm{s}}-\mathrm{N}\mathrm{C}\mathrm{C}{\mathrm{T}}_{\mathrm{t}\mathrm{h}\mathrm{r}\mathrm{o}\mathrm{m}\mathrm{b}\mathrm{u}\mathrm{s}}}{\mathrm{C}\mathrm{T}{\mathrm{A}}_{\mathrm{c}\mathrm{o}\mathrm{n}\mathrm{t}\mathrm{r}\mathrm{a}}-\mathrm{N}\mathrm{C}\mathrm{C}{\mathrm{T}}_{\mathrm{c}\mathrm{o}\mathrm{n}\mathrm{t}\mathrm{r}\mathrm{a}}}.}$$

Higher TPI values indicate greater contrast penetration and thus higher thrombus perviousness. The thrombus attenuation increase (TAI) was also calculated using the following formula to conduct sensitivity analyses:

$$\begin{array}{c}{\displaystyle }\mathrm{T}\mathrm{A}\mathrm{I}=\mathrm{C}\mathrm{T}{\mathrm{A}}_{\mathrm{t}\mathrm{h}\mathrm{r}\mathrm{o}\mathrm{m}\mathrm{b}\mathrm{u}\mathrm{s}}-\mathrm{N}\mathrm{C}\mathrm{C}{\mathrm{T}}_{\mathrm{t}\mathrm{h}\mathrm{r}\mathrm{o}\mathrm{m}\mathrm{b}\mathrm{u}\mathrm{s}}.\end{array}$$

Interobserver reliability was evaluated in 30 randomly selected cases, and the intraclass correlation coefficient showed excellent reproducibility (see Supplemental Methods for details).

Outcome Measures

Reperfusion treatment strategies were determined by a multidisciplinary stroke team based on clinical presentation, imaging findings, and contemporaneous guideline recommendations. In accordance with institutional protocols and prevailing stroke guidelines during the study period, patients presenting within the intravenous thrombolysis time window and without contraindications were preferentially treated with IVT as first-line therapy. Patients presenting beyond the thrombolysis time window or with specific indications underwent EVT. For analytic purposes, patients who underwent IVT followed by endovascular intervention were classified into the EVT group.

The primary outcome was functional independence at 90 days, defined as an mRS score ≤2. Secondary outcomes included the 180-day mRS score, symptomatic and asymptomatic intracranial hemorrhage, 90-day mortality, and, among EVT-treated patients, reperfusion status and distal embolization.

Statistical Analysis

All statistical analyses were performed using R software (version 4.4.2). Continuous variables were compared between groups using either the independent-samples t test or Mann-Whitney U test, as appropriate, and categorical variables were compared using either the χ² test or Fisher exact test. Multivariable logistic regression models were constructed to evaluate the independent associations of treatment modality, TPI, and other covariates with 90-day functional independence (mRS score ≤2). Covariates included age, sex, baseline NIHSS score, OTT, atrial fibrillation, hypertension, occlusion site, and metabolic parameters at admission. To examine whether thrombus perviousness acted as a modifier of treatment effects, an interaction term between treatment modality and TPI was incorporated into the primary regression model.¹⁶ To account for potential confounding owing to nonrandom treatment allocation, inverse probability of treatment weighting (IPTW) based on propensity scores was applied, as previously described.¹⁷ Covariate balance was assessed using standardized mean differences, with values <0.1 indicating acceptable balance. Restricted cubic spline analyses and descriptive assessment of mRS distributions were performed to further characterize outcome patterns across TPI levels. Additional sensitivity and robustness analyses are detailed in the Supplemental Methods. All statistical tests were 2-sided, and P<0.05 was considered statistically significant.

Results

Baseline Characteristics

A total of 255 patients meeting imaging-defined MeVO criteria were included, of whom 154 received IVT and 101 underwent EVT with or without prior IVT. The groups were comparable in age, sex, and most vascular risk factors (Table 1). Occlusion territory differed significantly, with middle cerebral artery occlusions more often treated with EVT and anterior cerebral artery/posterior cerebral artery occlusions more often treated with IVT (P<0.001). Among IVT-treated patients, 80.5% received alteplase and 19.5% received tenecteplase, with no difference across treatment pathways (P=0.29). Compared with IVT, EVT patients had higher baseline NIHSS scores (median 13 versus 10; P<0.001), more atrial fibrillation (27.7% versus 11.0%; P=0.003), and longer OTT (330 versus 185 minutes; P<0.001). Baseline Alberta Stroke Program Early Computed Tomography Score (ASPECTS) was lower in the EVT group than in the IVT group (median 8 versus 9; P<0.001), indicating greater baseline infarct burden in EVT-treated patients. CTA-derived thrombus density was slightly higher in the EVT group (P=0.007), whereas contralateral-normalized TPI did not differ (median, 0.09 in both; P=0.604; Figure S1).

Table 1.

Baseline Clinical and Imaging Characteristics of Patients Stratified by Treatment Modality (IVT Versus EVT)

Comparison of available baseline characteristics showed no significant differences in age, sex, baseline NIHSS, occlusion territory, or treatment modality between included patients and those excluded due to missing data (all P>0.05; Table S1).

Standardized mean differences were calculated before and after IPTW weighting. Before weighting, several baseline variables showed imbalance, particularly baseline NIHSS, OTT, atrial fibrillation, international normalized ratio, and occlusion site. After weighting, balance improved markedly, although slight residual imbalance remained for NIHSS and OTT; all standardized mean difference values were below 0.15, and most were below 0.10 (Table 2).

Table 2.

Covariate Balance Before and After Inverse Probability of Treatment Weighting Using PS-Logistic

During the early follow-up period (within 7 days postprocedure), NIHSS scores were significantly higher in the EVT group than in the IVT group (median, 7 versus 5, respectively; P=0.031), whereas the distributions of 7-day mRS scores were similar (median, 3 in both groups; P=0.260).

Interobserver reliability of TPI was excellent, with an intraclass correlation coefficient of 0.94 (95% CI, 0.88–0.97) based on 30 randomly selected cases independently assessed by 2 readers.

Primary Outcome

In the overall cohort, 90-day functional independence (mRS ≤ 2) did not differ between IVT and EVT (51.9% versus 51.5%, respectively; P=0.960). Univariable and multivariable logistic regression showed no independent association between treatment modality and 90-day outcome (adjusted odds ratio [OR], 1.34 [95% CI, 0.77–2.37]; P=0.309). Among covariates, only baseline NIHSS remained significant (OR, 0.93 [95% CI, 0.88–0.97]; Table S2).

After incorporating thrombus perviousness into the model, the association between treatment modality and the primary outcome changed materially. In the IPTW-weighted logistic regression model, the interaction between TPI and treatment modality reached statistical significance (interaction OR, 0.37 [95% CI, 0.24–0.56]; P<0.001). In the IPTW-weighted interaction model, the association between TPI and 90-day functional independence differed by reperfusion strategy used. Among IVT-treated patients, each 0.05-unit increase in TPI was associated with a higher likelihood of achieving functional independence (OR, 1.73 [95% CI, 1.19–2.53]; P=0.005). In contrast, in the EVT group, higher TPI was associated with a lower probability of functional independence (OR, 0.65 [95% CI, 0.49–0.87]; P=0.003; Table 3). The direction of effects was consistent in unweighted and IPTW-weighted models, indicating that the association between TPI and outcome was dependent on the treatment modality. Detailed coefficients are provided in Tables S3 through S5.

Table 3.

Interaction Between Thrombus Perviousness Index and Treatment Modality for 90-Day Functional Independence (mRS ≤2) Using IPTW

Further evidence of the interaction between TPI and treatment modality was demonstrated in restricted cubic spline analyses. In spline models incorporating the treatment-by-TPI interaction (Figure 3), the predicted probability of a favorable outcome increased with higher TPI values in the IVT group, whereas an opposite trend was observed in the EVT group. These patterns were concordant with the direction of the regression results and supported differential associations between thrombus perviousness and outcome across reperfusion strategies. The analysis of mRS score distribution (Figure S2) further demonstrated differences in functional outcomes across perviousness levels, with higher TPI values being associated with a greater proportion of lower disability grades. Consistent findings were observed in the analyses based on dichotomized TPI classifications; for example, in the EVT-treated patients, a higher rate of functional independence was observed in the low-TPI subgroup than in the high-TPI subgroup (60.4% versus 40.0%, respectively; P=0.029), whereas in the IVT-treated patients, more favorable outcomes were observed in the high-TPI subgroup (64.9% versus 43.4%; P=0.019; Table S6).

Figure 3.

Treatment-specific association between thrombus perviousness and 90-day functional independence. Restricted cubic spline curves showing the association between thrombus perviousness index (TPI) and the predicted probability of 90-day functional independence (modified Rankin Scale [mRS] score ≤2), stratified by treatment modality. Shaded areas indicate 95% CIs. EVT indicates endovascular therapy; and IVT, intravenous thrombolysis.

Extended Analysis of 180-Day Functional Outcome

At 180 days, functional outcomes were better in the IVT group in the unadjusted analysis (median mRS score, 1 versus 2; P=0.002). Using the same propensity score model as in the primary 90-day analysis, IPTW showed that EVT was associated with a lower likelihood of functional independence at 180 days in both univariable (OR, 0.43 [95% CI, 0.29–0.63]; P<0.001) and multivariable models (adjusted OR, 0.45 [95% CI, 0.30–0.68]; P<0.001).

In the IPTW-weighted interaction model, the TPI-by-treatment interaction remained significant (OR, 0.51 [95% CI, 0.32–0.80]; P=0.003), with higher TPI associated with improved outcomes in IVT and poorer outcomes in EVT. The main treatment effect was not significant in this parameterization (OR, 1.67 [95% CI, 0.65–4.26]; P=0.286). Detailed results are provided in Table S7.

Exploratory Threshold Analysis

An exploratory analysis was performed within the IVT subgroup to examine the shape of the association between TPI and 90-day functional independence. Analyses were conducted based on the IPTW-weighted sample, with trimming at the 2.5 and 97.5 percentiles, using restricted cubic splines in combination with 1000 bootstrap resamples.

In the weighted cohort, the predicted probability of a favorable outcome increased continuously and smoothly with rising TPI. A steeper gradient was observed in the low-perviousness range (approximate TPI range, 0.04–0.07), where probabilities generally remained below 50% across bootstrap replicates. At higher TPI values, the predicted probability increased and tended to plateau; although probabilities exceeding 70% were observed in some samples, no threshold interval showed consistent stability across resamples (Figure S3).

Overall, these findings suggested that among the IVT-treated patients, the association between the thrombus perviousness and outcome is better characterized as a continuous risk gradient rather than as a categorical variable defined by a single cutoff.

Safety Outcomes and EVT Subgroup Analyses

The rate of any hemorrhagic transformation was higher in the EVT group than in the IVT group (28.7% versus 13.0%, respectively; P=0.003), whereas neither symptomatic intracranial hemorrhage (3.0% versus 3.9%; P>0.990) nor 90-day mortality (5.9% versus 5.2%; P>0.990) differed significantly between groups. Because safety outcomes were infrequent, unweighted multivariable logistic regression models were used to maintain stability and avoid distortion from extreme IPTW weights. TPI was not associated with intracranial hemorrhage or hemorrhagic transformation. A modest association was observed between higher TPI and 90-day mortality (OR per 0.05 increase, 1.50 [95% CI, 1.01–2.22]; P=0.045), without altering the overall absence of mortality difference between treatment groups (Table 4).

Table 4.

Association Between Thrombus Perviousness Index and Safety Outcomes

Treatment-stratified analyses showed no mortality association in IVT-treated patients (OR per 0.05 increase, 0.92 [95% CI, 0.36–2.36]; P=0.86), but a significant association in EVT-treated patients (OR, 2.01 [95% CI, 1.01–4.02]; P=0.047), indicating that the overall signal was driven by the EVT subgroup.

In the EVT subgroup, TPI was not significantly associated with successful recanalization (OR, 0.84 [95% CI, 0.59–1.17]; P=0.296) or distal embolization (OR, 1.12 [95% CI, 0.74–1.59]; P=0.534; Table S8).

Sensitivity Analyses

The sensitivity analyses conducted using an alternative surrogate for thrombus permeability produced results consistent with those of the primary analysis. TAI was not independently associated with the 90-day functional independence outcome in the main-effect model (P=0.636); however, a significant interaction between TAI and treatment modality persisted in the interaction model (TAI × EVT OR, 0.89 [95% CI, 0.84–0.93]; P<0.001). Within this interaction framework, higher TAI was associated with greater benefit from IVT and reduced benefit from EVT (Table S9).

Using covariate balancing propensity score (CBPS) weighting further improved covariate balance, with most standardized mean differences approaching zero. Compared with propensity score-logistic IPTW, CBPS weighting modestly reduced the effective sample size (Table S10), without evidence of extreme weight concentration (Figure S4). In CBPS-weighted models, the TPI-by-treatment interaction remained significant (OR, 0.26 [95% CI, 0.18–0.37]; P<0.001). After accounting for treatment modality, higher TPI values were consistently associated with increased odds of 90-day functional independence in the IVT group (OR, 2.40 [95% CI, 1.75–3.29]; P<0.001), whereas an inverse association was observed in the EVT group (OR, 0.67 [95% CI, 0.50–0.89]; P=0.005; Table S11). Overall, these findings were consistent with the primary IPTW-based results, further supporting the robustness of the observed interaction.

The primary IPTW-weighted interaction models were re-estimated using heteroskedasticity-consistent robust standard errors and survey-weighted generalized linear models. Across alternative variance estimation approaches, the TPI-by-treatment interaction remained significant with only minor effect-size variations (Table S11). Treatment-stratified associations across permeability metrics and weighting strategies are summarized in Figure S5. Across analytic frameworks, higher thrombus perviousness remained associated with improved outcomes in IVT and poorer outcomes in EVT. Thrombus perviousness was not correlated with baseline NIHSS or OTT (all Spearman P>0.40; Table S12).

To address potential heterogeneity related to bridging therapy, we excluded patients who received intravenous thrombolysis before EVT (17/101, 16.8%). The TPI-by-treatment interaction remained significant (OR, 0.44 [95% CI, 0.26–0.76]; P=0.003). Given the reduced sample size after excluding patients who received bridging therapy, a conventional multivariable regression model was used for this sensitivity analysis (Table S13). Restriction to anterior circulation MeVO preserved a significant interaction in the IPTW-weighted model (OR, 0.34 [95% CI, 0.21–0.56]; P<0.001), indicating that the effect remained robust after accounting for vascular territory differences (Table S14). In addition, further adjustment for detailed occlusion territories (anterior cerebral artery, middle cerebral artery, posterior cerebral artery) within the IPTW-weighted interaction model did not materially alter the results, and the TPI-by-treatment interaction remained statistically significant (interaction OR, 0.44 [95% CI, 0.25–0.74]; P=0.003; Table S15).

Additional ASPECTS-adjusted sensitivity analyses showed consistent results. After adding ASPECTS to the IPTW-weighted interaction model, the TPI-by-treatment interaction remained significant (OR, 0.37 [95% CI, 0.24–0.57]; P<0.001). Re-estimating the propensity score model with ASPECTS yielded similar findings (OR, 0.37 [95% CI, 0.24–0.58]; P<0.001). The interaction also remained significant for 180-day functional independence (OR, 0.51 [95% CI, 0.32–0.80]; P=0.003; Table S16).

E-value analysis indicated that to fully explain away the observed interaction (OR, 0.37), an unmeasured confounder would need to be associated with both treatment selection and functional outcome by a risk ratio of at least 4.84 each, above and beyond the measured covariates (E-value for the upper confidence limit, 2.98).

Discussion

In this study of acute ischemic stroke due to MeVO, thrombus perviousness emerged not only as an outcome-associated imaging feature but also as a variable associated with differential treatment-outcome relationships across IVT and EVT. Most previous studies comparing reperfusion strategies have implicitly assumed that imaging characteristics influence different treatments in a uniform direction, whereas the present findings challenge this assumption. Without accounting for thrombus perviousness, 90-day functional independence outcomes of IVT and EVT were not clearly distinguishable; however, when thrombus perviousness was incorporated as a continuous variable with a treatment interaction, the direction of the treatment-outcome association varied substantially across perviousness levels. Importantly, TPI distributions were similar between the 2 treatment groups (median, 0.09 in both), suggesting that the observed interaction was unlikely to be driven by baseline differences in thrombus perviousness. Although no significant treatment difference was observed at 90 days, IVT was associated with better functional outcomes at 180 days after IPTW adjustment, possibly reflecting differences in longer-term recovery trajectories. However, residual confounding due to unmeasured differences between treatment groups cannot be excluded. The TPI-by-treatment interaction remained significant in the 180-day analysis, suggesting that the treatment-specific association between thrombus perviousness and functional outcome was directionally consistent across follow-up time points.

The biological plausibility of these findings is supported by differences in treatment mechanisms. IVT depends on the penetration and diffusion of administered thrombolytic agents within the fibrin network¹⁷; thrombi with higher perviousness likely contain fluid channels and looser fibrin architecture, facilitating intrathrombus drug distribution and enhancing the effectiveness of thrombolysis.¹⁸ In contrast, EVT is primarily reliant on mechanical engagement and en bloc clot retrieval.¹⁹ Histopathologic studies have shown that fibrin- and platelet-rich thrombi, which are often more pervious, are more prone to fragmentation and distal embolization during mechanical manipulation. Even when angiographic recanalization is achieved, microcirculatory reperfusion may remain impaired, potentially attenuating the functional benefits observed at higher perviousness levels.²⁰ Therefore, the opposite directional associations that were observed for IVT and EVT across perviousness levels likely reflect fundamental differences in treatment mechanisms. The robustness of the interaction was further supported by E-value analysis, suggesting that only a relatively strong unmeasured confounder could fully account for the observed association.

In the IVT subgroup, higher TPI values were associated with a monotonic increase in the probability of a favorable outcome, with no evidence of a distinct nonlinear inflection point. This pattern suggests that thrombus perviousness exerts a sustained modulatory effect on thrombolysis rather than acting within a restricted value range. Bootstrap analyses of the weighted sample indicated that the relationship between thrombus perviousness and outcomes in IVT-treated patients is better characterized as a continuous risk gradient rather than a threshold effect with a clearly defined cutoff. In routine clinical practice, standardized early vascular reassessment after intravenous thrombolysis is not systematically performed, particularly in patients with distal MeVO. Therefore, early recanalization after IVT could not be reliably assessed in this retrospective cohort. Future prospective studies incorporating standardized early vascular imaging are needed to clarify whether thrombus perviousness is associated with IVT responsiveness. Accordingly, this analysis should be regarded as exploratory and hypothesis-generating in nature, describing the trend between perviousness and thrombolytic benefits rather than supporting perviousness-based clinical stratification at the present stage of investigation.²¹ In contrast, in the EVT setting, although TPI exhibited a directional association with the primary outcome, no stable or reproducible nonlinear structure was observed across different perviousness ranges, suggesting that the effect of thrombus perviousness may be diminished by multiple determinants of mechanical thrombectomy effectiveness. Beyond thrombus structure, EVT outcomes are also influenced by device selection,²² thrombus length,²³ thrombus-vessel wall interactions, and procedural approaches.²⁴ Furthermore, distal embolization, microcirculatory hypoperfusion, and reperfusion injury may further weaken any direct association between a single imaging marker and functional outcomes. In the present cohort,²⁵ TPI was not significantly associated with either successful recanalization or distal embolization, indicating that in EVT, thrombus perviousness is more likely to reflect shifts in the relative advantage of reperfusion strategies rather than serving as a direct determinant of technical success.

The present findings are particularly relevant from a clinical decision-making perspective, suggesting that thrombus perviousness should not be interpreted as a binary criterion for selecting or excluding a specific reperfusion strategy, but rather as an imaging feature that informs the relative benefit of different treatment options. In patients with MeVO with highly pervious thrombi, IVT is more likely to achieve functional recovery and may represent a feasible and cost-effective treatment option, particularly when endovascular availability is limited or transfer delays are anticipated. In contrast, when imaging suggests a compact, low-perviousness thrombus, thrombolysis efficacy may be limited. If mechanical reperfusion is delayed under such circumstances, the window for salvaging potentially reversible ischemic tissue may be further narrowed; thus, early EVT performance may be more likely to confer a relative clinical advantage. This study focused on treatment-outcome patterns across the perviousness spectrum rather than defining a single bedside cutoff. While this graded pattern was directionally consistent with prior observations in studies of LVO showing an association of highly pervious thrombi with reduced urgency for EVT,²⁶ it also reveals a more nuanced continuum of treatment effects in the MeVO population.

With respect to safety, thrombus perviousness exhibited a weaker association with adverse outcomes than with treatment-related differences in effectiveness. Although the rate of hemorrhagic transformation was higher in the EVT group, intracranial hemorrhage and mortality rates were similar between groups. Thrombus perviousness was not an independent predictor of hemorrhagic complications, and it demonstrated only a modest association with mortality. These findings suggest that in MeVO, thrombus perviousness is better suited for determining relative treatment benefits rather than for stratifying hemorrhagic risk or excluding patients from reperfusion therapy. Notably, the observed positive association between higher TPI and mortality in the overall cohort was confined to the EVT subgroup, which helps reconcile the apparent discrepancy between improved functional outcomes with higher TPI in IVT-treated patients and the overall mortality findings. One possible explanation is that higher-perviousness thrombi may be more prone to fragmentation or impaired microcirculatory reperfusion during EVT.

From a methodological standpoint, logistic regression-based IPTW was selected as the primary analytical framework owing to its stability and interpretability in a modestly sized cohort with a limited set of covariates. Although CBPS weighting achieved slightly improved covariate balance, this came at the cost of a marked reduction in the effective sample size, highlighting a tradeoff between balance and estimation stability in the MeVO setting. Accordingly, the CBPS approach was reserved for the sensitivity analyses to assess the robustness of the findings rather than being used for primary inference. Furthermore, the consistency of the results across heteroskedasticity-robust standard errors and survey-weighted generalized linear models indicated that the observed interaction effect was not driven by variance specification.

It is important to emphasize that this study was conducted specifically in patients with acute ischemic stroke due to MeVO. Compared with LVO, MeVO differs substantially in terms of vessel caliber, anatomic variability, and procedural controllability, which limits the direct extrapolation of reperfusion-related evidence derived from populations with LVO. The continuous gradient observed between thrombus perviousness and treatment effects in the present study suggests that decision-making regarding reperfusion strategies in MeVO may require an imaging-based framework that is distinct from that established for LVO. If reproduced in larger multicenter prospective studies, thrombus perviousness assessment may support more individualized reperfusion strategies in MeVO.

Substantial baseline differences were observed between the IVT and EVT groups, reflecting real-world treatment selection. Although inverse probability weighting improved overall covariate balance, slight residual imbalance remained for certain variables, particularly baseline NIHSS and OTT. Residual confounding inherent to observational treatment assignment therefore cannot be fully excluded. Specifically, greater stroke severity or longer treatment delay may increase the likelihood of EVT selection while simultaneously being associated with poorer functional outcomes, which could theoretically influence the magnitude of the observed interaction. However, both baseline NIHSS and OTT were explicitly incorporated into the propensity score and multivariable interaction models, and postweighting covariate balance was acceptable. Moreover, thrombus perviousness was not significantly associated with OTT in the present cohort, and the TPI-by-treatment interaction remained robust after accounting for treatment timing. The TPI-by-treatment interaction also remained statistically significant across sensitivity analyses excluding bridging therapy and restricting the cohort to anterior circulation occlusions. Adjustment for detailed occlusion territory did not materially alter the TPI-by-treatment interaction, suggesting that the observed association remained robust after accounting for vascular territory heterogeneity. Additional adjustment for ASPECTS, including re-estimation of the propensity score model, did not materially change the TPI-by-treatment interaction, suggesting that the primary findings were robust after accounting for baseline infarct burden. Although residual confounding cannot be entirely ruled out, it is unlikely to fully account for the differential association.

In addition, statistical interaction testing generally requires larger sample sizes than main-effect analyses. Although the interaction between TPI and treatment modality demonstrated statistical significance with a relatively narrow CI in the primary model, the overall sample size remains modest for interaction assessment. Therefore, the present findings should be regarded as hypothesis-generating and require validation in larger, multicenter cohorts.

Limitations

Despite the positive findings, the results should be interpreted with caution. First, this was a single-center retrospective cohort study. Although the sample size supported interaction and continuous-effect analyses, the observed perviousness gradients and bootstrap-derived interval patterns may reflect center-specific characteristics. Accordingly, the findings should be regarded as trend-level observations requiring validation in independent prospective cohorts. Second, TPI calculation depends on specific CTA acquisition protocols, Hounsfield unit calibration, and contralateral normalization procedures. Absolute TPI values may vary across imaging platforms and reconstruction parameters, underscoring the need for workflow standardization and external calibration before multicenter application.²⁷ Third, thrombus length was not included in the present analysis. In MeVO, accurate thrombus length measurement is often challenging because of complex branch anatomy, simultaneous division involvement, or indistinct boundaries, which may limit reproducibility. In contrast, attenuation-based perviousness metrics derived from localized regions may offer greater feasibility and measurement consistency. Given the small caliber of MeVO vessels, larger ROIs may increase partial-volume effects by incorporating the vessel wall or surrounding tissue. Although smaller ROIs may be more sensitive to image noise, the use of spatially distributed sampling and averaging was intended to enhance signal stability. Alternative ROI sizes were not systematically evaluated and warrant future study. Fourth, the lack of standardized early vascular reassessment after IVT precluded reliable evaluation of early recanalization. This should be addressed in future prospective studies. Fifth, to reflect real-world clinical pathways, patients who underwent EVT with or without prior IVT were included in a single EVT group. Although weighted and sensitivity analyses showed consistent directions, treatment heterogeneity may have attenuated subtle between-group differences.²⁸ Finally, despite adjustment for multiple established prognostic factors, residual confounding could not be fully excluded, including unmeasured thrombus composition or postreperfusion microcirculatory factors. Accordingly, the observed effects of the TPI should be interpreted in the context of imaging phenotype-based treatment stratification rather than as a complete explanation of the underlying pathophysiological mechanisms.

Conclusions

In patients with acute ischemic stroke due to MeVO, thrombus perviousness was associated with differential functional outcomes across IVT and EVT. As thrombus perviousness increased, the functional benefits of IVT rose progressively, whereas endovascular or bridging therapies appeared to confer a relative advantage in patients with thrombi with low perviousness. Exploratory analyses indicated that the outcome probabilities varied across the thrombus perviousness spectrum; however, stable clinical thresholds were not identified. Overall, the findings suggest that thrombus perviousness is better conceptualized as a continuous imaging modifier that may inform stratified reperfusion strategies for MeVO.

ARTICLE INFORMATION

Acknowledgments

The authors thank Editage (www.editage.cn) for English language editing.

Author Contributions

Guarantors of the integrity of the entire study: Drs Jin and Guo. Study concept and design: Drs Zhang, Wu, and Jin. Data acquisition: Drs Zhang, Wu, Qu, Zhang, Jia, Zhang, and Wang. Imaging analysis and thrombus perviousness quantification: Drs Zhang, Wu, and Qu. Statistical analysis: Drs Zhang, Wu, and Guo. Manuscript drafting: Drs Zhang and Wu. Manuscript revision for important intellectual content: Drs Guo and Jin, all authors. Approval of the final manuscript for submission: all authors. Accountability statement: all authors agree to be accountable for all aspects of the work and to ensure that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. Drs Jin and Guo are the guarantors of this work. Provenance and peer review: Not commissioned; externally peer reviewed.

Disclosures

None.

Supplemental Material

Supplemental Methods

Tables S1–S16

Figures S1–S5

STROBE Checklist

Funding Statement

This work was supported by the Noncommunicable Chronic Diseases–National Science and Technology Major Project (grant nos. 2025ZD0546500 and 2025ZD0546503).