# Development and validation of a clinical model for predicting 90-day outcomes after endovascular therapy with adjunctive tirofiban in acute ischemic stroke

## TL;DR

This study creates a model to predict outcomes in stroke patients treated with endovascular therapy and tirofiban, helping doctors make better treatment decisions.

## Contribution

A new clinical prediction model for stroke outcomes using tirofiban and endovascular therapy was developed and validated.

## Key findings

- Stroke-associated pneumonia, high NIHSS scores, and smoking history were linked to poor outcomes.
- Successful reperfusion significantly reduced the risk of poor outcomes.
- The model showed strong discrimination with an AUC of 0.83.

## Abstract

Endovascular therapy (EVT) represents a cornerstone in the treatment of acute ischemic stroke due to large vessel occlusion (AIS-LVO). Despite high recanalization rates, ineffective microcirculatory reperfusion and early reocclusion can compromise clinical outcomes. The adjunctive use of tirofiban, a glycoprotein IIb/IIIa inhibitor, has been proposed to mitigate these risks, yet identification of patients who may benefit is uncertain. We aimed to develop and validate a clinical prediction model for 90-day poor functional outcome in AIS-LVO patients undergoing EVT with tirofiban.

We conducted a retrospective cohort study of 177 consecutive AIS-LVO patients who received EVT plus tirofiban at a single academic center. The primary outcome was a poor functional outcome, defined as modified Rankin Scale score 3–6 at 90 days. Secondary outcomes included successful reperfusion (mTICI 2b–3), symptomatic intracranial hemorrhage (sICH), and 90-day mortality. Using 70% of the cohort for model development, we constructed predictors via multivariable logistic regression and machine learning approaches (including XGBoost, Random Forest, and others). Predictors comprised baseline clinical, imaging, and procedural variables. Model performance was assessed by area under the curve (AUC), calibration plots, and decision curve analysis (DCA), sensitivity, specificity, precision.

Poor functional outcome was observed in 50.8% of patients. Multivariable analysis identified stroke-associated pneumonia (OR 7.56, 95% CI 2.75–20.77), higher baseline NIHSS score (OR 1.13, 95% CI 1.03–1.24), and smoking history (OR 2.86, 95% CI 1.19–6.85) as independent predictors of poor outcome, while successful reperfusion was protective (OR 0.06, 95% CI 0.01–0.57). The final nomogram model demonstrated good discrimination (AUC 0.83, 95% CI 0.75–0.90) and calibration (Hosmer–Lemeshow test, p = 0.539).

We developed and validated a pragmatic prediction model incorporating readily available clinical and procedural variables to estimate the risk of 90-day poor outcome in AIS-LVO patients treated with EVT and tirofiban. This tool may assist clinicians in individualized outcome prediction and inform adjunctive antithrombotic strategies in neurovascular care.