What is the clinical evidence from this study?

Study design: Cohort. Population: Acute ischaemic stroke (n=298). Intervention: Four-variable predictive model (PWV, TyG, sex, eGFR) vs. Pulse wave velocity (PWV) alone. Primary outcome: Discrimination of large-artery atherosclerosis vs small-vessel occlusion (AUC) (ΔAUC 0.058, 95% CI 0.623-0.747, p=0.043).

What does this research mean for the field?

A four-variable model combining pulse wave velocity, triglyceride-glucose index, sex, and eGFR achieves moderate discrimination (AUC 0.685) between large-artery atherosclerosis and small-vessel occlusion in acute ischemic stroke, significantly outperforming single markers. Novelty: ClaimNovelty.INCREMENTAL. Consensus alignment: ConsensusAlignment.NEUTRAL.

March 31, 2026Open Access

Combined pulse wave velocity and triglyceride–glucose index to discriminate large-artery atherosclerosis from small-vessel occlusion in acute ischaemic stroke: a single-centre retrospective study

Key Result

A four-variable model combining pulse wave velocity, triglyceride-glucose index, sex, and eGFR achieved an AUC of 0.685 for discriminating large-artery atherosclerosis from small-vessel occlusion, significantly outperforming pulse wave velocity alone.

Study Design

Type

Cohort (n=298)

Multicenter

Structured PICO

Does combining carotid pulse wave velocity and triglyceride-glucose index improve early in-hospital discrimination of large-artery atherosclerosis from small-vessel occlusion in acute ischaemic stroke patients?

Population

298 consecutive acute ischaemic stroke patients (179 with small-vessel occlusion, 119 with large-artery atherosclerosis), age ≥18 years, first-ever cerebrovascular event, last-known-well to admission ≤24 h. Mean age 64.71, 60.4% male.

Intervention

Predictive model combining carotid pulse wave velocity (PWV) and triglyceride-glucose (TyG) index (along with clinical indicators like sex and eGFR)

Comparator

Single markers (PWV alone or TyG alone)

Outcome

Discrimination of large-artery atherosclerosis (LAA) versus small-vessel occlusion (SVO) assessed by ROC AUCsurrogate

Combining carotid pulse wave velocity and the triglyceride-glucose index provides moderate, incremental value for early in-hospital discrimination of large-artery atherosclerosis from small-vessel occlusion in acute ischaemic stroke.

Main Result

Effect estimate: ΔAUC 0.058 (95% CI 0.623-0.747)

Absolute Event Rate: 0.685% vs 0.627%

p-value: p=0.043

Limitations

Single-center, retrospective design and relatively modest sample size
Device-derived local carotid pulse wave velocity is inherently dependent on platform-specific hardware and algorithms
Residual confounding may persist, particularly from pre-admission medications
A priori focus on differentiating LAA from SVO necessarily excluded other TOAST categories, narrowing the case mix
Detailed early in-hospital management at the time of measurement was not systematically captured
Some degree of residual selection bias due to incomplete data cannot be fully excluded

Abstract

Objective Early etiologic discrimination between large-artery atherosclerosis (LAA) and small-vessel occlusion (SVO) is clinically relevant when advanced imaging is unavailable or delayed. We examined whether combining carotid pulse wave velocity (PWV) with the triglyceride–glucose (TyG) index improves early in - hospital discrimination of LAA vs. SVO, and developed a predictive model integrating PWV, TyG, and clinical indicators. Methods We performed a single-centre retrospective study including consecutive acute ischaemic stroke patients from September 2021 to November 2023. Carotid PWV (Wv, m/s) was measured bilaterally and averaged; TyG was calculated from fasting triglycerides and glucose. Restricted cubic splines (RCS) tested non-linearity. Multivariable logistic regression was applied to (i) estimate the adjusted association of PWV and TyG with LAA versus SVO, and (ii) identify independent predictors via backward stepwise selection and establish predictive model. Discrimination was assessed by ROC AUC with DeLong tests; calibration used bootstrap-corrected curves; internal validation used 1,000-bootstrap optimism correction; decision-curve analysis appraised clinical utility. Results Of 892 admissions screened, 298 were analysed (SVO 179; LAA 119). LAA showed higher PWV (median 17.72 vs. 15.02 m/s; p 0.001) and higher TyG (8.85 vs. 8.64; p = 0.005) than SVO. RCS supported overall associations of PWV and TyG with LAA versus SVO without evidence of non-linearity. Compared with single markers (AUC: PWV = 0.627; TyG = 0.596), a two-marker model (PWV + TyG) achieved AUC 0.654, while a parsimonious four-variable model (PWV, TyG, sex, eGFR) reached AUC 0.685 and significantly outperformed either single marker (paired DeLong vs. PWV: ΔAUC 0.058, p = 0.043; vs. TyG: ΔAUC 0.089, p = 0.011). Calibration was acceptable after bootstrap correction. Conclusion PWV and TyG provide complementary, early in - hospital signals for moderate discrimination of LAA vs. SVO, with incremental value when combined. The predictive model constructed by combining PWV, TyG, and clinical indicators also performed well. These bedside markers provide adjunctive, early in-hospital risk stratification for suspected LAA versus SVO.

AI에게 질문

Bookmark

View Full Paper