Pulsatility index is a prognostic marker of cognitive outcome in patients with transient ischemic attack and mild acute stroke Authors: Nicoletta Brunelli, Claudia Altamura, Marilena Marcosano, Carlo Vico, Marcel Bach-Pages, Luigi d’Onofrio, Sergio Soeren Rossi, Carmelina Maria Costa, Simona Sacco, Mauro Silvestrini, Fabrizio Vernieri Published: 2026 ABSTRACT Highlights•Cognitive performance improved over time after stroke/TIA, with the greatest gains observed in MoCA scores during follow-up.•Lower cerebral vascular resistance, less severe stroke, and better baseline cognition were associated with more favorable long-term cognitive outcomes.•Higher pulsatility index (PI), reflecting increased cerebrovascular resistance, predicted poorer cognitive performance, highlighting the potential role of cerebral hemodynamics in post-stroke cognitive recovery. Introduction Stroke is the second-leading cause of death and the third-leading cause of death and disability combined, as measured by disability-adjusted life years (DALYs), throughout the world [1]. Although mortality rates have gradually decreased in recent years, an increasing number of stroke survivors are left to manage a variety of physical, cognitive, functional, and psychosocial challenges [2]. Among these, post-stroke cognitive impairment is especially prevalent, affecting up to 75% of patients, and it is independently linked to worse long-term outcomes, including reduced quality of life and increased disability [3,4]. Identifying patients with the potential for cognitive recovery could help guide their expectations and set more realistic, attainable goals in rehabilitation programs. However, predicting cognitive recovery following ischemic stroke remains inaccurate [5]. Emerging factors that may influence recovery include demographic variables (such as age, sex, and education level), stroke-related characteristics (such as location and severity), cognitive and emotional status, imaging findings (e.g., infarct volume and white matter hyperintensity) and vascular risk factors such as diabetes and smoking; however, results vary between studies [[6], [7], [8], [9], [10], [11]]. Vascular risk factors such as hypertension and diabetes can induce structural vessel changes in microcirculation, leading to lipohyalinosis and fibrinoid degeneration as well as in the intra- and extracranial macrocirculation, resulting in impairment of cerebral hemodynamics due to an increase in vascular resistance, reduced diastolic velocity, and increase in Pulsatility Index (PI) [12,13]. The PI was first described by Gosling and King [14] and characterizes the shape of the spectral waveform, independent of the angle of insonation of the transcranial doppler probe [15]. PI is measured as the difference between Peak Systolic Velocity (PSV) and End-Diastolic Velocity (EDV), divided by Mean Flow Velocity (MFV), and the normal range is between 0.6 and 1.1 [16]. It is hypothesized to reflect the vascular resistance in the artery distal to the probe [15]. An increase in distal resistance in the cerebral circulation, usually due to a downstream stenosis or to narrowing of the small vessels due to atherosclerosis, results in a reduction in diastolic velocity and an increase in PI [17]. Other conditions that can lead to an increase in PI are intracranial hypertension, hypocapnia, and chronic arterial hypertension [18]. PI has also been reported to increase in ageing, diabetes mellitus, Small Vessels Disease (SVD), and dementia [12,13,17,19] and to have a predictive value for cerebrovascular diseases and cognitive impairment [[20], [21], [22]]. In this scenario, we envisage that an increase in vascular resistance and a reduction in diastolic flow lead to an increase in PI and to cerebral hypoperfusion in the long run. Persistent cerebral hypoperfusion may not guarantee an adequate supply of oxygen to all brain areas, causing an alteration of cognitive and functional performance over time. Supporting this hypothesis, previous studies have demonstrated that PI is associated with cognitive performance in patients with lacunar infarcts and that an elevated PI might be related to impairment in several cognitive domains [23]. However, to our knowledge, no data are available about PI in other stroke subtypes. Currently, no prognostic markers are available in the acute phase to assess the cognitive outcome of patients with acute stroke or TIA with different ethiopatogenesis in the short and long term. Therefore, we aimed to explore the cerebral hemodynamics evaluated by PI in patients with TIA or acute stroke with mild symptoms to better understand its implication in the etiopathogenesis of different stroke subtypes and to assess if it may represent a prognostic marker of cognitive outcome. Methods This is a longitudinal observational study carried out at the Headache and Neurosonology Unit of the Neurology Unit at Fondazione Policlinico Universitario Campus Bio-Medico of Rome. The protocol was approved by the Ethics Committee of our hospital (Protocol: 53/18 OSS), and informed consent was obtained from the patients. Study population We consecutively considered patients admitted to the Neurology Unit between January 2020 and August 2022 for enrolment. We included patients who met the following inclusion criteria: age ≥ 18 years, diagnosis of TIA or acute mild stroke with NIHSS ≤ 8 [24], supratentorial stroke vascular territory, an adequate temporal acoustic window for ultrasound insonation, transcranial doppler (TCD) performed between 48 and 72 hours from stroke onset and the ability to provide informed consent. We excluded patients with the following characteristics: inadequate temporal bone window, hemodynamically significant stenoses (≥70%) or arterial occlusions of the intracranial and extracranial circulation, presence of severe aphasia or dysarthria, Alzheimer’s disease (AD) or other types of dementia, or receipt of revascularization therapy in the acute phase. Extra- and intra-cranial arteries were assessed by continuous wave Doppler and color flow B-mode Doppler ultrasound (Philips iU22, Bothell, WA, USA). We collected vascular risk factors through medical history interviews with all subjects. We defined hypertension as a history of high blood pressure, a systolic blood pressure ≥ 140 mmHg, diastolic blood pressure ≥ 90 mmHg, or the use of an antihypertensive; diabetes as a fasting blood glucose level of ≥ 126 mg/dL or current treatment for diabetes; hypercholesterolemia as a serum total cholesterol level of 200 mg/dL or the use of lipid-lowering medications. Participants were classified as smokers if current smokers. All patients were evaluated with diagnostic tests to define the etiopathogenesis according to TOAST (Trial of Org 10172 in Acute Stroke Treatment) classification system [24]. From an etiopathogenetic point of view, following the TOAST classification, ischemic stroke can be classified in five subtypes: large-artery atherosclerosis, cardioembolism, small-vessel occlusion, stroke of other determined etiology and stroke of undetermined etiology. Clinical evaluation The study consisted of four clinical evaluations conducted over one year (T1, T2, T3, T4): the first evaluation (T1, baseline) performed at admission between 48-72 hours after stroke/TIA onset) included TCD, cognitive assessments using Mini-Mental State Examination (MMSE) and Montreal Cognitive Assessment (MoCA), and the National Institute Stroke Scale (NIHSS); NIHSS at stroke onset was also obtained from clinical records. All cognitive test scores were adjusted according to standard correction procedures. MMSE scores were corrected for age and years of education, whereas MoCA scores were corrected for years of education. The second evaluation (T2) conducted at 1-month-follow up (±1 week), included MMSE and MoCA assessments. The third evaluation (T3) at 6-month-follow up (±1 week), included TCD along with MMSE and MoCA. The fourth evaluation (T4) at 12-month-follow up (±1 week) consisted of MMSE and MoCA assessments. Hemodynamic evaluation In our Neurosonology Lab, TCD examinations were performed according to guidelines [25]. Two transducers of transcranial doppler were placed on the temporal bone window with a stable angle of insonation to obtain a bilateral continuous measurement of flow velocity in MCA (unaffected hemisphere) and PCA (affected hemisphere). The PI of MCA and PCA was calculated by subtracting the EDV from PSV and then dividing by the MFV. We then divided the cohort into patients with a PI ≤ 1.1 and those with a PI > 1.1, as previously described [16]. Endpoints Primary endpoint −Evaluation of longitudinal changes in PI, MoCA scores, and MMSE scores throughout the follow-up period −Evaluation of longitudinal changes in PI, MoCA scores, and MMSE scores throughout the follow-up period Evaluation of longitudinal changes in PI, MoCA scores, and MMSE scores throughout the follow-up period Secondary endpoints −The association between baseline PI and longitudinal changes in cognitive performance, assessed by variations in MoCA and MMSE scores over the follow-up period.−The association between vascular risk factors and PI −The association between baseline PI and longitudinal changes in cognitive performance, assessed by variations in MoCA and MMSE scores over the follow-up period. The association between baseline PI and longitudinal changes in cognitive performance, assessed by variations in MoCA and MMSE scores over the follow-up period. −The association between vascular risk factors and PI The association between vascular risk factors and PI Statistical analysis The sample size was determined based on a previous study with a similar design in which the authors included 113 patients with a lacunar syndrome [23]. The interval variables between groups were compared using the Kruskal-Wallis test or Wilcoxon test (medians with interquartile range [IQR]) and using Spearman tests for correlations, according to the results of the Kolmogorov-Smirnov test for data distribution. Longitudinal changes of non-parametric variables were assessed with the Friedman test. A generalized linear model (GLM) was used to examine the association between the dependent variables T4 MOCA and T4 MMSE and the explanatory factors (PI at T1, and NIHSS at T1) while adjusting for potential covariates (T1 cognitive test score). Statistical significance was set at two-tailed p < 0.05. Statistical analyses were performed using SPSS version 31.0 (SPSS Inc., Chicago, IL, USA). Results Patient population and characteristics at baseline We consecutively screened 725 patients with TIA or acute stroke. Of these, 140 did not have a temporal acoustic window suitable for ultrasound insonation, 108 had an NIHSS > 8, 40 were diagnosed with stroke mimic conditions, 62 had an infratentorial stroke, 125 had a large vessel occlusion or hemodynamically significant stenoses (≥70%), 48 were admitted more than 72 hours from stroke onset, 26 had dementia or Alzheimer’s disease and 52 had severe aphasia or dysarthria. A total of 124 patients were finally included (71% men), median age of 66.0 (IQR 19.5) years, with TIA or acute stroke with mild symptoms (median NIHSS 2, IQR 2). Of note, 15 patients were lost at follow-up after baseline evaluation, and therefore, PI values at 6 months are available for 109 patients. Demographic characteristics, clinical and cognitive scales, hemodynamic parameters, and radiological data at baseline (T1) are summarized in Table 1. Percentages of vascular risk factors are represented in Fig. 1. The different stroke/TIA subtypes classified from an etiopathogenetic point of view according to TOAST classification [24] are shown in Fig. 2.Table 1Demographic characteristics, clinical and cognitive scales, hemodynamic parameters, and radiological data at baseline (T1).Table 1 dummy alt textPatients (N=124)Demographic characteristicsAge, median (IQR)66 (19.5)Sex, n (% of males)88 (71%)Clinical scalesNIHSS onset, median (IQR)2 (2)NIHSS at 48-72h (T1), median (IQR)1 (3)Cognitive scalesAdjusted MMSE, median (IQR)27.9 (3.7)MoCA, median (IQR)26.5 (4)Hemodynamic parametersMCA PI, median (IQR)1.1 (0.3)PCA PI, median (IQR)1.3 (0.5)Radiological dataDWI +, n (%)86 (69.4%)Subcortical infarcts, n (% of stroke)45 (52.3%)Cortical-subcortical infarcts, n (% of stroke)41 (47.7%)Involvement of the right hemisphere, n (%)50 (40.3%)Involvement of the left hemisphere, n (%)68 (54.9%)Involvement of both hemispheres, n (%)6 (4.8%)Fig. 1Vascular risk factors registered in the cohort. BMI: Body Mass Index; PFO: Patent Foramen Ovale; IHD: Ischemic Heart Disease.Fig 1 dummy alt textFig. 2Stroke etiopathogenesis of the cohort according to TOAST classification.Fig 2 dummy alt text Table 1Demographic characteristics, clinical and cognitive scales, hemodynamic parameters, and radiological data at baseline (T1).Table 1 dummy alt textPatients (N=124)Demographic characteristicsAge, median (IQR)66 (19.5)Sex, n (% of males)88 (71%)Clinical scalesNIHSS onset, median (IQR)2 (2)NIHSS at 48-72h (T1), median (IQR)1 (3)Cognitive scalesAdjusted MMSE, median (IQR)27.9 (3.7)MoCA, median (IQR)26.5 (4)Hemodynamic parametersMCA PI, median (IQR)1.1 (0.3)PCA PI, median (IQR)1.3 (0.5)Radiological dataDWI +, n (%)86 (69.4%)Subcortical infarcts, n (% of stroke)45 (52.3%)Cortical-subcortical infarcts, n (% of stroke)41 (47.7%)Involvement of the right hemisphere, n (%)50 (40.3%)Involvement of the left hemisphere, n (%)68 (54.9%)Involvement of both hemispheres, n (%)6 (4.8%) Demographic characteristics, clinical and cognitive scales, hemodynamic parameters, and radiological data at baseline (T1). Vascular risk factors registered in the cohort. BMI: Body Mass Index; PFO: Patent Foramen Ovale; IHD: Ischemic Heart Disease. Stroke etiopathogenesis of the cohort according to TOAST classification. Longitudinal variations of PI and cognitive assessments throughout the follow-up period MCA PI (p=0.630) and PCA PI (p=0.086) did not differ from baseline (T1) to 6 months (T3) (Fig. 3). A significant change in MoCA scores was observed between baseline (T1) and the 12-month-follow-up (T4) (p<0.001), whereas adjusted MMSE scores showed a trend toward significance (p=0.055) (Table 2).Fig. 3MCA and PCA PI at baseline (T1) and at 6-months (T3) follow up.Fig 3 dummy alt textTable 2Longitudinal changes in MCA and PCA PI from T1 to T3 and Adjusted MMSE and MoCA from T1 to T4.Table 2 dummy alt textPI and Cognitive assessmentsT1T2T3T4pMCA PI, median (IQR)1.1 (0.3)1.1 (0.4)0.630PCA PI, median (IQR)1.3 (0.5)1.3 (0.6)0.086Adjusted MMSE, median (IQR)27.9 (3.7)28.6 (3.3)29.5 (3.3)29.3 (2.8)0.055MoCA, median (IQR)26.5 (4)27 (3)27 (4)27 (4)p<0.001 MCA and PCA PI at baseline (T1) and at 6-months (T3) follow up. Table 2Longitudinal changes in MCA and PCA PI from T1 to T3 and Adjusted MMSE and MoCA from T1 to T4.Table 2 dummy alt textPI and Cognitive assessmentsT1T2T3T4pMCA PI, median (IQR)1.1 (0.3)1.1 (0.4)0.630PCA PI, median (IQR)1.3 (0.5)1.3 (0.6)0.086Adjusted MMSE, median (IQR)27.9 (3.7)28.6 (3.3)29.5 (3.3)29.3 (2.8)0.055MoCA, median (IQR)26.5 (4)27 (3)27 (4)27 (4)p<0.001 Longitudinal changes in MCA and PCA PI from T1 to T3 and Adjusted MMSE and MoCA from T1 to T4. Association of PI at T1 with longitudinal changes in MoCA and MMSE scores after adjustment for baseline cognitive performance and stroke severity The improvement of MoCA scores from T1 to T4 was significantly influenced by MCA PI ≤ 1.1 at T1 (p=0.014), MoCA at T1 (p<0.001) and NIHSS at T1 (p=0.037) (Table 3). Conversely, the group of patients with MCA PI > 1.1 at T1 shows lower MoCA scores at T4 (Fig. 4).Table 3Multivariable generalized linear model for MoCA score at T4, including PI category, NIHSS at T1, and baseline MoCA (T1).Table 3 dummy alt textParameterEstimateStd. Error95% Wald CI (Lower)95% Wald CI (Upper)Wald Chi-squarep-valuePI≤1.1-0.0470.0193-0.085-0.0095.9990.014NIHSS at T1-0.0120.0057-0.023-0.0014.3420.037MoCA at (T1)0.0290.00280.0240.035107.978<0.001Dependent variable: MoCA at T4.Model: Intercept, PI ≤1.1, NIHSS at T1, MoCA at T1.Fig. 4Cognitive performance expressed by MoCA in the two groups of patients with MCA PI ≤ 1.1 (blue line) and MCA PI > 1.1 (red line) considering PI at T1.Fig 4 dummy alt text Table 3Multivariable generalized linear model for MoCA score at T4, including PI category, NIHSS at T1, and baseline MoCA (T1).Table 3 dummy alt textParameterEstimateStd. Error95% Wald CI (Lower)95% Wald CI (Upper)Wald Chi-squarep-valuePI≤1.1-0.0470.0193-0.085-0.0095.9990.014NIHSS at T1-0.0120.0057-0.023-0.0014.3420.037MoCA at (T1)0.0290.00280.0240.035107.978<0.001Dependent variable: MoCA at T4.Model: Intercept, PI ≤1.1, NIHSS at T1, MoCA at T1. Multivariable generalized linear model for MoCA score at T4, including PI category, NIHSS at T1, and baseline MoCA (T1). Dependent variable: MoCA at T4. Model: Intercept, PI ≤1.1, NIHSS at T1, MoCA at T1. Cognitive performance expressed by MoCA in the two groups of patients with MCA PI ≤ 1.1 (blue line) and MCA PI > 1.1 (red line) considering PI at T1. Differently, the change in adjusted MMSE from T1 to T4 was significantly associated with NIHSS at T1 (p=0.042) and adjusted MMSE at T1 (p<0.001), with a trend toward significance for MCA PI ≤ 1.1 (p=0.073) (Table 4).Table 4Multivariable generalized linear model for adjusted MMSE score at T4, including PI category, NIHSS at T1, and baseline adjusted MMSE (T1).Table 4 dummy alt textParameterEstimateStd. Error95% Wald CI (Lower)95% Wald CI (Upper)Wald Chi-squarep-valuePI≤1.1-0.0270.0150-0.0560.0023.2250.073NIHSS at T1-0.0090.0044-0.0180.0004.1510.042Adjusted MMSE at T10.0130.00310.0070.01916.517<0.001Dependent variable: Adjusted MMSE at T4.Model: Intercept, PI≤1.1, NIHSS at T1, Adjusted MMSE at T1. Table 4Multivariable generalized linear model for adjusted MMSE score at T4, including PI category, NIHSS at T1, and baseline adjusted MMSE (T1).Table 4 dummy alt textParameterEstimateStd. Error95% Wald CI (Lower)95% Wald CI (Upper)Wald Chi-squarep-valuePI≤1.1-0.0270.0150-0.0560.0023.2250.073NIHSS at T1-0.0090.0044-0.0180.0004.1510.042Adjusted MMSE at T10.0130.00310.0070.01916.517<0.001Dependent variable: Adjusted MMSE at T4.Model: Intercept, PI≤1.1, NIHSS at T1, Adjusted MMSE at T1. Multivariable generalized linear model for adjusted MMSE score at T4, including PI category, NIHSS at T1, and baseline adjusted MMSE (T1). Dependent variable: Adjusted MMSE at T4. Model: Intercept, PI≤1.1, NIHSS at T1, Adjusted MMSE at T1. Association of PI at T3 with longitudinal changes in MoCA and MMSE scores from T3 to T4 after adjustment for cognitive performance at T3 and stroke severity When assessing the change in adjusted MMSE from T3 to T4 a significant association with MCA PI at T3 ≤ 1.1 (p=0.004) and adjusted MMSE at T3 (p<0.001) was found while a non-significant trend toward an association was observed for NIHSS at T1 (p=0.087) (Table 5).Table 5Multivariable generalized linear model for adjusted MMSE score at T4, including PI category, NIHSS at T1, and adjusted MMSE (T3).Table 5 dummy alt textParameterEstimateStd. Error95% Wald CI (Lower)95% Wald CI (Upper)Wald Chi-squarep-valueNIHSS at T1-0.0070.0041-0.0150.0012.9380.087PI≤1.1-0.0390.0135-0.065-0.0138.3770.004Adjusted MMSE at T30.0180.00330.0120.02430.672<0.001Dependent variable: MMSE at T4.Model: Intercept, NIHSS at T1, PI≤1.1, Adjusted MMSE at T3. Table 5Multivariable generalized linear model for adjusted MMSE score at T4, including PI category, NIHSS at T1, and adjusted MMSE (T3).Table 5 dummy alt textParameterEstimateStd. Error95% Wald CI (Lower)95% Wald CI (Upper)Wald Chi-squarep-valueNIHSS at T1-0.0070.0041-0.0150.0012.9380.087PI≤1.1-0.0390.0135-0.065-0.0138.3770.004Adjusted MMSE at T30.0180.00330.0120.02430.672<0.001Dependent variable: MMSE at T4.Model: Intercept, NIHSS at T1, PI≤1.1, Adjusted MMSE at T3. Multivariable generalized linear model for adjusted MMSE score at T4, including PI category, NIHSS at T1, and adjusted MMSE (T3). Dependent variable: MMSE at T4. Model: Intercept, NIHSS at T1, PI≤1.1, Adjusted MMSE at T3. Conversely, the group of patients with MCA PI > 1.1 at T3 shows lower MMSE scores at T4 (Fig. 5).Fig. 5Cognitive performance expressed by adjusted MMSE in the two groups of patients with MCA PI ≤ 1.1 (blue line) and MCA PI > 1.1 (red line) considering PI at T3.Fig 5 dummy alt text Cognitive performance expressed by adjusted MMSE in the two groups of patients with MCA PI ≤ 1.1 (blue line) and MCA PI > 1.1 (red line) considering PI at T3. When assessing the change in MoCA from T3 to T4 a significant association with MoCA at T3 (p<0.001) was found while a non-significant trend toward an association was observed for MCA PI at T3 ≤ 1.1 (p=0.082) (Table 6).Table 6Multivariable generalized linear model for MoCA at T4, including PI category, NIHSS at T1, and MoCA at T3.Table 6 dummy alt textParameterEstimateStd. Error95% Wald CI (Lower)95% Wald CI (Upper)Wald Chi-squarep-valueNIHSS at T1-0.0050.0046-0.0140.0041.1070.293PI≤1.1-0.0260.0152-0.0560.0033.0320.082MoCA at T30.0390.00210.0340.043335.954<0.001Dependent variable: MoCA at T4.Model: Intercept, NIHSS at T1, PI≤1.1, MoCA at T3. Table 6Multivariable generalized linear model for MoCA at T4, including PI category, NIHSS at T1, and MoCA at T3.Table 6 dummy alt textParameterEstimateStd. Error95% Wald CI (Lower)95% Wald CI (Upper)Wald Chi-squarep-valueNIHSS at T1-0.0050.0046-0.0140.0041.1070.293PI≤1.1-0.0260.0152-0.0560.0033.0320.082MoCA at T30.0390.00210.0340.043335.954<0.001Dependent variable: MoCA at T4.Model: Intercept, NIHSS at T1, PI≤1.1, MoCA at T3. Multivariable generalized linear model for MoCA at T4, including PI category, NIHSS at T1, and MoCA at T3. Dependent variable: MoCA at T4. Model: Intercept, NIHSS at T1, PI≤1.1, MoCA at T3. MCA PI and PCA PI were correlated with each other both at baseline and at the 6-month follow-up (p<0.001). Therefore, we did not consider it meaningful to perform the same longitudinal analyses for PCA PI and cognitive scores, given that any apparent effect of the PCA would likely reflect a statistical artifact driven by its correlation with the MCA. Influence of vascular risk factors on Pulsatility Index The multiple regression analysis revealed that age and diabetes were the main risk factors affecting MCA PI at T1 (R2=0.452; p<0.0001) (Supplemental Table 1), while age was the sole factor influencing PCA PI at T1 (R2=0.279; p<0.0001) (Supplemental Table 2). We also performed a regression analysis to understand if the single vascular risk factors could influence MCA PI and PCA PI at baseline (T1). We found a positive correlation between MCA PI and hypertension (R2=0.078; p=0.002) and between MCA PI and antihypertensive therapy at baseline (R2=0.083; p=0.001). We also found a positive correlation between MCA PI and anti-diabetic drugs at baseline (R2=0.232; p<0.0001), between MCA PI and dyslipidemia (R2=0.034; p=0.041), and between MCA PI and atrial fibrillation at baseline (R2=0.070; p=0.003). We observed a positive correlation between PCA PI and hypertension (R2=0.055; p=0.008) and between PCA PI and antihypertensive therapy at baseline (R2= 0.041; p=0.023). We also found a positive correlation between PCA PI and diabetes (R2=0.062; p=0.005) and between PCA PI and anti-diabetic drugs at baseline (R2=0.064; p=0.005). We also found a positive correlation between PCA PI and dyslipidemia (R2=0.047; p=0.015) and between PCA PI and atrial fibrillation (R2=0.047; p=0.016) at baseline. Pulsatility Index in stroke subgroups We also explored whether MCA and PCA PI at baseline, adjusted for age, were different in subgroups of stroke according to TOAST classification. We observed no significant differences in MCA and PCA PI at baseline between several groups (p=0.413 and p=0.234) (Fig. 6).Fig. 6MCA and PCA PI at baseline (T1) adjusted for age in several groups of stroke/TIA patients according to TOAST classification.Fig 6 dummy alt text MCA and PCA PI at baseline (T1) adjusted for age in several groups of stroke/TIA patients according to TOAST classification. Discussion This study investigated longitudinal changes in cerebral hemodynamics and cognitive outcomes after stroke/TIA. While MCA and PCA PI remained stable over time, cognitive performance improved during follow-up, particularly in MoCA scores. Overall, lower cerebral vascular resistance, together with less severe stroke and better baseline cognition, was associated with more favorable cognitive outcomes. Conversely, higher PI predicted worse cognitive performance. These findings suggest that cerebral hemodynamics may play a role in determining long-term cognitive outcome after stroke. The primary objective of this study was to investigate longitudinal changes in cerebral hemodynamics, assessed using MCA and PCA PI, along with cognitive outcomes over the follow-up period. MCA and PCA PI remained stable from baseline (T1) to 6 months (T3) while cognitive performance improved from T1 to 12 months (T4). MoCA scores increased significantly, whereas MMSE scores showed a positive but non-significant trend. As most prior studies have focused on PI in the acute stroke phase, these findings provide novel insight into longer-term hemodynamic patterns. The observed cognitive improvement aligns with longitudinal studies reporting recovery of global cognitive function after stroke, particularly when assessed with MoCA [26]. Previous studies have reported that most improvement occurs within the first three months, although cognitive gains may continue throughout the first year, as observed in our study [27]. In contrast, MMSE changes are often smaller and less sensitive to post-stroke cognitive recovery, especially in executive and visuospatial domains [28]. Overall, the significant increase in MoCA scores and the positive trend in MMSE scores in our cohort are consistent with evidence of gradual post-stroke cognitive improvement. We also investigated whether PI predicts cognitive outcomes. Improvement in MoCA scores from T1 to T4 was independently associated with MCA PI ≤ 1.1 at T1, indicative of lower vascular resistance, MoCA, and NIHSS scores at T1. In contrast, MCA PI > 1.1 at T1, indicative of higher vascular resistance, was linked to poorer cognitive performance, reflected by lower MoCA scores at T4. Similar associations were observed for MMSE, with baseline cognitive status and NIHSS as significant predictors and a trend toward better outcomes with lower MCA PI. Notably, lower MCA PI at T3 with MMSE at T3 independently predicted MMSE improvement from T3 to T4, whereas higher MCA PI at T3 was associated with lower MMSE scores at T4. Improvement in MoCA from T3 to T4 was primarily driven by MoCA at T3, with a trend toward greater improvement in patients with lower MCA PI at T3. Evidence on PI as a prognostic marker for post-stroke cognition remains limited. A recent study has linked TCD-derived hemodynamic measures, including neurovascular coupling, to long-term cognition after stroke, although it did not demonstrate a specific role for PI [29]. MCA PI has also been identified as a predictor of incident cognitive impairment in subclinical asymptomatic cervico-intracranial stenosis [30]. To our knowledge, no studies have shown that lower PI predicts long-term post-stroke cognition. In cognitively healthy older adults, lower resting PI has been linked to better fluid cognitive performance, suggesting a role for preserved hemodynamics [31]. Our findings extend this evidence by indicating that lower MCA PI may be associated with a more favorable post-stroke cognitive trajectory, supporting PI as a potential hemodynamic prognostic biomarker. This is, to our knowledge, the first study to assess cerebral hemodynamics both in the acute phase and at 6 months after stroke/TIA in relation to long-term cognition. Baseline and 6-month PI both predicted 12-month cognitive outcomes, supporting its prognostic value over time. In our cohort, baseline cognition independently predicted longitudinal change, consistent with previous findings that early post-stroke cognitive status is a strong predictor of long-term outcomes [32], likely reflecting cognitive reserve. Lower NIHSS scores were also associated with better cognitive trajectories, consistent with studies in minor stroke populations, although baseline cognition remains the strongest predictor of long-term outcome [33]. Despite limited evidence on PI as a prognostic marker of cognitive outcomes after stroke, several studies have reported an association between elevated PI and cognitive impairment, particularly in patients with Alzheimer’s disease, dementia, vascular dementia, and lacunar infarcts [34,35]. PI reflects cerebral vascular resistance and may serve as a marker of SVD, with elevated PI identified as an independent predictor of SVD [17]. To our knowledge, the relationship between PI and cognitive impairment has not been investigated in stroke subtypes beyond those related to SVD. Although SVD accounted for 49% of cases in our cohort, the association between higher PI and poorer cognitive performance at 12 months was observed across all stroke subtypes [24]. Moreover, MCA and PCA PI values did not differ according to stroke etiology. We then explored the association between increased PI and cognitive impairment. PI reflects cerebral vascular resistance, and higher resistance reflects limited capability of the cerebral vessels to damp the pulsatile cerebral blood flow, resulting in increased pulsatility [36]. Structural changes associated with SVD, including lipohyalinosis and microatherosclerosis can narrow small vessels and increase downstream resistance. In addition, a thickened fibrotic wall may lose its elastic properties, becoming rigid and unable to adequately dilate and constrict as needed. These mechanisms suggest that factors affecting vascular resistance may contribute to elevated PI. Therefore, we wondered which factors could influence PI and affect the ability of the vessels to dilate and constrict adequately. Consequently, we performed a multiple regression analysis to explore which vascular risk factors could influence the MCA and PCA PI at baseline. We found that age and diabetes were the main determinants of MCA PI at baseline whereas age was the only factor influencing PCA PI. We also examined the relationship between vascular risk factors and MCA and PCA PI. MCA/PCA PI was positively correlated with hypertension and the use of antihypertensive medication at baseline. Both intracranial hypertension and chronic arterial hypertension have been associated with increased PI [18]. Previous studies have shown that intensive antihypertensive treatment can improve cerebral blood flow [[37], [38], [39]]. Furthermore, a meta-analysis of vasodilator therapies reported that, despite significantly lowering systolic blood pressure, these agents did not reduce absolute cerebral blood flow and were associated with improved PI [40]. MCA/PCA PI was also positively correlated with diabetes and the use of antidiabetic drugs at baseline. An increase in PI has been reported in diabetes, a major risk factor for ischemic stroke. Diabetes-related vascular pathology includes both macroangiopathy and microangiopathy affecting cerebral and systemic vessels [41]. Histopathological studies in diabetic patients and experimental models have demonstrated structural abnormalities in small cerebral vessels, including endothelial cell necrosis and thickening of capillary basement membranes [42,43]. In support of this evidence, the use of some antidiabetic drugs (sodium-glucose cotransporter-2 inhibitor, GLP-1R, and DPP-IV) has been shown to improve cerebral blood flow in diabetic animal models [[44], [45], [46]]. We further observed a positive correlation between MCA/PCA PI and dyslipidaemia. Although cerebral hemodynamics has been investigated in hyperlipidaemic patients using transcranial Doppler, previous studies found no significant differences in cerebral blood flow velocity or PI compared with controls [47]. Finally, MCA/PCA PI was positively correlated with atrial fibrillation at baseline. To our knowledge, there are no data on the relationship between persistent atrial fibrillation and PI. Some studies have examined cerebral hemodynamics during and after paroxysmal atrial fibrillation in small patient cohorts, reporting a reduction in PI across all arteries, although these findings did not reach statistical significance [48]. This study has several limitations. First, its single-center design may limit the generalizability of the findings. Second, the observational nature of the study precludes causal inferences. Therefore, larger multicenter studies are warranted to validate our findings. Finally, although standardized correction procedures were applied to the cognitive test scores, the significant association between age and MCA and PCA PI, as demonstrated by our multiple regression analysis, may still be reflected in patients' cognitive performance. Previous studies have shown that PI increases with ageing, diabetes mellitus, hypertension, SVD and vascular dementia [12,13,17,19]. In this scenario we envisage that in stroke patients damage to the vascular wall secondary to hypertension, diabetes, physiological ageing processes or endothelial dysfunction may lead to increased PI. All these mechanisms may alter cerebral hemodynamics, thus causing cerebral hypoperfusion and, consequently, impaired cognitive performance. Conversely, patients with low vascular resistance and low PI exhibit more favorable hemodynamics, with improved cerebral perfusion and, consequently, better cognitive performance. In conclusion, our findings suggest that MCA and PCA PI values in the acute phase and at 6 months may represent prognostic markers of cognitive outcome in patients with TIA and ischemic stroke with mild symptoms. Data availability statement Anonymized data will be shared on request from any qualified investigator. Sources of funding Study costs were covered by the Neurosonology Unit of Fondazione Policlinico Universitario Campus Bio-Medico. CRediT authorship contribution statement Nicoletta Brunelli: Writing – original draft, Methodology, Investigation, Formal analysis, Conceptualization. Claudia Altamura: Writing – review & editing, Methodology, Formal analysis. Marilena Marcosano: Writing – review & editing, Investigation. Carlo Vico: Writing – review & editing, Investigation. Marcel Bach-Pages: Writing – review & editing, Formal analysis. Luigi d’Onofrio: Writing – review & editing, Investigation. Sergio Soeren Rossi: Writing – review & editing, Investigation. Carmelina Maria Costa: Writing – review & editing, Investigation. Simona Sacco: Writing – review & editing, Methodology. Mauro Silvestrini: Writing – review & editing, Methodology. Fabrizio Vernieri: Writing – review & editing, Methodology. Declaration of competing interest The authors have no disclosures.