Peripheral Blood Regulatory T Cells and IL-6 are Associated with Cognitive Impairment After Acute Ischemic Stroke Authors: Yueyue Chang, Xuke Zhang, Chuanqing Yu, Lei Zhu, Shiyu Sun, Zhongwu Sun Published: 2026 ABSTRACT BackgroundAbstract Objective: Regulatory T cells (Tregs) and interleukin (IL)-6 are known to be involved in the progression of cognitive impairment but the association with post-stroke cognitive impairment remains little studied. Levels of Tregs and IL-6 in peripheral blood serum of patients with acute ischemic stroke (AIS) were investigated to assess the association with post-stroke cognitive impairment (PSCI).MethodsA total of 256 AIS patients were enrolled with 19 lost to follow-up after six months. Participants were divided into those without post-stroke cognitive impairment (PSNCI, n=129) and those with post-stroke cognitive impairment (PSCI, n=108). PSCI patients were subdivided into those with post-stroke cognitive impairment, no dementia (PSCIND, n=87) and those with post-stroke dementia (PSD, n=21). The percentage of peripheral blood CD4+ CD25+ CD127 (low) regulatory T cells (Tregs%) out of total CD4+ lymphocytes was evaluated by flow cytometry. IL-6 was measured by enzyme-linked immunosorbent assay (ELISA).Results(1) Increased percentages of peripheral blood Tregs and IL-6 levels were found in PSCI, PSCIND and PSD patients relative to PSNCI and differences were particularly significant for PSCI and PSD groups. (2) Logistic regression analysis was performed after controlling for these risk factors and Tregs% (OR=1.122, 95% CI: 1.006–1.251, p=0.038) identified as an independent risk factor for cognitive impairment after AIS. ROC analysis gave an area under the curve (AUC) of 0.763 (95% CI: 0.653–0.873) for the combination of Tregs% and IL-6 levels in predicting cognitive impairment after AIS.(3) IL-6 level (OR=1.307, 95% CI: 1.058–1.614, p=0.013) was found to be an independent risk factor for dementia after AIS with an AUC of 0.805 (95% CI: 0.681–0.929).ConclusionElevated peripheral Tregs% and IL-6 levels are risk factors for cognitive impairment after AIS. Their combined predictive power is greater than individual predictive power, exposing new therapeutic targets for post-stroke cognitive impairment. Introduction Acute ischemic stroke (AIS) is a cerebrovascular condition that accounts for high rates of disability and death. About one-third of stroke patients experience cognitive decline or post-stroke cognitive impairment (PSCI) and post-stroke dementia (PSD), creating substantial family burdens.1 AIS activates microglia and inflammatory factors, initiating neuroinflammatory responses, and increases the permeability of the blood-brain barrier (BBB), promoting PSCI progression.2 Tregs have been suggested as early biomarkers of cognitive impairment and may be involved in the dysregulation responsible. The adoptive transfer of Tregs improved cognitive deficits in animal models of Alzheimer’s disease (AD), indicating the potential for Treg cell therapy.3 T cell receptor (TCR)-engineered Tregs decreased amyloid-beta (Aβ) accumulation in brain tissue and enhanced behavioral outcomes in the APP/PS1 AD mouse model, illustrating that TCR Aβ-specific Tregs may improve cognitive function. TCR Aβ-Tregs may regulate Aβ levels and neuroinflammation by modulating the immune environment, reversing neurotoxicity and enhancing cognition.4 Studies in AD patients have shown reduced neuroinflammation and improved cognitive abilities after transplant of Tregs functionally enhanced by in vitro expansion.5 Research on roles of Tregs in post-stroke cognitive impairment remains limited, although an imbalance of Th17 and Treg cells has been suggested to exacerbate cognitive decline.6 Tregs may function as memory cells identified by brain cells, reducing neuroinflammation, supporting repair of white matter and accelerating post-stroke neurological recovery.7 IL-6 is a pro-inflammatory cytokine involved in the progression of cognitive decline. It causes chronic neuroinflammatory responses within the central nervous system by an effect on the permeability of the blood-brain barrier (BBB), which impairs neuronal function and promotes the deterioration of cognitive abilities.8 Elevated serum IL-6 has been linked to cognitive decline and reduced executive function in older adults, substantiating the role of neuroinflammation. Indeed, IL-6 and the pro-inflammatory tumor necrosis factor-alpha (TNF-α) have both been implicated in the pathological inflammation linked to cognitive impairment.9 Patients with IL-6 levels in the highest quartile of a 1003 patient cohort with AIS and transient ischemic attack (TIA) had a greater risk of cognitive decline compared with those in the lowest quartile (25.90%vs16.80%).10 Patients experiencing cognitive impairment after stroke are generally considered to have increased IL-6, which may be a predictive marker for this dysfunction.11 Neuroinflammation is acknowledged to be involved in PSCI development, although mechanisms remain unclear. The causes of PSCI require urgent clarification to enable targeted intervention strategies. Tregs and IL-6 play crucial roles in modulating inflammation after stroke. Recent studies have shown that amphiregulin (AREG) regulates the IL-6 and STAT3 signaling pathway in microglia and astrocytes, which is a critical aspect of the neuroprotective function of brain Tregs.12 Additionally, excessive IL-6 production inhibits Foxp3 expression and Tregs development.13 However, Nevertheless, it is still unclear whether Tregs’ effects related to IL-6 contribute to PSCI, and if so, through which mechanisms. The relationship between peripheral blood Tregs, IL-6 and cognitive impairment in AIS patients was investigated during the current work to allow evaluation of Treg percentage and IL-6 as biomarkers of PSCI to explore its underlying mechanism. The aim was to contribute to immunotherapeutic approaches for PSCI. Materials and Methods Study Design and Subjects A cohort of 256 patients, diagnosed with AIS and admitted to the Department of Neurology at the First Affiliated Hospital of Anhui University of Science and Technology (Huainan First People’s Hospital) between January 2023 and October 2024, were enrolled in the current retrospective and single-center study. Patients with acute cerebral infarction underwent neuropsychological evaluations using the Montreal Cognitive Assessment (MoCA) and Clinical Dementia Rating (CDR) scales and clinical examination within one week of stroke onset. The blood sample was collected within 24 hours of admission. Follow-up assessments using the same scales were conducted six months post-stroke. Patients were classified into four groups of post-stroke no cognitive impairment (PSNCI; n=129), post-stroke cognitive impairment (PSCI; n=108), post-stroke cognitive impairment with no dementia (PSCIN; n=87) and post-stroke dementia (PSD; n=21). Inclusion Criteria: Diagnosis of AIS confirmed by diffusion-weighted imaging (DWI) within 72 hours of symptom onset;Written informed consent obtained:Aged between 40 and 88 years;Ability to independently participate in all neuropsychological assessments;Pre-stroke modified Rankin Scale (mRS) score of 2 or less. Diagnosis of AIS confirmed by diffusion-weighted imaging (DWI) within 72 hours of symptom onset; Diagnosis of AIS confirmed by diffusion-weighted imaging (DWI) within 72 hours of symptom onset; Written informed consent obtained: Written informed consent obtained: Aged between 40 and 88 years; Aged between 40 and 88 years; Ability to independently participate in all neuropsychological assessments; Ability to independently participate in all neuropsychological assessments; Pre-stroke modified Rankin Scale (mRS) score of 2 or less. Pre-stroke modified Rankin Scale (mRS) score of 2 or less. Exclusion Criteria: Presence of cognitive impairment prior to stroke onset;Cognitive dysfunction attributable to other neurological or systemic conditions associated with cognitive decline, including Parkinson’s disease, multiple system atrophy, metabolic disorders, non-vascular white matter disease, neurosyphilis and psychiatric illnesses;Systemic illnesses, such as malignancies, hematologic disorders or autoimmune diseases;Inability to cooperate with neuropsychological assessments;History of severe stroke-related sequelae;Severe cardiac, hepatic or renal dysfunction or metabolic diseases within the preceding three months;Refusal to provide informed consent or unwillingness to comply with study procedures and follow-up assessments. Presence of cognitive impairment prior to stroke onset; Presence of cognitive impairment prior to stroke onset; Cognitive dysfunction attributable to other neurological or systemic conditions associated with cognitive decline, including Parkinson’s disease, multiple system atrophy, metabolic disorders, non-vascular white matter disease, neurosyphilis and psychiatric illnesses; Cognitive dysfunction attributable to other neurological or systemic conditions associated with cognitive decline, including Parkinson’s disease, multiple system atrophy, metabolic disorders, non-vascular white matter disease, neurosyphilis and psychiatric illnesses; Systemic illnesses, such as malignancies, hematologic disorders or autoimmune diseases; Systemic illnesses, such as malignancies, hematologic disorders or autoimmune diseases; Inability to cooperate with neuropsychological assessments; Inability to cooperate with neuropsychological assessments; History of severe stroke-related sequelae; History of severe stroke-related sequelae; Severe cardiac, hepatic or renal dysfunction or metabolic diseases within the preceding three months; Severe cardiac, hepatic or renal dysfunction or metabolic diseases within the preceding three months; Refusal to provide informed consent or unwillingness to comply with study procedures and follow-up assessments. Refusal to provide informed consent or unwillingness to comply with study procedures and follow-up assessments. Evaluation of Cognitive Impairment Two to three neurologists with expertise in cognitive assessment scales conducted face-to-face evaluations of all eligible AIS patients using the MoCA and CDR scales. Assessments were made in a quiet environment and scheduled at times when patients were deemed to be in an optimal state to complete the evaluations. The same cognitive assessments were repeated six months after patient discharge. Patients with a MoCA score < 22 were classified as having PSCI.14 Patients with CDR scores between 0.5 and 1 were categorized as having PSCIND15 and those with scores between 2 to 3 as having PSD. Imaging Assessment All participants underwent detailed cranial magnetic resonance imaging (MRI) using either 1.5T or 3.0T scanners (made in the United States). The imaging protocol included T1-weighted, T2-weighted, diffusion-weighted imaging (DWI) and fluid-attenuated inversion recovery (FLAIR) sequences, complemented by vascular imaging of the head and neck. White Matter Hyperintensities (WMH) WMH16 were evaluated by modified Fazekas grading scale (range: 0 to 6 points) applied to T2-weighted or FLAIR MRI scans with separate assessment of periventricular and deep WMH. Lesions involving deep gray WMH were classified as subcortical hyperintensities to align with the broader definition of subcortical regions, similar to recent subcortical small infarcts. Individual scores from the two regions were aggregated to yield a total score which was categorized into three grades of mild (0–2 points), moderate (3–4 points) or severe (5–6 points). Vital Brain Infarction (VBI) Infarctions located in the left thalamus, left frontotemporal lobe and right parietal lobe were designated as vital brain infarction (VBI)17 from DWI findings. Infarct volume in AIS patients was quantified using the Pullicito18 formula and dichotomized into volumes ≥5 cm3 or <5 cm3. Stroke subtypes were categorized according to TOAST classification, including large artery atherosclerosis (LAA), cardioembolic infarction (CE), small artery occlusion (SAO), stroke of other determined etiology (SOE) and stroke of undetermined etiology (SUE). SOE and SUE were combined into a single category of other types (OT) in the current analysis. AIS was classified according to the Oxfordshire Community Stroke Project (OCSP) criteria into total anterior circulation infarction (TACI), partial anterior circulation infarction (PACI), posterior circulation infarction (POCI) and lacunar infarction (LACI). Brain atrophy was assessed by Global Cortical Atrophy (GCA) scale (range: grade 0–3),19 reflecting the severity of cortical atrophy across the entire brain. Two grades of brain atrophy were considered: mild (0–1 points) or moderate to severe (2–3 points). Measurement of Peripheral Blood Tregs and IL-6 Mouse anti-human monoclonal antibodies targeting CD4-PC, CD3-PC, CD8-PC, CD45-PC and CD25-PE-CY7C were purchased from Beckman Coulter and used to analyze samples of 2mL peripheral whole blood anticoagulated with EDTA, at room temperature and usually processed within 24 hours. Samples that were processed after 24 hours were stored at 4–8°C. During the sample preparation, a clean test tube was used to which 100 µL of whole blood specimen was added. Subsequently, fluorescently labeled monoclonal antibodies (CD4-PE, CD127-PE-CY5, CD25-PE-CY7) were introduced. The mixture was gently vortexed and incubated at room temperature in the dark for 15 minutes. 300 µL of BC OptiLyse C lysing solution was added to mix thoroughly by vortexing in the dark for 5 minutes. The sample was then centrifuged at 1700 rpm, the supernatant discarded, and the pellet washed with 2 mL of PBS by centrifugation at 1700 rpm. After discarding the supernatant, 600 µL of PBS was added to resuspend the cells, and the sample was mixed thoroughly prior to analysis. Tregs were detected by flow cytometry (Beckman Coulter, USA) and data acquired by Beckman CXP software and analyzed by Beckman Kaluza software to quantify proportions of CD4+CD25++CD127low regulatory T cells within the total CD4+ lymphocyte population (expressed as Tregs%). IL-6 was quantified by enzyme-linked immunosorbent assay (ELISA) kit (Siemens, UK). Statistical Analysis Data were processed using SPSS version 26.0, and graphical presentations were created using GraphPad Prism 9.5.0. Variable normality and homogeneity of variance were assessed. Continuous variables with normal distribution are reported as mean ± standard deviation and intergroup comparisons were conducted by one-way analysis of variance (ANOVA). Categorical variables are presented as frequencies and percentages [n (%)] with intergroup differences evaluated by chi-square test. Non-normally distributed continuous variables are expressed as median (first quartile, third quartile) [M (Q1, Q3)] and intergroup comparisons were performed by Mann–Whitney U-test. Univariate analyses followed by binary logistic regression were utilized to identify risk factors associated with the development of cognitive impairment after AIS. Diagnostic accuracy for post-stroke cognitive impairment was assessed by receiver operating characteristic (ROC) curve analysis. Statistical significance was defined as a p-value < 0.05. Results 256 AIS patients were initially enrolled, 19 of whom were lost during the six-month follow-up, leaving 237 patients with complete clinical data and successful follow-up. These patients were categorized into 108 PSCI patients and 129 PSNCI patients. The first group was subdivided into 87 patients with PSCIND and 21 patients with PSD. Baseline clinical characteristics, including demographic variables, cerebrovascular risk factors and hematological parameters, are summarized in Table 1. Statistically significant differences were found between PSNCI and PSCI patients in age (p = 0.048), uric acid (UA) levels (p = 0.031), presence of diabetes and admission NIHSS scores (p < 0.001). Significant PSNCI vs. PSCIND differences were also found in the prevalence of coronary heart disease (p = 0.027) and admission NIHSS scores (p < 0.001) and PSNCI vs. PSD differences in age (p = 0.017), diabetes status (p = 0.031) and admission NIHSS scores (p < 0.001, Table 1).Table 1Clinical and Demographic Patient CharacteristicsParametersPSNCI (n=129)PSCI (n=108)PSCIND (n=87)PSD (n=21)p-valuePSNCI vs PSCIPSNCI vs PSCINDPSNCI vs PSDDemographic characteristicsAge, year62.20±10.8667.15±11.2363.95±10.9269.84±11.000.0480.5890.017Sex (male, n, %)72(55.81)49(45.37)40(45.98)9(42.85)0.1090.1560.269Educational level8.41±2.427.72±2.038.93±3.417.98±2.560.1200.3740.269Vascular risk factorsSBP (mmHg)140(121,154)143(123,160)136(129,145)145(133,155)0.8230.5780.146DBP (mmHg)86 (77, 99)84(72,95)87(70,91)90(75,100)0.5260.8730.126Diabetes, n (%)37(28.68)47(43.51)36(41.38)11(52.38)0.0170.0530.031Coronary heart disease, n (%)27(20.93)34(31.48)30(34.48)4(19.05)0.0640.0270.843Stroke (n %)21(16.28)22(20.37)16(18.39)6(28.57)0.4350.6860.174Hypertension (n, %)73(56.59)49(45.37)39(44.83)10(47.62)0.0850.0900.443Smoking, n (%)51(39.53)30(27.78)25(28.74)5(23.81)0.0570.1030.167Alcohol consumption, n (%)43(33.33)32(29.63)28(25.93)4(19.05)0.5410.8600.191Admission NIHSS scores4.74±2.318.58±3.125.68±2.2410.87±4.60<0.001<0.001<0.001Laboratory dataWBC (×109/L)6.03 (5.14, 7.61)5.94(4.87,7.30)6.14(4.03,7.37)6.37(5.21,7.84)0.7310.8290.594PLT (×109/L)226(147,242)203(152,274)237(168,302)213(180,239)0.4520.2120.347HB (g/L)138(119,147)131(120,145)136(117,143)129(112,138)0.5340.8740.932GLU (mmol·L−1)6.89±3.526.44±2.596.53±2.546.36±2.680.5440.7110.550TC (mmol·L−1)4.47±1.074.43±1.024.57±1.014.28±1.010.8000.7760.513TG (mmol·L−1)1.53(1.07,1.89)1.28(0.95,1.88)1.37(1.03,2.23)1.18±(0.89,1.1)0.5590.5540.572HDL (mmol·L−1)1.03±0.331.07±0.321.10±0.261.05±0.350.5940.4340.964LDL (mmol·L−1)3.04±1.053.09±0.973.21±0.903.03±0.960.7650.5540.846UA (μmol·L−1)317.08±68.73274.43±85.32288.62±62.58285.88±100.150.0310.0900.205HCY (μmol·L−1)11.39±5.8513.65±5.2913.26±5.2713.98±5.400.1020.2650.112IL-611.76±3.2016.25±4.7214.27±5.1217.91±5.990.0300.144<0.001Tregs%2.00±0.832.35±0.872.04±0.792.54±0.950.0140.4990.007Abbreviations: PSC, Post-stroke cognitive impairment; PSNCI, post-stroke non-cognitive impairment; PSCIND, post-stroke cognitive impairment no dementia; PSD, post-stroke dementia. WBC, Leukocyte; PLT, Platelet Count; HB, Hemoglobin; GLU, Glucose; A; UA, Uric Acid; HCY, Homocysteine; TC, Total Cholesterol; TG, Total Triglyceride; HDL-C, High-Density Lipoprotein Cholesterol; LDL-C, Low-Density Lipoprotein Cholesterol. Table 1Clinical and Demographic Patient CharacteristicsParametersPSNCI (n=129)PSCI (n=108)PSCIND (n=87)PSD (n=21)p-valuePSNCI vs PSCIPSNCI vs PSCINDPSNCI vs PSDDemographic characteristicsAge, year62.20±10.8667.15±11.2363.95±10.9269.84±11.000.0480.5890.017Sex (male, n, %)72(55.81)49(45.37)40(45.98)9(42.85)0.1090.1560.269Educational level8.41±2.427.72±2.038.93±3.417.98±2.560.1200.3740.269Vascular risk factorsSBP (mmHg)140(121,154)143(123,160)136(129,145)145(133,155)0.8230.5780.146DBP (mmHg)86 (77, 99)84(72,95)87(70,91)90(75,100)0.5260.8730.126Diabetes, n (%)37(28.68)47(43.51)36(41.38)11(52.38)0.0170.0530.031Coronary heart disease, n (%)27(20.93)34(31.48)30(34.48)4(19.05)0.0640.0270.843Stroke (n %)21(16.28)22(20.37)16(18.39)6(28.57)0.4350.6860.174Hypertension (n, %)73(56.59)49(45.37)39(44.83)10(47.62)0.0850.0900.443Smoking, n (%)51(39.53)30(27.78)25(28.74)5(23.81)0.0570.1030.167Alcohol consumption, n (%)43(33.33)32(29.63)28(25.93)4(19.05)0.5410.8600.191Admission NIHSS scores4.74±2.318.58±3.125.68±2.2410.87±4.60<0.001<0.001<0.001Laboratory dataWBC (×109/L)6.03 (5.14, 7.61)5.94(4.87,7.30)6.14(4.03,7.37)6.37(5.21,7.84)0.7310.8290.594PLT (×109/L)226(147,242)203(152,274)237(168,302)213(180,239)0.4520.2120.347HB (g/L)138(119,147)131(120,145)136(117,143)129(112,138)0.5340.8740.932GLU (mmol·L−1)6.89±3.526.44±2.596.53±2.546.36±2.680.5440.7110.550TC (mmol·L−1)4.47±1.074.43±1.024.57±1.014.28±1.010.8000.7760.513TG (mmol·L−1)1.53(1.07,1.89)1.28(0.95,1.88)1.37(1.03,2.23)1.18±(0.89,1.1)0.5590.5540.572HDL (mmol·L−1)1.03±0.331.07±0.321.10±0.261.05±0.350.5940.4340.964LDL (mmol·L−1)3.04±1.053.09±0.973.21±0.903.03±0.960.7650.5540.846UA (μmol·L−1)317.08±68.73274.43±85.32288.62±62.58285.88±100.150.0310.0900.205HCY (μmol·L−1)11.39±5.8513.65±5.2913.26±5.2713.98±5.400.1020.2650.112IL-611.76±3.2016.25±4.7214.27±5.1217.91±5.990.0300.144<0.001Tregs%2.00±0.832.35±0.872.04±0.792.54±0.950.0140.4990.007Abbreviations: PSC, Post-stroke cognitive impairment; PSNCI, post-stroke non-cognitive impairment; PSCIND, post-stroke cognitive impairment no dementia; PSD, post-stroke dementia. WBC, Leukocyte; PLT, Platelet Count; HB, Hemoglobin; GLU, Glucose; A; UA, Uric Acid; HCY, Homocysteine; TC, Total Cholesterol; TG, Total Triglyceride; HDL-C, High-Density Lipoprotein Cholesterol; LDL-C, Low-Density Lipoprotein Cholesterol. Clinical and Demographic Patient Characteristics Abbreviations: PSC, Post-stroke cognitive impairment; PSNCI, post-stroke non-cognitive impairment; PSCIND, post-stroke cognitive impairment no dementia; PSD, post-stroke dementia. WBC, Leukocyte; PLT, Platelet Count; HB, Hemoglobin; GLU, Glucose; A; UA, Uric Acid; HCY, Homocysteine; TC, Total Cholesterol; TG, Total Triglyceride; HDL-C, High-Density Lipoprotein Cholesterol; LDL-C, Low-Density Lipoprotein Cholesterol. Comparative analysis of imaging findings between PSNCI and PSCI patients demonstrated significant differences in VBI (p < 0.001) and WMH (p = 0.010) but no significant differences in infarct volume (p=0.115), TOAST (p=0.874) or brain atrophy (p=0.789). Significant PSNCI vs. PSCIND differences were identified in VBI (p<0.001) and WMH (p = 0.027) and significant PSNCI vs. PSD differences in VBI (p < 0.001), WMH (p = 0.025) and brain atrophy (p = 0.003, Table 2). Tregs% was elevated in PSCI (2.35±0.87), PSCIND (2.04±0.79) and PSD (2.54±0.95) relative to PSNCI (2.00±0.83) and significant differences were observed between the PSNCI and PSCI groups (p = 0.014) and between PSNCI and PSD (p = 0.007). IL-6 concentrations were also elevated in PSCI, PSCIND and PSD patients relative to PSNCI with significant differences in PSNCI vs. PSCI (p = 0.030) and PSNCI vs. PSD (p < 0.001).Table 2Clinical Imaging Characteristics [n (%)]ImagingPSNCI (n=129)PSCI (n=108)PSCIND (n=87)PSD (n=21)p-valuePSNCI vs PSCIPSNCI vs PSCINDPSNCI vs PSDVBI24(18.60)51(47.22)42(48.28)9(42.86)<0.001<0.001<0.001Infarct volume0.1150.1810.212<5cm380(62.02)56(51.85)46(52.87)10(47.62)≧5cm349(37.98)52(48.15)41(47.13)11(52.38)OCSP0.8990.9350.281TACI47(36.43)37(34.26)28(32.18)9(42.86)PACI23(17.83)22(20.37)17(19.54)5(23.81)POCI21(16.28)20(18.52)15(17.24)5(23.81)LACI38(29.46)29(26.85)27(31.04)2(9.52)TOAST0.8740.6770.340LAA68(52.71)56(51.85)41(47.13)15(71.43)CE27(20.93)24(22.22)20(22.99)4(19.05)SAO21(16.28)20(18.52)19(17.59)1(4.76)CE13(10.08)8(7.41)7(8.05)1(4.76)WMH0.0100.0270.025mild85(65,89)53(49.07)43(49.43)10(47.62)moderate33(25.58)33(30.56)28(32.18)5(23.81)severe11(8.53)22(20.37)16(18.39)6(28.57)Brain atrophy0.7890.8250.003Mild96(74.42)79(73.15)70(80.46)9(42.86)Moderate to Severe33(25.58)29(26.85)17(19.64)12(57.14)Abbreviations: VBI, vital brain infarction; WMH, White matter hyperintensities; LAA, large artery atherosclerosis; CE, cardioembolic infarction; SAO, small artery occlusion; OT, other types; TACI, total anterior circulation infarction; PACI, partial anterior circulation infarction; POCI, posterior circulation infarction; LACI, lacunar infarction; OCSP, Oxfordshire Community Stroke Project. Table 2Clinical Imaging Characteristics [n (%)]ImagingPSNCI (n=129)PSCI (n=108)PSCIND (n=87)PSD (n=21)p-valuePSNCI vs PSCIPSNCI vs PSCINDPSNCI vs PSDVBI24(18.60)51(47.22)42(48.28)9(42.86)<0.001<0.001<0.001Infarct volume0.1150.1810.212<5cm380(62.02)56(51.85)46(52.87)10(47.62)≧5cm349(37.98)52(48.15)41(47.13)11(52.38)OCSP0.8990.9350.281TACI47(36.43)37(34.26)28(32.18)9(42.86)PACI23(17.83)22(20.37)17(19.54)5(23.81)POCI21(16.28)20(18.52)15(17.24)5(23.81)LACI38(29.46)29(26.85)27(31.04)2(9.52)TOAST0.8740.6770.340LAA68(52.71)56(51.85)41(47.13)15(71.43)CE27(20.93)24(22.22)20(22.99)4(19.05)SAO21(16.28)20(18.52)19(17.59)1(4.76)CE13(10.08)8(7.41)7(8.05)1(4.76)WMH0.0100.0270.025mild85(65,89)53(49.07)43(49.43)10(47.62)moderate33(25.58)33(30.56)28(32.18)5(23.81)severe11(8.53)22(20.37)16(18.39)6(28.57)Brain atrophy0.7890.8250.003Mild96(74.42)79(73.15)70(80.46)9(42.86)Moderate to Severe33(25.58)29(26.85)17(19.64)12(57.14)Abbreviations: VBI, vital brain infarction; WMH, White matter hyperintensities; LAA, large artery atherosclerosis; CE, cardioembolic infarction; SAO, small artery occlusion; OT, other types; TACI, total anterior circulation infarction; PACI, partial anterior circulation infarction; POCI, posterior circulation infarction; LACI, lacunar infarction; OCSP, Oxfordshire Community Stroke Project. Clinical Imaging Characteristics [n (%)] Abbreviations: VBI, vital brain infarction; WMH, White matter hyperintensities; LAA, large artery atherosclerosis; CE, cardioembolic infarction; SAO, small artery occlusion; OT, other types; TACI, total anterior circulation infarction; PACI, partial anterior circulation infarction; POCI, posterior circulation infarction; LACI, lacunar infarction; OCSP, Oxfordshire Community Stroke Project. Binary logistic regression analysis was conducted to evaluate the association between Tregs% and PSCI. After adjustment for confounding risk factors, Tregs% (OR=1.122; 95% CI: 1.006–1.251; p=0.038) was found to be an independent risk factor for PSCI (Table 3). The AUC values for IL-6 and Tregs% in predicting cognitive impairment in PSCI patients were 0.660 and 0.718, respectively, and, when the two risk factors were combined, the AUC was 0.763 with a sensitivity of 87.0% and specificity of 67.1% (Table 4 and Figure 1).Table 3Logistic Regression Analysis of Tregs and IL-6 in Cognitive Impairment After AISVariablesUnadjustedAdjusted Model 1Adjusted Model 2OR 95.0% CIpOR 95.0% CIpOR 95.0% CIpTregs%1.163(1.047–1.288)0.0051.155(1.037–1.286)0.0091.122(1.006–1.251)0.038IL-60.787(0.634–0.976)0.0300.786(0.621–0.995)0.0460.539(0.262–1.111)0.044Note: Model 1 accounted for sex and age while Model 2 was built upon Model. 1 by adjusting for risk factors such as UA, diabetes, NIHSS, VBI and WMH.Abbreviations: AIS, acute ischemic stroke; OR, odds ratio; CI, confidence interval; Tregs, Regulatory T cells; IL-6, interleukin-6. Table 4Evaluation of the Efficacy of Tregs% and IL-6 in Predicting Cognitive Impairment After AISVariablesCut off ValueAUC95% CISensitivitySpecificityYouden IndexIL-612.330.6600.526–0.7930.7390.6400.379Tregs%1.610.7180.600–0.8360.8480.5600.408Tregs% and IL-6-0.7630.653–0.8730.8700.6710.541 Figure 1ROC curve of Tregs% and IL-6 in cognitive impairment after AIS. Predictive values for PSCI,Area under the curve 0.763 for Tregs% and IL-6.A multi-line ROC curve graph of IL minus 6, Tregs percent and Tregs percent and IL minus 6.Abbreviations: AIS, acute ischemic stroke; PSCI, post-stroke cognitive impairment; Tregs, Regulatory T cells; IL-6, interleukin-6. Table 3Logistic Regression Analysis of Tregs and IL-6 in Cognitive Impairment After AISVariablesUnadjustedAdjusted Model 1Adjusted Model 2OR 95.0% CIpOR 95.0% CIpOR 95.0% CIpTregs%1.163(1.047–1.288)0.0051.155(1.037–1.286)0.0091.122(1.006–1.251)0.038IL-60.787(0.634–0.976)0.0300.786(0.621–0.995)0.0460.539(0.262–1.111)0.044Note: Model 1 accounted for sex and age while Model 2 was built upon Model. 1 by adjusting for risk factors such as UA, diabetes, NIHSS, VBI and WMH.Abbreviations: AIS, acute ischemic stroke; OR, odds ratio; CI, confidence interval; Tregs, Regulatory T cells; IL-6, interleukin-6. Logistic Regression Analysis of Tregs and IL-6 in Cognitive Impairment After AIS Note: Model 1 accounted for sex and age while Model 2 was built upon Model. 1 by adjusting for risk factors such as UA, diabetes, NIHSS, VBI and WMH. Abbreviations: AIS, acute ischemic stroke; OR, odds ratio; CI, confidence interval; Tregs, Regulatory T cells; IL-6, interleukin-6. Table 4Evaluation of the Efficacy of Tregs% and IL-6 in Predicting Cognitive Impairment After AISVariablesCut off ValueAUC95% CISensitivitySpecificityYouden IndexIL-612.330.6600.526–0.7930.7390.6400.379Tregs%1.610.7180.600–0.8360.8480.5600.408Tregs% and IL-6-0.7630.653–0.8730.8700.6710.541 Evaluation of the Efficacy of Tregs% and IL-6 in Predicting Cognitive Impairment After AIS ROC curve of Tregs% and IL-6 in cognitive impairment after AIS. Predictive values for PSCI,Area under the curve 0.763 for Tregs% and IL-6. Binary logistic regression identified IL-6 (OR=1.269; 95% CI: 1.079–1.492; p=0.004) as a significant predictor of PSD and this relationship remained after adjustment for confounding risk factors (OR=1.307; 95% CI: 1.058–1.614; p=0.013, Table 5). ROC curve analysis demonstrated an AUC of 0.805 (95% CI: 0.681–0.929; p<0.01) and the optimal cutoff value for IL-6 was 12.40pg/mL, corresponding to a sensitivity of 84% and specificity of 78% (Figure 2). Multivariate logistic regression analysis demonstrated relationship between Tregs%, IL-6 and the severity of PSCI, Tregs% (P=0.006) IL-6 (P=0.001) were associated with the PSD Group after adjusting for potential confounding factors, unassociated with the PSNCI group (Table 6).Table 5Logistic Regression Analysis of Tregs% and IL-6 in Dementia in AISVariablesUnadjustedAdjusted Model 3Adjusted Model 4OR 95.0% CIpOR 95.0% CIpOR 95.0% CIpTregs%0.540(0.188–1.556)0.2540.943(0.842–1.055)0.3041.059(0.395–2.842)0.909IL-61.269(1.079–1.492)0.0040.257(0.09–0.886)0.0301.307(1.058–1.614)0.013Note: Model 3 accounted for age, while Model 4 was built upon Model. 3 by adjusting for risk factors such as diabetes, NIHSS, VBI, brain atrophy and WMH.Abbreviations: AIS, acute ischemic stroke; OR, odds ratio; CI, confidence interval; Tregs, Regulatory T cells; IL-6, interleukin-6. Table 6Logistic Regression Analysis of Tregs% and IL-6 Across Different Degrees of PSCIVariablesUnadjustedAdjustedOR 95.0% CIPOR 95.0% CIPTregs%PSCIND1.866(0.797–4.370)0.1511.714 (0.714–4.115)0.228PSD4.248(1.673–10.785)0.0023.845(1.476–10.016)0.006IL-6PSCIND1.104(0.983–1.239)0.0941.113 (0.968–1.256)0.083PSD1.244(1.091–1.418)0.0011.258(1.097–1.443)0.001Note: adjusting for risk factors such as age, diabetes, UA,NIHSS, VBI and WMH.Abbreviations: PSCI, post-stroke cognitive impairment; OR, odds ratio; CI, confidence interval; Tregs, Regulatory T cells; IL-6, interleukin-6. Figure 2ROC curve for predicting PSD using IL-6. Predictive values for PSD, Area under the curve 0.805 for IL-6.A line graph showing ROC curve for predicting PSD using IL-6.Abbreviations: PSD, post-stroke dementias; IL-6, interleukin-6. Table 5Logistic Regression Analysis of Tregs% and IL-6 in Dementia in AISVariablesUnadjustedAdjusted Model 3Adjusted Model 4OR 95.0% CIpOR 95.0% CIpOR 95.0% CIpTregs%0.540(0.188–1.556)0.2540.943(0.842–1.055)0.3041.059(0.395–2.842)0.909IL-61.269(1.079–1.492)0.0040.257(0.09–0.886)0.0301.307(1.058–1.614)0.013Note: Model 3 accounted for age, while Model 4 was built upon Model. 3 by adjusting for risk factors such as diabetes, NIHSS, VBI, brain atrophy and WMH.Abbreviations: AIS, acute ischemic stroke; OR, odds ratio; CI, confidence interval; Tregs, Regulatory T cells; IL-6, interleukin-6. Logistic Regression Analysis of Tregs% and IL-6 in Dementia in AIS Note: Model 3 accounted for age, while Model 4 was built upon Model. 3 by adjusting for risk factors such as diabetes, NIHSS, VBI, brain atrophy and WMH. Abbreviations: AIS, acute ischemic stroke; OR, odds ratio; CI, confidence interval; Tregs, Regulatory T cells; IL-6, interleukin-6. Table 6Logistic Regression Analysis of Tregs% and IL-6 Across Different Degrees of PSCIVariablesUnadjustedAdjustedOR 95.0% CIPOR 95.0% CIPTregs%PSCIND1.866(0.797–4.370)0.1511.714 (0.714–4.115)0.228PSD4.248(1.673–10.785)0.0023.845(1.476–10.016)0.006IL-6PSCIND1.104(0.983–1.239)0.0941.113 (0.968–1.256)0.083PSD1.244(1.091–1.418)0.0011.258(1.097–1.443)0.001Note: adjusting for risk factors such as age, diabetes, UA,NIHSS, VBI and WMH.Abbreviations: PSCI, post-stroke cognitive impairment; OR, odds ratio; CI, confidence interval; Tregs, Regulatory T cells; IL-6, interleukin-6. Logistic Regression Analysis of Tregs% and IL-6 Across Different Degrees of PSCI Note: adjusting for risk factors such as age, diabetes, UA,NIHSS, VBI and WMH. Abbreviations: PSCI, post-stroke cognitive impairment; OR, odds ratio; CI, confidence interval; Tregs, Regulatory T cells; IL-6, interleukin-6. ROC curve for predicting PSD using IL-6. Predictive values for PSD, Area under the curve 0.805 for IL-6. A subgroup analysis was performed to investigate the Association between age, vascular risk factors, Tregs%, and IL-6. In This study, none of the subgroups, including age, sex, the presence of hypertension or diabetes or Coronary heart disease or Stroke, as well as smoking and Alcohol consumption, significantly modified the association between Tregs%, IL-6, and PSCI (all P values for interaction > 0.05, Figure 3). Figure 3Subgroup analyses of the association between Tregs%, IL-6 level and post-stroke cognitive. In the multivariate models, confounding factors, such as age, sex, Diabetes, Coronary heart disease, Stroke, Hypertension, Smoking, Alcohol consumption, UA, NIHSS, VBI and WMH were included unless the variables were used as a subgroup study.Subgroup Sex: Treg percent Male 49(45.37) odds ratio 1.187(0.693-2.037); Female 59(54.63) 0.841(0.492-1.442); P for interaction 0.238. Interleukin 6 Male 49(45.37) 0.992(0.919-1.070); Female 59(54.63) 0.441(0.635-1.388); P 0.586. Subgroup Ag, year: Treg percent less than 65 46(42.59) 0.857(0.493-1.493); greater than or equal to 65 62(57.41) 1.166(0.671-2.027); P 0.532. Interleukin 6 less than 65 46(42.59) 0.953(0.581-1.032); greater than or equal to 65 62(57.41) 1.049(0.369-1.131); P 0.829. Subgroup Diabetes: Treg percent Yes 47(43.51) 0.994(0.409-2.179); No 61(56.49) 1.059(0.459-2.444); P 0.893. Interleukin 6 Yes 47(43.51) 1.002(0.672-1.321); No 61(56.49) 0.930(0.501-1.111); P 0.978. Subgroup Coronary heart disease: Treg percent Yes 34(31.48) 1.096(0.604-1.988); No 74(68.52) 0.913(0.503-1.656); P 0.764. Interleukin 6 Yes 34(31.48) 0.979(0.603-1.670); No 74(68.52) 1.021(0.437-1.313); P 0.734. Subgroup Stroke: Treg percent Yes 22(20.37) 1.156(0.699-1.914); No 86(79.63) 0.865(0.523-1.431); P 0.572. Interleukin 6 Yes 22(20.37) 0.995(0.452-1.472); No 86(79.63) 1.121(0.523-1.175); P 0.890. Subgroup Hypertension: Treg percent Yes 49(45.37) 0.914(0.540-1.548); No 59(54.63) 1.094(0.646-1.852); P 0.738. Interleukin 6 Yes 49(45.37) 0.975(0.513-1.466); No 59(54.63) 0.986(0.612-1.266); P 0.721. Subgroup Smoking: Treg percent Yes 30(27.78) 1.058(0.763-1.789); No 78(72.22) 0.643(0.337-1.225); P 0.179. Interleukin 6 Yes 30(27.78) 1.058(0.763-1.789); No 78(72.22) 0.945(0.860-1.239); P 0.241. Subgroup Alcohol consumption: Treg percent Yes 32(29.63) 0.469(0.132-1.663); No 76(70.37) 2.132(0.601-4.560); P 0.352. Interleukin 6 Yes 32(29.63) 1.004(0.652-1.378); No 76(70.23) 0.998(0.548-1.374); P 0.979. The forest plot x-axis label is (unclear) with tick labels 0, 1, 2, 3, 4 (unit not shown). The y-axis label is Subgroup (unit not applicable). Each row shows a point estimate with a horizontal confidence interval bar, aligned to the listed odds ratio and 95 percent confidence interval values. The detailed data points are as follows: - For subgroup Sex, biomarker Treg percent, category Male, the number of patients was 49 (45.37 percent), the odds ratio was 1.187, the 95 percent CI lower was 0.693, the 95 percent CI upper was 2.037 and the P for interaction was 0.238. - For subgroup Sex, biomarker Treg percent, category Female, the number of patients was 59 (54.63 percent), the odds ratio was 0.841, the 95 percent CI lower was 0.492, the 95 percent CI upper was 1.442 and the P for interaction was 0.238. - For subgroup Sex, biomarker IL-6, category Male, the number of patients was 49 (45.37 percent), the odds ratio was 0.992, the 95 percent CI lower was 0.919, the 95 percent CI upper was 1.070 and the P for interaction was 0.586. - For subgroup Sex, biomarker IL-6, category Female, the number of patients was 59 (54.63 percent), the odds ratio was 0.441, the 95 percent CI lower was 0.635, the 95 percent CI upper was 1.388 and the P for interaction was 0.586. - For subgroup Age, year, biomarker Treg percent, category less than 65, the number of patients was 46 (42.59 percent), the odds ratio was 0.857, the 95 percent CI lower was 0.493, the 95 percent CI upper was 1.493 and the P for interaction was 0.532. - For subgroup Age, year, biomarker Treg percent, category greater than or equal to 65, the number of patients was 62 (57.41 percent), the odds ratio was 1.166, the 95 percent CI lower was 0.671, the 95 percent CI upper was 2.027 and the P for interaction was 0.532. - For subgroup Age, year, biomarker IL-6, category less than 65, the number of patients was 46 (42.59 percent), the odds ratio was 0.953, the 95 percent CI lower was 0.581, the 95 percent CI upper was 1.032 and the P for interaction was 0.829. - For subgroup Age, year, biomarker IL-6, category greater than or equal to 65, the number of patients was 62 (57.41 percent), the odds ratio was 1.049, the 95 percent CI lower was 0.369, the 95 percent CI upper was 1.131 and the P for interaction was 0.829. - For subgroup Diabetes, biomarker Treg percent, category Yes, the number of patients was 47 (43.51 percent), the odds ratio was 0.994, the 95 percent CI lower was 0.409, the 95 percent CI upper was 2.179 and the P for interaction was 0.893. - For subgroup Diabetes, biomarker Treg percent, category No, the number of patients was 61 (56.49 percent), the odds ratio was 1.059, the 95 percent CI lower was 0.459, the 95 percent CI upper was 2.444 and the P for interaction was 0.893. - For subgroup Diabetes, biomarker IL-6, category Yes, the number of patients was 47 (43.51 percent), the odds ratio was 1.002, the 95 percent CI lower was 0.672, the 95 percent CI upper was 1.321 and the P for interaction was 0.978. - For subgroup Diabetes, biomarker IL-6, category No, the number of patients was 61 (56.49 percent), the odds ratio was 0.930, the 95 percent CI lower was 0.501, the 95 percent CI upper was 1.111 and the P for interaction was 0.978. - For subgroup Coronary heart disease, biomarker Treg percent, category Yes, the number of patients was 34 (31.48 percent), the odds ratio was 1.096, the 95 percent CI lower was 0.604, the 95 percent CI upper was 1.988 and the P for interaction was 0.764. - For subgroup Coronary heart disease, biomarker Treg percent, category No, the number of patients was 74 (68.52 percent), the odds ratio was 0.913, the 95 percent CI lower was 0.503, the 95 percent CI upper was 1.656 and the P for interaction was 0.764. - For subgroup Coronary heart disease, biomarker IL-6, category Yes, the number of patients was 34 (31.48 percent), the odds ratio was 0.979, the 95 percent CI lower was 0.603, the 95 percent CI upper was 1.670 and the P for interaction was 0.734. - For subgroup Coronary heart disease, biomarker IL-6, category No, the number of patients was 74 (68.52 percent), the odds ratio was 1.021, the 95 percent CI lower was 0.437, the 95 percent CI upper was 1.313 and the P for interaction was 0.734. - For subgroup Stroke, biomarker Treg percent, category Yes, the number of patients was 22 (20.37 percent), the odds ratio was 1.156, the 95 percent CI lower was 0.699, the 95 percent CI upper was 1.914 and the P for interaction was 0.572. - For subgroup Stroke, biomarker Treg percent, category No, the number of patients was 86 (79.63 percent), the odds ratio was 0.865, the 95 percent CI lower was 0.523, the 95 percent CI upper was 1.431 and the P for interaction was 0.572. - For subgroup Stroke, biomarker IL-6, category Yes, the number of patients was 22 (20.37 percent), the odds ratio was 0.995, the 95 percent CI lower was 0.452, the 95 percent CI upper was 1.472 and the P for interaction was 0.890. - For subgroup Stroke, biomarker IL-6, category No, the number of patients was 86 (79.63 percent), the odds ratio was 1.121, the 95 percent CI lower was 0.523, the 95 percent CI upper was 1.175 and the P for interaction was 0.890. - For subgroup Hypertension, biomarker Treg percent, category Yes, the number of patients was 49 (45.37 percent), the odds ratio was 0.914, the 95 percent CI lower was 0.540, the 95 percent CI upper was 1.548 and the P for interaction was 0.738. - For subgroup Hypertension, biomarker Treg percent, category No, the number of patients was 59 (54.63 percent), the odds ratio was 1.094, the 95 percent CI lower was 0.646, the 95 percent CI upper was 1.852 and the P for interaction was 0.738. - For subgroup Hypertension, biomarker IL-6, category Yes, the number of patients was 49 (45.37 percent), the odds ratio was 0.975, the 95 percent CI lower was 0.513, the 95 percent CI upper was 1.466 and the P for interaction was 0.721. - For subgroup Hypertension, biomarker IL-6, category No, the number of patients was 59 (45.63 percent), the odds ratio was 0.986, the 95 percent CI lower was 0.612, the 95 percent CI upper was 1.266 and the P for interaction was 0.721. - For subgroup Smoking, biomarker Treg percent, category Yes, the number of patients was 30 (27.78 percent), the odds ratio was 1.058, the 95 percent CI lower was 0.763, the 95 percent CI upper was 1.789 and the P for interaction was 0.179. - For subgroup Smoking, biomarker Treg percent, category No, the number of patients was 78 (72.22 percent), the odds ratio was 0.643, the 95 percent CI lower was 0.337, the 95 percent CI upper was 1.225 and the P for interaction was 0.179. - For subgroup Smoking, biomarker IL-6, category Yes, the number of patients was 30 (27.78 percent), the odds ratio was 1.058, the 95 percent CI lower was 0.763, the 95 percent CI upper was 1.789 and the P for interaction was 0.241. - For subgroup Smoking, biomarker IL-6, category No, the number of patients was 78 (72.22 percent), the odds ratio was 0.945, the 95 percent CI lower was 0.860, the 95 percent CI upper was 1.239 and the P for interaction was 0.241. - For subgroup Alcohol consumption, biomarker Treg percent, category Yes, the number of patients was 32 (29.63 percent), the odds ratio was 0.469, the 95 percent CI lower was 0.132, the 95 percent CI upper was 1.663 and the P for interaction was 0.352. - For subgroup Alcohol consumption, biomarker Treg percent, category No, the number of patients was 76 (70.37 percent), the odds ratio was 2.132, the 95 percent CI lower was 0.601, the 95 percent CI upper was 4.560 and the P for interaction was 0.352. - For subgroup Alcohol consumption, biomarker IL-6, category Yes, the number of patients was 32 (29.63 percent), the odds ratio was 1.004, the 95 percent CI lower was 0.652, the 95 percent CI upper was 1.378 and the P for interaction was 0.979. - For subgroup Alcohol consumption, biomarker IL-6, category No, the number of patients was 76 (70.23 percent), the odds ratio was 0.998, the 95 percent CI lower was 0.548, the 95 percent CI upper was 1.374 and the P for interaction was 0.979.A forest plot of odds ratios for Treg percent and Interleukin 6 across clinical subgroups.Abbreviations: Tregs, Regulatory T cells; IL-6, interleukin-6; OR, odds ratio; CI, confidence intervals; VBI, vital brain infarction; WMH, White matter hyperintensities; UA, Uric Acid. Subgroup analyses of the association between Tregs%, IL-6 level and post-stroke cognitive. In the multivariate models, confounding factors, such as age, sex, Diabetes, Coronary heart disease, Stroke, Hypertension, Smoking, Alcohol consumption, UA, NIHSS, VBI and WMH were included unless the variables were used as a subgroup study. Discussion Tregs% and IL-6 were found to be elevated in the peripheral blood of the present cohort of PSCI patients with increases being particularly pronounced in PSD. Tregs% was identified as an independent risk factor for PSCI and the combination of Tregs% and IL-6 showed enhanced predictive accuracy for PSCI. IL-6 was also found to be an independent risk factor for PSD and may be a biomarker for PSD prediction. The involvement of Tregs in the pathophysiology of AIS is acknowledged to be significant, but a complex dual nature is shown, where the impact may be either protective or may exacerbate stroke-related injury. Empirical evidence suggests Tregs numbers to fluctuate during different AIS stages, with low numbers in the initial stages and elevated levels beyond one month post-stroke. Temporal variations influence stroke outcomes and neurological recovery trajectories.20–22 Tregs contribute to BBB preservation in the acute phase by mitigating inflammatory responses, but excessive numbers during later stages may impair BBB function, impeding neurological recuperation.23 Early post-stroke increases in Tregs populations have sometimes been reported, but chronic inflammatory conditions may induce Tregs dysfunction or exhaustion, diminishing the protective effect and facilitating the development and progression of cognitive impairment.24–26 Indeed,Tregs secrete anti-inflammatory cytokines, such as interleukin-10 (IL-10) and transforming growth factor-beta (TGF-β),27 which inhibit microglial polarization toward the pro-inflammatory M1 phenotype while promoting the anti-inflammatory M2 phenotype, attenuating neuroinflammation-associated cognitive deficits.28,29 Tregs have also been shown to suppress activation of the NLRP3 inflammasome and reduce the secretion pro-inflammatory cytokines, including IL-1β and IL-6, alleviating inflammatory burden within cerebral tissue and mitigating neuronal death and synaptic dysfunction.30,31 Tregs facilitate tissue repair and stimulate immune tolerance by modulating the activity of microglia and macrophages, mitigating white matter injury. The depletion of endogenous Tregs has been shown to delay functional recovery after stroke and to exacerbate white matter damage.32,33 This observation implies that the augmentation of Tregs populations may enhance white matter integrity.34 In our study, Elevated Tregs percentages in the early post-stroke stages were shown to be a risk factor for cognitive impairment after AIS, with an AUC value of 0.718 (95% CI: 0.600–0.836), and Tregs levels may be a valuable biomarker for early prediction and evaluation of PSCI risk, enabling timely therapeutic intervention. In fact, emerging evidence suggests that the consistent expression of FoxP3 does not fully explain the functional stability of Tregs. This is attributed to the considerable phenotypic and functional heterogeneity observed, which has prompted the classification of distinct Tregs subsets. Among them, there is a group of Tregs called “fragile Tregs”, which are characterized by the expression of FoxP3 and inflammatory cytokines but no immunosuppressive functions. This fragile subset of Tregs may directly contribute to the initiation of cytokine storms and exacerbate disease severity by promoting proinflammatory responses35,36 While the potent immunosuppressive properties of Tregs serve to mitigate detrimental inflammatory responses, excessive immunosuppression can induce systemic immune compromise, thereby increasing patients’ vulnerability to infectious complications. Helios-positive regulatory T cells (H+Tregs) had favor of pro-inflammatory than Helios-negative regulatory T cells (H-Tregs) in the early day after stroke. Notably, H+Tregs on day 3 may contribute to an elevated risk of infection due to their strong immunosuppressive properties.37 Therefor, The function of Tregs depends not on their quantity, but on their heterogeneity, functionality, and different subtypes. Our findings indicate that elevated Tregs are correlated with PSCI, which may contradict previous findings. It is evident that fragile Tregs could promote severe inflammation and impair the immunosuppressive function of Treg cells in acute and chronic inflammation. Current studies have shown that patients with severe COVID-19 have an increased proportion of Treg cells (CD3+CD4+CD25+FoxP3+), which is associated with a worse prognosis compared to those with mild COVID-19.38 Similarly, it has been reported that Tregs fragility may significantly contribute to the pathogenesis and progression of moyamoya disease (MMD) under chronic inflammation. Notably, in MMD, Tregs showed a marked increase, specifically in the proportions of unstable and fragile Tregs, accompanied by a reduction in the proportion of stable Tregs.39 One potential explanation for this discrepancy is that the proportion of Tregs measured may not accurately reflect their functional capacity in PSCI. It is plausible that during the acute phase of AIS, the heightened inflammatory response leads to elevated Tregs levels but gradually becoming more fragile Tregs that promote the development of PSCI. An initial elevation in the population of Tregs could promote a pro-inflammatory effect and may contribute to the development of PSCI. Significant challenges remain that require further investigation regarding the complexity and stability of Tregs function. IL-6 is a pleiotropic inflammatory cytokine with both pro- and anti-inflammatory properties. Excessive IL-6 expression activates neurocircuits that contribute to neuronal injury and precipitate PSCI,40 and elevated IL-6 has been positively correlated with the severity of cognitive deficits following stroke. A multicenter study involving 7053 patients with ischemic stroke and transient ischemic attack (TIA) demonstrated the association of early elevated IL-6 levels with increased stroke severity, higher recurrence rates and greater disability.41 IL-6 increases during the acute post-stroke phase as a consequence of ischemic brain injury, triggering an inflammatory response. Empirical evidence indicates that elevated IL-6 compromises BBB integrity, allowing the infiltration of immune cells and neurotoxic substances into brain parenchyma.42 Cerebrospinal fluid (CSF) and serum IL-6 are thus considered potential biomarkers for post-stroke cognitive decline.43 Longitudinal studies have identified significant associations between IL-6 concentrations at 3 and 6 months post-stroke and cognitive performance, assessed by instruments such as the MoCA and the Mini-Mental State Examination (MMSE),41,44 and cytokine profiling has correlated increased IL-6, IL-1β, IL-8, IL-10 and IL-12 with poorer cognitive outcomes. For instance, a study of 455 ischemic stroke patients reported a relationship between IL-6 levels and MoCA scores at 18 months post-event.45 Investigations into inflammatory markers have also identified associations between erythrocyte sedimentation rate, serum and CSF concentrations of C-reactive protein (CRP), IL-6, IL-10 and IL-12 and the incidence of PSCI. Stroke survivors with cognitive impairment also present with higher IL-6 levels than those in whom cognitive function is preserved, substantiating IL-6 as a prognostic indicator for PSCI. IL-6 expression was found to be elevated in the early phase of AIS in the current cohort of patients who later developed cognitive impairment, particularly those with PSD in this study. Indeed, IL-6 was a risk factor for PSD with an AUC value of 0.805 (95% CI: 0.681–0.929, p < 0.01). The optimal cutoff value was 12.40pg/mL, corresponding to a sensitivity of 84% and specificity of 78%. IL-6 (P=0.001) and Tregs% (P=0.006) are significantly associated with severe PSCI, especially in PSD. These findings suggest that early elevated IL-6 expression may be a valuable prognostic tool for assessing the risk of post-stroke cognitive decline. IL-6 is also a potential target for therapeutic intervention aimed at mitigating cognitive sequelae of stroke. IL-6 combined with Tregs provides a more effective prediction of PSCI than individual prediction. IL-6 plays a critical role in the differentiation, functionality, and stability of Tregs. Elevated levels of IL-6 were frequently observed in various autoimmune disorders and malignancies, where it contributes to disease progression by promoting Th17 differentiation and inhibiting Tregs function.46 Notably, IL-6 receptor (IL-6R) signaling within Tregs is essential for maintaining epithelial homeostasis and orchestrating protective inflammatory responses in inflammatory bowel disease (IBD).47 IL-6 signaling pathway is critically involved in the functions of Tregs. IL-6 in vitro plays a vital role in the balance between Th17 cells and Foxp3+Tregs because of IL-6 inhabiting the TGF-β induced generation of Foxp3+ Tregs and driving Th17 cell.48 It has been demonstrated that elevated levels of IL-6 negate the suppressive function of thymus-derived Tregs (nTregs). Meanwhile, it has been shown that these nTregs lost Foxp3 expression as well as their suppressive capacity in the simulation of increased IL-6. In IL-6 transgenic mice,which serve as a model for excessive IL-6 production during chronic inflammation,IL-6 may suppress peripherally derived Tregs(iTreg)generation,while may increase and retain self-tolerance by maintaining Foxp3+nTregs development.49 In mouse model with elective deletion IL-6R on Tregs (IL-6RaKO), while IL-6R-deficient Tregs maintain immune-suppressive capacity, mice lacking IL-6R on Tregs were more susceptible to inflammatory epithelial injury.47 Regarding the underlying mechanisms for PSCI involving IL-6 and Tregs remain to be fully elucidated. It is hypothesized that elevated IL-6 concentrations may provoke a pronounced inflammatory response that suppresses Tregs functional expression increasing fragility to promote PSCI. Nonetheless, this proposed mechanism warrants further empirical validation through comprehensive experimental studies. Consequently, targeting the IL-6 signaling pathway holds considerable promise as a therapeutic strategy for stroke management in the future. We acknowledge several limitations to the current study. Firstly, longitudinal monitoring of Treg levels and functional activity was not performed. Tregs and IL-6 fluctuate at various time points following AIS. The present investigation examined early expression but did not assess roles of Tregs in the different stroke phases of acute, subacute and chronic. Mechanisms of interaction among other Tregs subsets,IL-10 and TGF-β were not elucidated, particularly within the complex pathological framework of post-stroke cognitive impairment. The safety profile and therapeutic efficacy of Tregs-based interventions require further evaluation prior to clinical translation. The close association between fragile Tregs and severe inflammation highlights the potential for developing novel preventive and therapeutic strategies. Secondly, peripheral venous blood samples were examined but inflammatory cytokine levels in cerebrospinal fluid were not. This omission may limit the accuracy of inferences regarding central nervous system inflammatory status. Lastly, the relatively small sample size and absence of multicenter validation limit interpretations of external validity and generalizability. Conclusion Elevated levels of Tregs and IL-6 were predictors of PSCI. IL-6 was identified as having a strong impact on PSD, indicating that both Tregs and IL-6 may be biomarkers for PSCI prediction. IL-6 is a key cytokine that determines the balance of Tregs. While overproduction of IL-6 does not suppressed proliferation of Tregs,it could increasing fragile Tregs in PSCI. Thus,more in-depth studies are needed to elaborate mechanism between IL-6 signaling and fragile Tregs in PSCI. These results may inform early treatment methods and pave the way for development of immunotherapy treatments.