Lipoprotein(a) -Related Cognitive Vulnerability After Stroke Beyond High-Sensitivity C-Reactive Protein Authors: Dandan Wang, Yue Zheng, Yuesong Pan, Mengxing Wang, Tingfang Ai, Xingquan Zhao, Yongjun Wang, Jingjing Li Published: 2026 ABSTRACT AbstractPost-stroke cognitive impairment (PSCI) reflects vulnerability to cognitive decline beyond acute brain injury. Lipoprotein(a) (Lp(a)) is a genetically determined vascular risk factor with pro-inflammatory and pro-atherogenic properties, but the pathways linking Lp(a) to cognitive outcomes after stroke remain unclear. We examined whether systemic inflammation mediates or modifies Lp(a)-related cognitive risk after ischemic stroke. We analyzed data from a prespecified substudy of the Third China National Stroke Registry. Baseline serum Lp(a) and high-sensitivity C-reactive protein (hs-CRP) were measured. PSCI was defined as a Montreal Cognitive Assessment score ≤ 22 at 1 year. Multivariable logistic regression, interaction, joint exposure, and causal mediation analyses were performed to evaluate the role of systemic inflammation in the Lp(a)–PSCI association. Among 954 patients with acute ischemic stroke, higher baseline Lp(a) was associated with increased PSCI risk after adjustment for baseline cognitive performance and clinical covariates (highest vs. lowest tertile: OR, 1.479; 95% CI, 1.007–2.170). In contrast, hs-CRP was not independently associated with PSCI after multivariable adjustment. No significant multiplicative interaction between Lp(a) and hs-CRP was observed (P = 0.2026). Although joint exposure analyses showed the highest PSCI risk among patients with both elevated Lp(a) and hs-CRP, mediation analysis did not support hs-CRP as a meaningful mediator (proportion mediated = 1.46%). These findings suggest that the observed association between Lp(a) and PSCI was not substantially explained by hs-CRP-indexed peripheral systemic inflammation. Elevated Lp(a) was associated with PSCI independently of baseline cognitive performance, but this association was not mediated or significantly modified by hs-CRP. Lp(a)-related cognitive vulnerability after ischemic stroke may therefore reflect vascular susceptibility not captured by hs-CRP alone, with hs-CRP-defined systemic inflammation contributing to cumulative risk rather than serving as the primary mechanistic pathway.Supplementary InformationThe online version contains supplementary material available at https://doi.org/10.1007/s12975-026-01483-0.Keywords: Post-stroke cognitive impairment, Lipoprotein(a), Ischemic stroke, Cognitive decline, Systemic inflammation Introduction Post-stroke cognitive impairment (PSCI) is a frequent and disabling complication of acute stroke, affecting up to 60% of survivors within the first year [1]. PSCI is associated with reduced functional independence, poorer quality of life, and increased long-term morbidity and mortality [2, 3]. Despite its clinical importance, early identification of patients at high risk for PSCI remains challenging, and effective preventive strategies are limited. Increasing evidence suggests that cognitive outcomes after stroke are influenced not only by acute lesion-related factors, but also by individual susceptibility to secondary brain injury and cognitive decline [1, 4, 5]. However, clinical markers that capture this underlying vulnerability remain insufficiently characterized. Lipoprotein(a) (Lp(a)) is a genetically determined lipoprotein with well-established pro-atherogenic and pro-thrombotic properties and has been consistently associated with ischemic stroke and cardiovascular events [6–8]. Beyond its role in large-vessel atherosclerosis, Lp(a) has been implicated in chronic cerebrovascular injury, microvascular dysfunction, impaired fibrinolysis, and pro-inflammatory processes, all of which may contribute to long-term brain vulnerability. Prior studies examining Lp(a) and cognitive outcomes have yielded inconsistent findings, and most have focused on overall associations without distinguishing whether Lp(a) reflects pre-existing cognitive impairment, stroke-related injury, or an underlying susceptibility to post-stroke cognitive decline [9–11]. Systemic inflammation has been proposed as a potential pathway linking vascular risk factors, including Lp(a), to post-stroke brain injury and cognitive impairment. High-sensitivity C-reactive protein (hs-CRP), a widely used marker of systemic inflammatory activation, is commonly elevated after ischemic stroke and has been associated with stroke severity and unfavorable outcomes [12, 13]. However, it remains unclear whether systemic inflammation mediates or modifies the relationship between Lp(a) and PSCI, or whether Lp(a)-related cognitive risk is not fully explained by hs-CRP-indexed systemic inflammation. In this study, we investigated the association between baseline Lp(a) levels and the risk of PSCI at one year in patients with acute ischemic stroke. Given the proposed link between Lp(a) and inflammatory processes, we specifically examined whether systemic inflammation mediates or modifies this association. By doing so, we aimed to test whether Lp(a)-related cognitive risk is explained by hs-CRP-indexed peripheral systemic inflammation or reflects vascular susceptibility not captured by hs-CRP alone. Methods Study Design and Population The Third China National Stroke Registry (CNSR-III) is a nationwide, prospective clinical registry that consecutively enrolled patients with acute ischemic stroke (AIS) or transient ischemic attack (TIA) between August 2015 and January 2018. The Impairment of Cognition and Sleep after acute ischemic stroke or transient ischemic attack in Chinese patients (ICONS) study is a prespecified subgroup of CNSR-III focusing on cognitive and sleep outcomes after stroke. A total of 40 participating centers contributed to the ICONS study. Participants in ICONS met the same core inclusion criteria as CNSR-III, including age ≥ 18 years and a diagnosis of AIS or TIA with hospital admission within 7 days of symptom onset. To ensure the validity of cognitive assessments, ICONS applied additional exclusion criteria, including a history of cognitive impairment or psychotic disorders, illiteracy, and neurological or sensory conditions that could interfere with cognitive evaluation. The ICONS study was conducted in accordance with the Declaration of Helsinki and approved by the central institutional review board, with written informed consent obtained from all participants. The study design, enrollment procedures, and detailed inclusion and exclusion criteria of ICONS have been described previously [14, 15]. Basic and Clinical Data Collection Baseline information was obtained at enrollment by trained research staff using a unified electronic data capture system, integrating face-to-face interviews with medical record abstraction. Demographic characteristics, including age, sex, whether educational level less than 9 years, and body mass index (BMI), were recorded, along with lifestyle factors such as smoking status and alcohol consumption. Medical history related to vascular risk, including prior stroke, hypertension, diabetes mellitus, and hyperlipidemia, was also documented [14]. Neurological deficit at admission was evaluated using the National Institutes of Health Stroke Scale (NIHSS), and prestroke functional status was assessed with the modified Rankin Scale (mRS). Stroke etiology was determined according to the Trial of ORG 10,172 in Acute Stroke Treatment (TOAST) classification system, including large-artery atherosclerosis (LAA), cardioembolism (CE), small-artery occlusion (SAO), subtype of other determined etiology (SOE), and subtype of undetermined etiology (SUE). Laboratory Measurements Fasting blood samples were collected from all participants within 24 h of hospital admission. Samples were processed according to a standardized protocol and transported through a cold chain to the central laboratory at Beijing Tiantan Hospital, where serum specimens were stored at − 80 °C until analysis. Serum Lp(a) concentrations were measured using a commercially available enzyme-linked immunosorbent assay (ELISA) based on a sandwich technique targeting distinct epitopes of apolipoprotein(a). Hs-CRP levels were quantified using a high-sensitivity immunoassay in a certified central laboratory, with laboratory personnel blinded to clinical information. Additional baseline laboratory measurements included total cholesterol (TC), low-density lipoprotein cholesterol (LDL-C), fasting glucose, and glycated hemoglobin (GHb), which were analyzed using standard automated methods. All laboratory analyses were performed in accordance with the manufacturers’ instructions and established quality control procedures. Assessment of Cognition Cognitive performance was assessed using the Montreal Cognitive Assessment (MoCA) by trained and certified examiners following standardized assessment procedures. Baseline cognitive assessment was performed during the index hospitalization after clinical stabilization, and follow-up cognitive evaluation was conducted at 1 year after stroke onset. PSCI was defined as a MoCA score ≤ 22 at 1-year follow-up, consistent with prior studies in Chinese patients with stroke or transient ischemic attack using the MoCA-Beijing version and with previous ICONS-based analyses [15, 16]. Baseline MoCA scores were recorded and incorporated as covariates in multivariable analyses to account for pre-existing cognitive status. Statistical Analysis Baseline characteristics were summarized according to PSCI status. Continuous variables with non-normal distributions were presented as medians with interquartile ranges and compared using the Wilcoxon rank-sum test. Categorical variables were expressed as counts with percentages and compared using the χ² test or Fisher exact test, as appropriate. Serum Lp(a) levels were categorized into tertiles based on their distribution in the study population. Hs-CRP was dichotomized using a cutoff of 3 mg/L (≤ 3 vs. > 3 mg/L), a commonly used threshold for high inflammatory risk in cardiovascular risk stratification [17]. In our research, hs-CRP was also analyzed as a log-transformed continuous variable in sensitivity analyses. Multivariable logistic regression models were applied to estimate odds ratios (ORs) and 95% confidence intervals (CIs) for the association between baseline Lp(a) levels and PSCI at 1 year. Two adjusted models were constructed. Model 1 included age, sex, educational level, prior stroke, hypertension, diabetes mellitus, hyperlipidemia, NIHSS score at admission, and TOAST classification. Model 2 further adjusted for baseline MoCA score. Additional analyses were performed to address potential selection bias, baseline cognitive status, hs-CRP modeling, and stroke subtype heterogeneity. Baseline characteristics were compared between included and excluded patients from the ICONS cognitive cohort. Sensitivity analyses were conducted by restricting the sample to patients with baseline MoCA > 22 and by modeling change in MoCA score from baseline to 1 year as a continuous outcome. Hs-CRP was also analyzed as a log-transformed continuous variable because of its skewed distribution. Exploratory analyses were further performed according to TOAST subtype. Effect modification by systemic inflammation was evaluated by including an interaction term between Lp(a) and hs-CRP in the regression models. Joint associations of Lp(a) tertiles and hs-CRP categories with PSCI were further examined using combined exposure groups, with the low Lp(a) and low hs-CRP group as the reference. Causal mediation analysis under the counterfactual framework was performed to assess whether hs-CRP mediated the association between Lp(a) and PSCI, adjusting for covariates included in Model 1. All analyses were conducted using SAS software version 9.4 (SAS Institute Inc.), and a two-sided P value < 0.05 was considered statistically significant. Results Baseline Characteristics Among 15 166 patients enrolled in the CNSR-III, 4 475 patients were screened for inclusion in the ICONS substudy. A total of 2,625 patients were included in the cognitive cohort. After exclusion of patients with incomplete data, 954 patients were eligible for the present analysis (Fig. 1). Flowchart of study participant selection. CNSR-III: The Third China National Stroke Registry, ICONS: The Impairment of Cognition and Sleep after acute ischemic stroke or transient ischemic attack in Chinese patients, Lp(a): lipoprotein(a), hs-CRP: high-sensitivity C-reactive protein, MoCA: Montreal Cognitive Assessment The mean age of the study population was 61.49 ± 10.96 years, and 693 patients (72.64%) were male. Baseline characteristics of patients with and without PSCI are shown in Table 1. Compared with patients without PSCI, those who had PSCI at 1 year were older, had lower educational attainment, lower baseline MoCA scores, and higher baseline Lp(a) and hs-CRP levels. Notably, admission NIHSS and prestroke mRS did not differ significantly between groups, suggesting that subsequent cognitive impairment was not simply explained by greater baseline neurological deficit or functional disability. These baseline patterns support the concept that PSCI may reflect inter-individual vulnerability beyond acute stroke severity alone. At the same time, the marked difference in baseline MoCA scores highlights the importance of accounting for baseline cognitive performance, which was addressed in multivariable models and sensitivity analyses. Table 1Basic characteristics of patients between different cognitive groups at baselinePSCIN= 315Non-PSCIN= 639P valueAge (quartile), y65.00(58.00–73.00)60.00(52.00–67.00)< 0.0001Male (n), %215(68.25%)478(74.80%)0.0328Low Education (n), %238(75.56%)384(60.09%)< 0.0001BMI (quartile)24.77(22.60–26.99)24.66(22.86–26.70)0.7728Smoking (n), %100(31.75%)232(36.31%)0.1643Alcohol consumption (n), %57(18.10%)99(15.49%)0.3068Past history of Stroke (n), %79(25.08%)122(19.09%)0.0330 Hypertension (n), %208(66.03%)402(62.91%)0.3451 Diabetes mellitus (n), %84(26.67%)145(22.69%)0.1764 Hyperlipidemia (n), %32(10.16%)76(11.89%)0.4264mRS (quartile)0.00(0.00–1.00)0.00(0.00–1.00)0.6823NIHSS (quartile)3.00(1.00–5.00)3.00(1.00–5.00)0.8621TOAST classification0.0375 LAA (n), %96(30.48%)148(23.16%) CE (n), %23(7.30%)36(5.63%) SAO (n), %61(19.37%)161(25.20%) SOE (n), %4(1.27%)4(0.63%) SUE (n), %131(41.59%)290(45.38%)Lp(a)167.08(84.02–335.98)138.44(63.72–282.05)0.0015Hs-CRP1.70(0.72–4.86)1.40(0.69–3.30)0.0457LDL-C2.36(1.71–3.04)2.20(1.63–2.96)0.0568TC4.00(3.24–4.63)3.78(3.18–4.52)0.0564Fasting glucose5.61(4.96–7.07)5.49(4.90–6.90)0.2612GHb6.00(5.50–7.50)5.98(5.50–7.00)0.3471MoCA at baseline18.00(13.00–22.00)24.00(21.00–27.00)< 0.0001PSCI post-stroke cognitive impairment, BMI body mass index, NIHSS National Institutes of Health Stroke Scale, TOAST Trial of ORG 10,172 in Acute Stroke Treatment, LAA large-artery atherosclerosis, CE cardioembolism, SAO small-artery occlusion, SOE subtype of other determined etiology, SUE and subtype of undetermined etiology, Lp(a) lipoprotein(a), Hs-CRP high-sensitivity C-reactive protein, LDL-C low-density lipoprotein cholesterol, TC total cholesterol, GHb glycated hemoglobin, MoCA Montreal Cognitive Assessment Basic characteristics of patients between different cognitive groups at baseline PSCI post-stroke cognitive impairment, BMI body mass index, NIHSS National Institutes of Health Stroke Scale, TOAST Trial of ORG 10,172 in Acute Stroke Treatment, LAA large-artery atherosclerosis, CE cardioembolism, SAO small-artery occlusion, SOE subtype of other determined etiology, SUE and subtype of undetermined etiology, Lp(a) lipoprotein(a), Hs-CRP high-sensitivity C-reactive protein, LDL-C low-density lipoprotein cholesterol, TC total cholesterol, GHb glycated hemoglobin, MoCA Montreal Cognitive Assessment Compared with excluded patients, included patients showed generally similar baseline demographic and clinical characteristics, including age, sex, education, NIHSS score, TOAST classification, baseline MoCA score, and hs-CRP levels (Table S1). Although several laboratory variables differed statistically, all standardized mean differences were < 0.10, suggesting no substantial imbalance between included and excluded patients. Sequential Models Testing the Association of Lp(a), hs-CRP, and Their Interaction With PSCI Higher baseline Lp(a) levels were associated with an increased risk of PSCI (Table 2). In unadjusted analyses, patients in the medium and high Lp(a) tertiles had higher odds of PSCI compared with those in the lowest tertile. After adjustment for age, sex, educational level, prior stroke, hypertension, diabetes mellitus, hyperlipidemia, NIHSS at admission, and TOAST classification (Model 1), the association between elevated Lp(a) and PSCI remained evident, with a significant increase in PSCI risk observed in the highest Lp(a) tertile. Further adjustment for baseline MoCA (Model 2) slightly attenuated the association, but participants in the highest Lp(a) tertile continued to have a significantly higher risk of PSCI compared with those in the lowest tertile. The persistence of this association after adjustment for baseline MoCA suggests that elevated Lp(a) was not merely a marker of poorer cognitive performance at the time of stroke, but was associated with subsequent cognitive impairment. This interpretation was further supported by sensitivity analyses restricted to patients with baseline MoCA > 22 and analyses of longitudinal MoCA change. Table 2Sequential models testing the association of Lp(a), hs-CRP, and their interaction with PSCIExposure/ModelOR95% CIP valueLp(a), medium vs. low Crude1.5211.084–2.1350.0152 Model 11.3730.962–1.9600.0812 Model 21.4250.967–2.1000.0732Lp(a), high vs. low Crude1.6361.168–2.2940.0042 Model 11.5341.076–2.1860.0179 Model 21.4791.007–2.1700.0458Hs-CRP, high vs. low Crude1.3811.034–1.8450.0289 Model 11.1650.854–1.5880.3553 Model 21.2020.857–1.6860.2855Lp(a)* hs-CRPP for interaction = 0.2026Lp(a) tertiles were defined as follows: tertile low: ≤93.9 mg/dl, tertile medium: 93.9–223.4 mg/dl, and tertile high: >223.4 mg/dlModel 1: adjusted by age, sex, education, past history of stroke, hypertension, diabetes, and hyperlipidemia, NIHSS, and TOAST classificationModel 2: adjusted by age, sex, education, past history of stroke, hypertension, diabetes, and hyperlipidemia, NIHSS, TOAST classification, and MoCA at baseline Sequential models testing the association of Lp(a), hs-CRP, and their interaction with PSCI Lp(a) tertiles were defined as follows: tertile low: ≤93.9 mg/dl, tertile medium: 93.9–223.4 mg/dl, and tertile high: >223.4 mg/dl Model 1: adjusted by age, sex, education, past history of stroke, hypertension, diabetes, and hyperlipidemia, NIHSS, and TOAST classification Model 2: adjusted by age, sex, education, past history of stroke, hypertension, diabetes, and hyperlipidemia, NIHSS, TOAST classification, and MoCA at baseline Baseline hs-CRP levels were modestly higher among patients who had PSCI at 1 year and were associated with PSCI in unadjusted analyses. However, after adjustment for Model 1, the association between hs-CRP and PSCI was attenuated and no longer statistically significant. Further adjustment for Model 2 did not materially change the results (Table 2). In sensitivity analyses treating hs-CRP as a continuous variable, log-transformed hs-CRP was associated with PSCI in the crude model (OR (95% CI) = 1.126 (1.012–1.253)), but was no longer statistically significant after adjustment for clinical covariates and baseline MoCA in Model 2 (OR (95% CI) = 1.116 (0.986–1.264)). This attenuation suggests that hs-CRP may reflect the broader clinical risk profile of patients who had PSCI at 1 year, rather than serving as an independent determinant of cognitive outcome in this cohort. To examine whether systemic inflammation modified the association between Lp(a) and PSCI, an interaction term between Lp(a) tertiles and hs-CRP categories was included in multivariable models. No statistically significant interaction was observed between Lp(a) and hs-CRP on the risk of PSCI (P for interaction = 0.2026, Table 2). Thus, the association between Lp(a) and PSCI did not appear to depend significantly on the level of systemic inflammation as measured by hs-CRP. In a sensitivity analysis restricted to patients with baseline MoCA > 22, 64 of 483 patients had PSCI at 1 year. The association between high Lp(a) and PSCI remained directionally consistent with the primary analysis but did not reach statistical significance after multivariable adjustment (high vs. low tertile: OR (95% CI) = 1.825 (0.924–3.605)). Hs-CRP remained unassociated with PSCI, and no significant interaction between Lp(a) and hs-CRP was observed (Table S2). In an additional sensitivity analysis using change in MoCA score from baseline to 1 year as a continuous outcome, higher Lp(a) levels were directionally associated with less favorable cognitive change after adjustment for baseline MoCA, with a borderline overall association across tertiles (Table S3). In exploratory analyses stratified by TOAST subtype, the association between high Lp(a) and PSCI appeared most pronounced among patients with large-artery atherosclerosis. In the LAA subtype, high versus low Lp(a) remained associated with PSCI after adjustment for baseline MoCA and clinical covariates. No consistent association was observed in CE, SAO, or SUE subtypes, and estimates for SOE were not interpreted because of sparse data (Table S4). When Lp(a) and hs-CRP were jointly considered, a graded increase in PSCI risk was observed across increasing Lp(a) tertiles within both hs-CRP strata (Fig. 2). Compared with the reference group (low Lp(a) and low hs-CRP), the highest risk was observed among participants with concomitantly elevated Lp(a) and hs-CRP (OR (95% CI) = 2.079 (1.231–3.509)). In stratified analyses, higher Lp(a) tertiles were generally associated with increased PSCI risk within each hs-CRP category, although the magnitude of association varied across subgroups. Together with the nonsignificant interaction term, this joint exposure pattern suggests cumulative risk rather than a synergistic effect between elevated Lp(a) and systemic inflammation. Joint exposure pattern of Lp(a) and hs-CRP in relation to PSCI. Forest plot showing odds ratios (ORs) and 95% confidence intervals (CIs) for PSCI according to combined categories of Lp(a) tertiles and hs-CRP levels. The reference group was low Lp(a) and low hs-CRP. Estimates were derived from multivariable logistic regression models adjusted for age, sex, education, prior stroke, hypertension, diabetes, hyperlipidemia, NIHSS, and TOAST classification. The graded increase in PSCI risk across Lp(a) tertiles within both hs-CRP strata, together with the nonsignificant multiplicative interaction, suggests a cumulative rather than synergistic pattern of risk Causal Mediation Analysis Causal mediation analysis was performed to assess whether systemic inflammation, as indexed by hs-CRP, mediated the association between baseline Lp(a) levels and PSCI. The natural indirect effect through hs-CRP was not statistically significant, and the proportion of the total association mediated by hs-CRP was small (1.46%). In contrast, the natural direct effect of Lp(a) on PSCI remained statistically significant and was comparable in magnitude to the total effect, indicating that the association between Lp(a) and PSCI was not meaningfully explained by systemic inflammation (Table 3). These findings indicate that hs-CRP did not meaningfully explain the association between Lp(a) and PSCI. Collectively, the persistence of the Lp(a) association after baseline cognitive adjustment, the attenuation of the hs-CRP association after clinical adjustment, the absence of significant effect modification, and the negligible mediated proportion suggest that hs-CRP-indexed peripheral systemic inflammation did not substantially explain the observed Lp(a)–PSCI association. Table 3Causal mediation analysis testing hs-CRP as an explanatory pathway between Lp(a) and PSCIEffectEstimate95% CIP valueProportion mediatedTotal effect0.000150.00004–0.000260.008Direct effect (ADE)0.000150.00004–0.000260.010—Indirect effect (ACME)0.000000.00000–0.000000.4421.46%A negligible and nonsignificant ACME, together with a preserved direct effect, indicates that hs-CRP did not meaningfully mediate the Lp(a)–PSCI associationADE average direct effect, ACME average causal mediation effect*All adjusted by age, sex, education, past history of stroke, hypertension, diabetes, hyperlipidemia, NIHSS, and TOAST classification Causal mediation analysis testing hs-CRP as an explanatory pathway between Lp(a) and PSCI A negligible and nonsignificant ACME, together with a preserved direct effect, indicates that hs-CRP did not meaningfully mediate the Lp(a)–PSCI association ADE average direct effect, ACME average causal mediation effect *All adjusted by age, sex, education, past history of stroke, hypertension, diabetes, hyperlipidemia, NIHSS, and TOAST classification Discussion In this large prospective cohort of patients with acute ischemic stroke, elevated baseline Lp(a) was associated with an increased risk of PSCI at 1 year, independent of baseline cognitive performance. Given prior evidence linking Lp(a) to both vascular and inflammatory processes, we examined whether systemic inflammation, as indexed by hs-CRP, explained this association. Hs-CRP was not independently associated with PSCI after multivariable adjustment, did not significantly modify the Lp(a)–PSCI association, and did not meaningfully mediate this relationship. These findings suggest that Lp(a)-related cognitive vulnerability after ischemic stroke is not primarily explained by hs-CRP-indexed peripheral systemic inflammation and may instead reflect vascular susceptibility not captured by hs-CRP alone. Accumulating evidence suggests that the association between Lp(a) and cognitive outcomes is complex and context dependent. In population-based studies, relationships between Lp(a) levels and global cognitive performance or dementia have generally been weak or inconsistent, indicating that Lp(a)-related cognitive effects may not be readily detectable in the absence of substantial cerebrovascular burden [18–20]. By contrast, in cohorts enriched for vascular disease, particularly ischemic stroke, elevated Lp(a) has been more consistently linked to adverse cognitive outcomes [9, 15, 21]. These observations are biologically plausible given the established pro-atherogenic, pro-thrombotic, and microvascular effects of Lp(a), which may contribute to cumulative brain vulnerability rather than acute focal injury alone [6, 22–24]. In this context, our findings extend prior work by focusing not only on the presence of an association, but also on the potential pathways underlying Lp(a)-related cognitive risk, particularly in relation to systemic inflammation. Although systemic inflammation has been proposed as a potential pathway linking vascular risk factors to PSCI, our findings suggest a limited explanatory role of hs-CRP-indexed peripheral systemic inflammation in the association between Lp(a) and PSCI. Specifically, hs-CRP did not mediate the effect of Lp(a) on PSCI, as indicated by the negligible indirect effect and the very small proportion mediated in causal mediation analysis. It should be noted that hs-CRP represents a nonspecific marker of systemic inflammation and may not capture more proximal inflammatory pathways, such as cytokine signaling or immune cell activation, which have been implicated in PSCI in prior studies [25, 26]. At the same time, no statistically significant multiplicative interaction between Lp(a) and hs-CRP was detected. However, joint exposure analyses revealed a graded increase in PSCI risk across increasing Lp(a) levels within both hs-CRP strata, with the highest risk observed among patients with concomitantly elevated Lp(a) and hs-CRP. These results do not support a primary mediating role of hs-CRP-indexed systemic inflammation in the association between Lp(a) and PSCI. Instead, they are more consistent with a pattern of risk accumulation rather than synergistic interaction on the multiplicative scale. This pattern aligns with the interpretation that Lp(a) may represent a relatively stable vascular susceptibility, whereas systemic inflammation reflects a contextual or amplifying factor that influences the clinical expression of this risk [22, 23, 27]. In this context, hs-CRP may indicate an adverse vascular or metabolic milieu in which the effects of elevated Lp(a) on cerebral microcirculation, endothelial function, or thrombotic tendency become more apparent, without serving as a direct mediator of the Lp(a)–PSCI association [5, 28, 29]. This interpretation is further supported by the observation that elevated Lp(a) was associated with increased PSCI risk even among patients with low hs-CRP levels, indicating that the Lp(a)–PSCI association was not restricted to patients with elevated hs-CRP. Conversely, the greater effect size observed in the presence of elevated hs-CRP suggests risk accumulation rather than multiplicative interaction, a pattern consistent with current models of PSCI emphasizing cumulative brain vulnerability [1, 30–32]. From a clinical perspective, our findings refine the interpretation of Lp(a)-related PSCI risk by suggesting that this risk is unlikely to be primarily driven by acute systemic inflammation. Given its strong genetic determination and relative temporal stability, Lp(a) may help identify patients at increased risk for long-term cognitive decline following ischemic stroke, complementing conventional clinical and vascular risk factors. Importantly, the persistence of the Lp(a)–PSCI association after adjustment for baseline cognitive performance, together with directionally consistent findings in sensitivity analyses restricted to patients with baseline MoCA > 22 and analyses of MoCA change, supports the possibility that Lp(a) captures vulnerability beyond pre-existing cognitive status, underscoring its potential value in early risk stratification. From a research standpoint, the lack of a meaningful mediating effect of hs-CRP highlights the need to consider mechanisms beyond systemic inflammation when investigating the link between vascular risk factors and cognitive outcomes after stroke. Future studies may benefit from integrating imaging markers of cerebral small vessel disease, microstructural white matter integrity, or cerebral perfusion, as well as longitudinal assessments of cognitive trajectories, to better delineate the mechanisms underlying Lp(a)-related cognitive vulnerability. Such approaches may help clarify whether Lp(a) exerts its effects primarily through chronic microvascular dysfunction, impaired cerebrovascular reserve, or cumulative structural brain injury, rather than through acute systemic inflammatory responses. Several limitations of this study should be acknowledged. First, hs-CRP was measured at a single time point during the acute phase of stroke, which may not fully capture chronic inflammatory status or longitudinal inflammatory dynamics relevant to cognitive decline. In addition, hs-CRP is only a nonspecific peripheral marker and cannot fully represent the complex, multi-system inflammatory response after stroke. Second, cognitive outcomes were assessed using the MoCA, a widely used screening tool that may not fully characterize domain-specific cognitive deficits. Although a MoCA cutoff of ≤ 22 has been used in prior Chinese stroke cohorts, a single cutoff may still be relatively coarse and may not fully account for individual differences in age and education. Third, although baseline MoCA was adjusted for and sensitivity analyses were performed among patients with baseline MoCA > 22 and using MoCA change, residual influence of baseline or persistent cognitive impairment cannot be completely excluded. Fourth, the final analysis was based on complete cases; however, included and excluded patients showed generally similar baseline characteristics, with all standardized mean differences < 0.10. Finally, we did not incorporate neuroimaging markers of cerebral small vessel disease or microstructural integrity, which may have provided additional mechanistic insights. Despite these limitations, the prospective multicenter design and standardized assessment of cognitive outcomes at 1 year support the overall robustness of the findings. Conclusion Elevated baseline Lp(a) levels are associated with an increased risk of PSCI at 1 year, independent of baseline cognitive performance. This association is not meaningfully mediated or modified by hs-CRP, suggesting that hs-CRP-indexed peripheral systemic inflammation does not substantially explain the observed Lp(a)–PSCI association. Supplementary Information Below is the link to the electronic supplementary material. Supplementary Material 1 (DOCX 24.2 KB) Acknowledgements We thank all the participants in the study. Author Contributions DW conceived and designed the study, interpreted the data, and drafted the initial manuscript. YZ revised the manuscript. YP, MW, and TA contributed to data acquisition, performed the statistical analyses, and assisted with data interpretation. XZ provided overall supervision of the research project. YW and JL conceived and designed the study, critically revised the manuscript for important intellectual content, and were responsible for funding acquisition. All authors contributed to the article and approved the submitted version. Funding This study was supported by 8th Central Healthcare Research Program of China (No. 2022YB65), National Science and Technology Major Project for the Prevention and Treatment of Cancer, Cardiovascular, Respiratory, and Metabolic Diseases(2026ZD0557802), and capital health research and development of special public health project (2026-1-2013). Data Availability All data generated or analyzed during this study are included in this article. Further inquiries can be directed to the corresponding authors. Declarations Competing interests The authors declare no competing interests. Footnotes Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Contributor Information Yongjun Wang, Email: yongjunwang@ncrcnd.org.cn. Jingjing Li, Email: jingjingli0606@126.com. References Associated Data Data Availability Statement All data generated or analyzed during this study are included in this article. Further inquiries can be directed to the corresponding authors.