This real-world cross-sectional study utilized prospectively collected data from the National Metabolic Management Center at Longyan First Affiliated Hospital of Fujian Medical University and Fuqing City Hospital Affiliated with Fujian Medical University. A consecutive recruitment approach was employed to enroll participants with T2DM aged over 18, who were admitted between January 2023 and December 2023. Prior to their enrollment, written informed consent was diligently obtained from all participants, aligning with the ethical guidelines set forth by the Ethical Committee of Longyan First Affiliated Hospital of Fujian Medical University. Throughout the study, adherence to the principles outlined in the Declaration of Helsinki was ensured, governing the conduct of all procedures. Exclusion criteria were diligently applied to ensure the appropriate selection of participants. Specifically, individuals were excluded if they met any of the following conditions: 1. Concurrent presence of acute and chronic infections, acute diabetes complications, or experiencing acute stress states. 2. Coexistence of systemic diseases that may impede immune system function (e.g., blood, rheumatic diseases, and malignant tumors). 3. Current usage of medications known to potentially induce liver steatosis or interfere with the blood system (e.g., propylthiouracil, estrogens, tamoxifen, methimazole, and glucocorticoids). 4. Heavy alcohol consumption, defined as a daily intake of ≥ 40 g for men and ≥ 20 g for women, persisted for more than 5 years. 5. Presence of other liver comorbidities capable of inducing liver steatosis (e.g., viral hepatitis, autoimmune liver disease, or hereditary liver disease). 6. a history of cerebrovascular disease. 7. Pregnancy status or inadequate data availability. Following the recruitment process, a total of 830 participants with T2DM were included in the final analysis. Subsequently, all enrolled participants were conducted with non-enhanced CT and B-mode carotid ultrasonography to facilitate the identification and assessment of MAFLD and SCAS.

MAFLD served as the exposure variable. The definition of MAFLD in T2DM adhered to the latest expert consensus statement. This definition emphasized the identification of hepatic steatosis through various methods such as imaging techniques, blood biomarkers, or liver histology (18). After enrollment, all participants underwent non-enhanced abdominal CT scans to evaluate hepatic steatosis utilizing the CT liver-spleen attenuation measurement (CT_L-S). This CT index is specifically developed for accurately assessing hepatic steatosis (fatty liver). The CT_L-S was derived by dividing the mean liver attenuation by the mean spleen attenuation. A mean CTL-S value below 1.0 was indicative of hepatic steatosis. Two experienced radiologists were involved in the CTL-S measurement process to ensure consistency and reduce inter-operator variability.

SII is identified as the primary mediator, offering valuable insights into local immune response and systemic inflammation. SII was derived through the formula: platelet count x neutrophil count/lymphocyte count, a calculation method consistent with previous relevant research studies (13). These blood cells count was determined utilizing the Coulter LH 780 Analyzer (Beckman Coulter Ireland, Galway, Ireland).

SCAS is identified as the outcome variable in this study. As established in previous reviews, SCAS was defined as the presence of cIMT exceeding 1.0mm or the presence of arterial plaque, either independently or in combination (19). Carotid artery evaluations were performed using carotid ultrasonography with an L12–5 MHz ultrasound probe, encompassing the common carotid artery, internal carotid artery, external carotid artery, and carotid bifurcation. The average intima-media thickness (IMT) of the distal vascular wall at the proximal end of the carotid artery glomus was calculated to determine the carotid intima-media thickness (cIMT). Carotid plaque was defined as a focal structure that protruded at least 0.5 mm into the arterial lumen, exhibited greater than 50% thickness compared to the surrounding IMT, or had an IMT exceeding 1.5 mm (20). All procedures were executed by skilled ultrasonologists following the standardized measurement protocol endorsed by the Society for Vascular Medicine (21).

Multivariable models were constructed to adjust for the potential confounding variables that may influence the association between MAFLD and SCAS. Comprehensive demographic data, including sex, age, diabetic duration, smoking and drinking status, waist circumference (WC), systolic blood pressure (SBP), and diastolic blood pressure (DBP), was meticulously collected by trained research personnel and recorded in the information collecting system. The laboratory assessments encompassed a wide range of measurements, including serum levels of creatinine, alanine aminotransferase, aspartate aminotransferase (AST), uric acid (UA), fasting plasma glucose (FBG), serum insulin levels, triglyceride, total cholesterol (TC), low-density lipoprotein cholesterol (LDL-c), high-density lipoprotein cholesterol (HDL-c), and hemoglobin A1c (HbA1c). Biochemical parameters were conducted using an auto-biochemical analyzer (Roche Diagnostics Corporation). HbA1c levels were assessed through high-performance liquid chromatography with a D10 set (Bio-Rad). Homeostatic model assessment of insulin resistance (HOMA-IR) was calculated using the formula: fasting serum insulin (µU/ml) x FBG (mmol/l)/22.5.

Continuous variables were reported as means ± standard deviation (SD), while discrete variables were presented as frequency tables (N, %). To assess differences among the SII quartiles, ANOVA or the K-W test was utilized for continuous variables. The chi-square or Fisher exact test was employed to compare categorical variables. The relationships between the SII, cIMT, and CT_L-S were assessed using Spearman correlation analysis. Subsequently, these associations underwent further scrutiny through weighted multivariable regression analyses within three distinct models. Model 1 encompassed adjustments for age, gender, diabetic duration, smoking, and drinking. In Model 2, additional adjustments were made considering cardiometabolic variables including SBP, DBP, WC, BMI, HbA1c, TG, TC, HDL-c, LDL-c, UA, and HOMA-IR. Model 3 featured supplementary adjustments for liver functional variables such as ALT and AST. Weighted logistic regression was employed to determine odds ratios (OR) and 95% confidence intervals (CI) for the correlations between SII quartiles and the risks of MAFLD and SCAS. Moreover, restricted cubic spline (RCS) analyses were carried out to explore the relationship between Ln (SII) and the risk of MAFLD and SCAS. Given the non-uniform distribution of SII data, the Ln (SII) was utilized to render the statistical analysis more appropriate. Mediation models using bootstrapping calculations were then employed to evaluate the direct impact of MAFLD on SCAS risk, as well as the indirect effect mediated by SII and Ln (SII). Finally, sensitivity analyses were performed, focusing on the associations as mentioned above in men to account for potential variations in SII categorizations and further enhance the robustness of the study findings. All statistical analyses were executed using R language 4.2.3 software. The significance level was set at P < 0.05 (two-tailed).

A total of 830 individuals with T2DM were recruited, with 50.5% being male and an average age of 53.0 ± 8.3 years. The prevalence of MAFLD and SCAS were found to be 47.2% and 62.4%, respectively. Table 1 provides a comprehensive overview of the clinical characteristics among participants based on SII quartiles. Significant differences were observed across the SII quartiles in variables such as WC, BMI, SBP, DBP, TG, TC, HDL-c, LDL-c, UA, HOMA-IR, cIMT, and CT_L-S, as well as the prevalence of SCAS and MAFLD. Notably, higher SII quartiles were associated with elevated cIMT values and decreased CT_L-S values compared to the lower quartiles (P < 0.05). Moreover, there was a notable increase in the prevalence of both MAFLD and SCAS as the SII quartiles increased (P < 0.05).

Figure 1 depicts the association between SII, cIMT, and CT_L-S, as analyzed using Spearman correlation analysis. The results showed that CT_L-S was negatively correlated with SII (r=-0.407, P<0.001) and CT_L-S (r=-0.507, P<0.001). Conversely, a positive association was observed between SII and cIMT (r=0.424, P<0.001). To further investigate these correlations, weighted multiple linear regression analyses were conducted. As outlined in Table 2 . As outlined in Table 2 , CT_L-S remained negatively correlated with SII (β=-0.147, P<0.001) and cIMT (β=-0.243, P<0.001) even after conducting full adjustments in Model 3. Similarly, the positive correlation between SII and cIMT persisted (β=0.168, P<0.001).

In all three models, MAFLD contributed an independent variable for SCAS risk. The ORs for SCAS risk were 4.79 (95%CI: 3.30–6.97, P < 0.001) in Model 1, 2.12 (95%CI: 1.29–3.48, P < 0.001) in Model 2, and 2.15 (95%CI: 1.31–3.53, P < 0.001) in Model 3, respectively. Table 3 presents the findings of weighted binomial logistic regression analysis for the correlations of SII with SCAS and MAFLD risk. The results showed that higher quartiles of SII were positively associated with increased SCAS and MAFLD risk compared to the first quartile, across all three models (P<0.05). Meanwhile, Ln (SII) was independently correlated with increased SCAS and MAFLD risk (P < 0.05). In the fully adjusted Model 3, the ORs for SCAS and MAFLD risk were 2.26 (95% CI: 1.23–4.13, P=0.008) and 2.11 (95% CI: 1.14–3.92, P=0.017), respectively. Similar results were also observed when translating Ln (SII) into SII to increase by 1SD in this analysis. Additionally, a significant dose-response relationship was observed across all three models (P for trend <0.001). As shown in Figure 2 , the RCS analyses demonstrated a linear correlation of Ln (SII) with SCAS and MAFLD risk in Model 3 (P for nonlinearity >0.05).

The diagnostic performance of SII for MAFLD and SCAS was assessed by the ROC curves analysis. The results indicate that SII exhibits a favorable diagnostic value for MAFLD and SCAS. The AUCs (95%CI) of SII for identifying MAFLD and SCAS were 0.847 (0.819–0.874) and 0.741(0.708–0.774). The optimal cut-off values of SII were 518.8 (sensitivity: 52.7%, specificity: 87.2%) for SCAS and 399.1 (sensitivity: 84.7%, specificity: 79.7%) for MAFLD.

Figure 3 displays the mediation analysis examining the role of SII and Ln (SII) in the association between MAFLD and SCAS. The findings indicate that a significant positive indirect effect of MAFLD associated with SCAS through SII and Ln (SII) was observed after adjustment for Model 3. The proportion mediated effect was 7.8% for SII (P<0.001), and 10.9% for Ln (SII) (P<0.001).

As shown in Supplementary Tables 1, 2 , similar results were obtained when sensitivity analysis was performed in men. The findings revealed an inverse correlation between CT_L-S and SII (β=-0.160, P<0.001), as well as CT_L-S (β=-0.264, P<0.001). Similarly, the positive correlation between SII and cIMT remained consistent (β=0.159, P<0.001). Meanwhile, Ln (SII) exhibited an independent correlation with increased SCAS and MAFLD risk (P < 0.05). In the fully adjusted Model 3, the ORs for SCAS and MAFLD risk were 1.79(95% CI: 1.24–2.59, P=0.002) and 1.35 (95% CI: 1.17–1.56, P<0.001), respectively. Furthermore, the proportion mediated effects of SII and Ln (SII) were 9.9% (P<0.001) and 13.8% (P<0.001) in men ( Supplementary Figure 1 ).