Autonomic-inflammatory crosstalk in diabetic atherogenesis: a neuroimmune triad (HRV-LMR-hsCRP) predicts carotid plaque risk in type 2 diabetes Authors: Xinrui Zhou, Xiaowei Bai, Li Ding, Shuai Zhang, Ya Li Published: 2025 ABSTRACT Background and aimType 2 diabetes mellitus (T2D) is associated with a high risk of cardiovascular complications, including carotid atherosclerosis (CAS). This study aimed to investigate the link between CAS and T2D by identifying novel risk factors and examining the relationship between autonomic dysfunction (via heart rate variability, HRV) and systemic inflammation.MethodsWe conducted a retrospective observational study of 232 T2D patients, categorized into three groups based on carotid ultrasound: normal arteries (n = 47), intima-media thickening (n = 49), and carotid plaques (n = 136). Differences in clinical and inflammatory markers across groups were analyzed. Independent risk factors for CAS were identified using multivariate logistic regression, and a predictive model was developed and evaluated by Receiver Operating Characteristic (ROC) curve analysis. To infer causality, a bidirectional two-sample Mendelian randomization (MR) analysis was performed using summary-level data from large-scale genome-wide association studies (GWAS).ResultsSignificant differences were observed in age, insulin usage, and inflammatory markers (NLR, PLR, LMR) among the groups (all p < 0.05). Logistic regression identified age, SDNN (a measure of HRV), LMR, and hs-CRP as independent risk factors for CAS. The combined model integrating hs-CRP, SDNN, and LMR demonstrated exceptional predictive accuracy for CAS (AUC = 0.941; 95% CI: 0.912–0.969). MR analysis provided genetic evidence supporting a causal relationship between genetically predicted HRV reduction and increased carotid intima-media thickness.ConclusionOur findings underscore the critical interplay between autonomic dysfunction and inflammation in the development of CAS in T2D. The proposed integrative model shows high potential for risk stratification. Future large-scale multicenter studies are warranted to validate these findings and elucidate the underlying mechanisms.Supplementary InformationThe online version contains supplementary material available at 10.1186/s40842-025-00259-z. Introduction Type 2 diabetes mellitus (T2D), characterized by insulin resistance and β-cell dysfunction, leads to chronic hyperglycemia with distinct pathophysiological sequelae [1]. These metabolic disturbances drive both microvascular complications (retinopathy, nephropathy) and accelerated macrovascular disease progression, particularly in the context of atherosclerotic cardiovascular events [2, 3]. With global prevalence tripling since 2000 in parallel with obesity epidemics [4], T2D now accounts for 19.1% of global cardiovascular mortality, with accelerated growth in age-standardized rates since 2019 [5], necessitating innovative strategies to decode its vascular pathophysiology [6]. The synergistic interplay between chronic hyperglycemia and systemic inflammation creates a self-perpetuating cycle that exacerbates both β-cell failure and vascular endothelial dysfunction [7]. Furthermore, emerging evidence indicates that a neuroimmune connection is involved in diabetic vascular pathology [8]. In this context, autonomic dysfunction—measured by heart rate variability (HRV)—intensifies systemic inflammation through sympathetic overactivity and vagal withdrawal [9]. This neural‒immune crosstalk creates a self-perpetuating cycle, accelerating endothelial injury and plaque formation. However, clinical tools for quantifying neuroimmune dysregulation in atherosclerosis remain underdeveloped. Emerging evidence highlights the potential of inflammatory biomarkers as early diagnostic tools, underscoring the need for integrated clinical and research approaches [10]. Current T2D strategies emphasize drugs and lifestyle changes for glycemic control and cardiovascular health but often fail to prevent disease progression and complications [11]. Despite herbal remedies such as Corchorus olitorius that improve metabolic markers in T2D [12], insufficient attention has been given to the systemic cardiovascular impact in diabetic patients, as current evidence mainly relies on rodent studies. Recent studies have illuminated the association between T2D and carotid atherosclerosis, a significant predictor of cardiovascular events [13]. Carotid intima-media thickness (cIMT) and plaque formation have emerged as crucial markers for assessing cardiovascular risk in diabetic patients [13]. The T2D group exhibited a more pronounced cIMT increase, indicating a heightened risk for atherosclerosis and cardiovascular complications compared to the non-DM group [14, 15]. These findings emphasize the need for early diagnostic strategies that can identify vascular changes in T2D patients, facilitating timely intervention. Thus, it is crucial to identify easily accessible, repeatable, and minimally technician-influenced risk factors associated with carotid artery atherosclerosis in patients with T2D and to implement interventions to prevent carotid lesions and serious vascular complications. HRV analysis is a widely used noninvasive method to identify changes in the autonomic nervous system [16]. A decrease in HRV indicates early signs of cardiac autonomic neuropathy (CAN), which arises from dysregulation of both the parasympathetic and the sympathetic nervous systems [17]. The characteristics of decreased HRV include increased proinflammatory sympathetic outflow and reduced anti-inflammatory parasympathetic outflow [18]. Therefore, changes in HRV may also influence the degree of carotid atherosclerosis through inflammation. Currently, there are few studies on the correlation between carotid atherosclerosis and HRV in T2D patients. On the other hand, elevated levels of various inflammatory markers associated with systemic inflammation are positively correlated with the degree of carotid atherosclerosis [19], and their progression may increase the risk of mortality [20]. However, most current studies have focused on the correlation between carotid atherosclerosis and inflammatory markers in the general population [21], whereas research specific to T2D patients is relatively lacking. Most prior studies have been observational, and owing to the limitations of confounding factors and sample size in observational studies, the accuracy of the results can be affected. This study included 232 patients with TD2 in a retrospective analysis, with groups compared based on carotid ultrasound and dynamic electrocardiography. The primary aim was to examine risk factors for carotid atherosclerosis in T2D patients and to assess the relationships between HRV and inflammatory biomarkers. To extend beyond observational associations and infer causality, Mendelian randomization (MR) was applied—a method that uses genetic variants as instrumental variables to address confounding and reverse causation, thereby offering more reliable causal evidence [22]. By integrating clinical measurements with genetic causal inference, this study seeks to uncover potential therapeutic targets and guide prevention strategies for cardiovascular complications in T2D. Materials and methods Study patients and study design This was a retrospective observational study with 232 T2D patients hospitalized in the Department of Endocrinology and Metabolism and the Department of Cardiology at the First Affiliated Hospital of Xi’an Medical University from April 2021 to September 2023. The inclusion criteria were as follows: (1) meeting the diagnostic criteria for T2D per the China guidelines (2020) [23] and (2) completing carotid ultrasound, 24-hour electrocardiogram measurement, and full medical documentation. The exclusion criteria were as follows: (1) diabetes-related factors: non-T2D diabetes (type 1/special type) or pregnancy; acute complications (diabetic ketoacidosis, hyperosmolar coma, lactic acidosis); (2) cardiovascular/neurological conditions: acute cardiovascular/cerebrovascular events, arrhythmias, or cardiac pacemaker implantation; (3) organ dysfunction: severe hepatic/renal impairment (ALT/AST ≥ 3× ULN); interstitial/obstructive lung diseases, dialysis-dependent renal failure; (4) systemic disorders: active malignancies, hematologic malignancies, acute/chronic inflammatory/autoimmune diseases; (5) metabolic/laboratory abnormalities: thyroid/parathyroid dysfunction, serum potassium imbalance; (6) behavioral/neurological factors: alcohol/caffeine addiction; mental/cognitive disorders; and (7) structural abnormalities: carotid artery stenosis >50%. In total, 69 patients with T2D and acute inflammatory conditions were excluded: 15 with diabetic ketoacidosis, 5 with hyperosmolar coma, 243 with acute cardiovascular diseases, 33 with arrhythmia, 3 with implanted cardiac pacemakers, 38 with abnormal thyroid function, 10 with carotid artery stenosis, 5 with tumors, 12 with pulmonary interstitial fibrosis, and 28 with abnormal serum potassium levels. All participants were subjected to a standard baseline protocol that included a thorough clinical‒laboratory evaluation. Figure 1 displays the patient selection process used in this study. Fig. 1Flowchart of patient enrollment. NCA group: normal carotid artery group (n = 47); CAT group: carotid artery intima‒media thickness group (n = 49); CAP group: carotid artery plaque group (n = 136) Flowchart of patient enrollment. NCA group: normal carotid artery group (n = 47); CAT group: carotid artery intima‒media thickness group (n = 49); CAP group: carotid artery plaque group (n = 136) The general information of the patient included age, sex, history of smoking and alcohol consumption, body mass index (BMI in kg/m²), duration of diabetes (in years), family history of diabetes, and relevant medical history, such as hypertension, coronary heart disease (CHD), cerebral infarction, previous percutaneous coronary intervention (PCI), and readings of both systolic (SBP) and diastolic blood pressure (DBP). After admission, the patient rests for 30 min before a nurse measures blood pressure with a mercury sphygmomanometer, with additional readings at 6:00 AM on days 2 and 3, averaging three measurements for the study. CHD diagnosis followed the 1979 WHO criteria, requiring a history of myocardial infarction, typical angina, or significant coronary artery stenosis (> 50%). A history of hypertension was defined as either a documented prior diagnosis in the patient’s medical record or, in the absence of antihypertensive medication, having three separate resting SBP readings of ≥ 140 mmHg and/or DBP readings of ≥ 90 mmHg obtained on different days prior to enrollment. The cerebral infarction diagnosis followed Chinese guidelines, with a history of infarction and confirmed CT or MRI findings while other conditions were excluded. PCI indicates prior angioplasty for CHD. The height and weight of the nursing staff were measured to calculate BMI via the following formula: BMI = weight (kg)/height (m)². Baseline measurements All enrolled patients underwent a blood test after fasting overnight for 8 h. Venous blood samples were collected to assess routine blood parameters, liver function, kidney function, coagulation, and lipid profiles. The results of the routine blood tests were then used to calculate the neutrophil-to-lymphocyte ratio (NLR), lymphocyte-to-monocyte ratio (LMR), platelet-to-lymphocyte ratio (PLR), and systemic immune-inflammation index (SII) values, which are defined as follows: NLR = neutrophil count/lymphocyte count; LMR = lymphocyte count/monocyte count; PLR = platelet count/lymphocyte count; and SII = platelet count × (neutrophil count/lymphocyte count). 24-h ambulatory ECG measurement All patients underwent a 24-hour Holter electrocardiogram (ECG) examination with 12 leads (MMROC-H, M&B Electronic Instruments, Beijing, China). On the day of the test, the participants were requested to maintain their normal diet and to avoid strenuous physical activity, alcohol, smoking, and caffeinated beverages. Heart rate variability (HRV) was analyzed in both the time domain and the frequency domain via computer software. Time domain measurements consisted of the following: (i) SDNN, which is the standard deviation of all recorded normal RR intervals and indicates primarily sympathetic nervous activity; (ii) SDANN, the standard deviation of average normal NN intervals calculated over 5-minute intervals within a 24-hour period; (iii) RMSSD, the square root of the mean squared differences between adjacent RR intervals, indicating vagal nerve activity; and (iv) PNN50, the percentage of heartbeats where the difference between adjacent RR intervals exceeds 50 ms, indicating vagal nerve activity. The frequency domain measurements include (i) high-frequency power (HF), which ranges from 0.15 to 0.4 Hz and indicates vagal nerve tension; (ii) low-frequency power (LF), which spans 0.04 to 0.15 Hz and reflects the combined influence of both the sympathetic and vagal nervous systems; (iii) very low-frequency power (VLF), which ranges from 0.003 to 0.04 Hz; (iv) total power (TP); and (v) the LF/HF ratio, which effectively indicates the balance between sympathetic and vagal regulation. Low heart rate variability (HRV) is defined as a value falling below the 25th percentile [24]. To ensure the reliability and consistency of HRV measurements throughout this study, a standardized protocol was implemented, complemented by a quality assurance process. All HRV analyses were conducted using the same version of dedicated analysis software to maintain uniformity in data processing. The analysis adhered to consistent, pre-defined protocols for both time-domain and frequency-domain assessments. A key quality control measure involved the evaluation of the signal-to-noise ratio for each data segment to ensure the integrity of the signals included in the final analysis. The quality assurance framework was applied to validate the entire analytical procedure. Inter-observer reliability for critical artifact detection was high, with a kappa (κ) coefficient of 0.92, indicating strong agreement between independent analysts. Furthermore, the intra-class correlation coefficient (ICC) for all key HRV parameters exceeded 0.95. Finally, to confirm the physiological plausibility of our findings, the derived HRV parameters were systematically validated against established normative values available in published literature. Carotid ultrasound and cIMT measurement The examination was conducted by experienced ultrasound physicians via the GE S8 color ultrasound diagnostic system (GE HealthCare, US), with the subjects resting for 15–20 min. The carotid artery was examined by turning the head 45 degrees, placing the probe behind the mandible, and capturing longitudinal images. The posterior wall displayed two low echogenic lines indicating the lumen‒intima and media‒adventitia interfaces, with the vertical distance representing the intima‒media thickness. Measurements were taken at the end of the cardiac cycle to assess arterial smoothness, thickening, plaque, and lumen narrowing. Patients were categorized into three groups on the basis of their intima‒media thickness [19]: below 1.0 mm, 1.0–1.2 mm indicates thickening, and 1.3 mm or greater suggests plaque formation. Statistical analysis For non-normally distributed continuous data, the median and interquartile range (IQR) were used for description, and group differences were assessed using the Mann-Whitney U test (for two groups) or the Kruskal-Wallis H test (for three or more groups). Categorical variables were described as frequencies and percentages, and comparisons were made using the Chi-square test. Spearman correlation was used to analyze bivariate associations, whereas logistic regression was used to identify risk factors for carotid atherosclerosis in T2D patients. ROC curve analysis was used to evaluate the predictive power of the SDNN, LMR, and hs-CRP, and the AUC values for clinical diagnosis were compared, with p < 0.05 considered significant. To ensure the robustness of the combined predictive model, multicollinearity among the key predictors (hs-CRP, SDNN, LMR) was assessed prior to model construction using Variance Inflation Factor (VIF), Tolerance, and Condition Index diagnostics. All analyses were performed via SPSS 25. Mendelian randomization studies To provide a comprehensive analytical framework, our study comprises two main parts: (1) conventional observational analyses of clinical and biomarker data, and (2) genetic causal inference using Mendelian randomization (MR). This section describes the second part. We conducted a bidirectional two-sample MR analysis using Genome-Wide Association Study (GWAS) data to explore causal links between inflammatory proteins, heart rate variability, and carotid atherosclerosis. This approach helps address issues of confounding and reverse causation inherent in observational studies. Although the GWAS data were not derived from a specific type 2 diabetes (T2D) population, they provide valuable insights into general disease mechanisms. Methodologically, we applied five MR methods, with the inverse variance weighted (IVW) method serving as the primary analysis. Sensitivity analyses were performed to assess horizontal pleiotropy and heterogeneity, ensuring the robustness of our findings. Our findings revealed a significant causal relationship between heart rate variability (SDNN, RMSSD, and RMSSDc) and carotid intima–media thickness (cIMT). RMSSDc or SDNNc stands for calculating HRV values corrected for the influence of the mean interbeat interval [25]. We further hypothesized that circulating inflammatory proteins might mediate the relationship between HRV and cIMT. However, mediation analysis did not support this pathway; while three inflammatory factors were significantly associated with cIMT, they did not act as intermediaries in the causal pathway between HRV and cIMT. Data sources Heart rate variability data from 46,075 Europeans in the UK Biobank included SDNN, SDNNc, RMSSD, and RMSSDc (corresponding to GWAS accession numbers GCST90281263 to GCST90281266) [25]. The carotid atherosclerosis data used cIMT from 45,185 participants from the UK Biobank (GWAS accession number GCST90100574) [26]. Genetic data for circulating inflammatory proteins were obtained from a meta-analysis of 14,824 Europeans, which analyzed 91 inflammatory proteins (accession numbers GCST90274758 to GCST90274848) [27]. Selection of instrumental variables (IVs) We set a threshold of 5 × 10− 8 to filter SNPs associated with heart rate variability, grouping genetic differences within 10,000 kb on the basis of LD (r2 = 0.001). Palindromic SNPs and those with p = 5 × 10− 8 were excluded, and F statistics were calculated to assess each SNP’s strength related to exposure, retaining only those with F > 10 for MR analysis. SNPs linked to inflammatory proteins were selected at p < 1 × 10 − 5. Statistical analysis We employed five MR methods—IVW, MR‒Egger, simple mode, weighted median, and weighted mode—to explore the causal link between heart rate variability and carotid atherosclerosis. The IVW method effectively estimates causal effects and addresses heterogeneity, which is evaluated via Cochran’s Q test for each SNP. A random effects IVW model was used for heterogeneous SNPs (p < 0.05), whereas a fixed effects model was used for those with p values greater than 0.05. To check for horizontal pleiotropy, we used the MR-PRESSO global test and the MR‒Egger intercept test. We concluded that there was no pleiotropy if the p value exceeded 0.05. Leave-one-out analysis was used to assess the influence of individual SNP drivers on the random estimates. Scatter plots and funnel plots were generated to further validate the robustness of the results. Additionally, steer directionality tests and reverse MR analyses were conducted to determine the direction of the causal relationship between exposure and outcome. When anomalies were detected, reverse causation was assumed. To identify the inflammatory factors that act as mediators, we performed two-step MR analyses on significant inflammatory proteins to distinguish between direct and indirect effects. The first step involved identifying inflammatory factors with causal effects on carotid atherosclerosis; the second step involved analyzing the causal relationships between heart rate variability and the selected inflammatory factors. Data analysis was conducted via R (version 4.3.3) with the TwoSampleMR (version 0.5.7) and MRPRESSO (version 1.0) packages. Results Demographic and baseline clinical characteristics of patients This study involved 232 T2D patients divided into three groups on the basis of carotid artery status: 47 in the normal carotid artery (NCA) group, 49 in the carotid artery intima‒media thickness (CAT) group, and 136 in the carotid artery plaque (CAP) group. Potential confounders such as age (p < 0.001), coronary heart disease (CHD) history (p = 0.031), and cerebral infarction history (p = 0.018), and insulin use (p = 0.043) showed significant differences among groups (Supplementary Table 1) and were adjusted for in multivariate analyses. However, we did not explicitly adjust for all medication types due to retrospective data limitations. Comparison of inflammatory data among T2D patients in the NCA, CAT and CAP groups The comparison of inflammatory data among T2D patients with normal carotid arteries (the NCA group), carotid intima‒media thickening (the CAT group), and carotid plaque groups (the CAP group) revealed significant differences (p < 0.001) in the following parameters: neutrophil count (N), lymphocyte count (L), platelet count (PLT), neutrophil‒lymphocyte ratio (NLR), platelet‒lymphocyte ratio (PLR), lymphocyte‒monocyte ratio (LMR), systemic immune‒inflammation index (SII), and high-sensitivity C-reactive protein (hs-CRP). In contrast, there were no statistically significant differences in white blood cell count (WBC) (p = 0.240) or monocyte count (M) (p = 0.235) among the three groups (Fig. 2, Supplementary Table 2). Fig. 2Comparison of inflammatory marker levels among T2D patients stratified by carotid Intima-media thickness. Study participants were categorized into three groups: those with normal carotid arteries (NCA group, n = 47), those with carotid intima-media thickening (CAT group, n = 49), and those with carotid plaque (CAP group, n = 136). N: neutrophil; L: lymphocyte; hs-CRP: high-sensitivity C-reactive protein; NLR: neutrophil-to-lymphocyte ratio; LMR: lymphocyte-to-monocyte ratio; PLR: platelet-to-lymphocyte ratio; SII: systemic immune-inflammation index. **p < 0.001 Comparison of inflammatory marker levels among T2D patients stratified by carotid Intima-media thickness. Study participants were categorized into three groups: those with normal carotid arteries (NCA group, n = 47), those with carotid intima-media thickening (CAT group, n = 49), and those with carotid plaque (CAP group, n = 136). N: neutrophil; L: lymphocyte; hs-CRP: high-sensitivity C-reactive protein; NLR: neutrophil-to-lymphocyte ratio; LMR: lymphocyte-to-monocyte ratio; PLR: platelet-to-lymphocyte ratio; SII: systemic immune-inflammation index. **p < 0.001 Comparison of biochemical data among T2D patients in the NCA, CAT and CAP groups The biochemical data comparison revealed statistically significant differences among the three groups for aspartate transaminase (AST, p = 0.035), creatinine (CR, p = 0.002), D-dimer (p = 0.003), and HbA1c (p = 0.011). On the other hand, no statistically significant differences were detected in red blood cell (RBC, p = 0.186), red cell distribution width standard deviation (RDW-SD, p = 0.101), Hemoglobin (Hb, p = 0.232), alanine aminotransferase (ALT, p = 0.649), albumin (ALB, p = 0.952), direct bilirubin (DBIL, p = 0.073), indirect bilirubin (IBIL, p = 0.334), total cholesterol (TC, p = 0.636), triglyceride (TG, 0.275), LDL-Cholesterol (LDL-C, p = 0.818), HDL-Cholesterol (HDL-C, p = 0.138), or fibrinogen (FIB, p = 0.054) among the three groups (Supplementary Table 3). Comparison of time-domain and frequency-domain indicators of heart rate variability among T2D patients in the NCA, CAT and CAP groups The analysis of HRV indicators across the NCA, CAT, and CAP groups revealed that only the SDNN was statistically significant (p = 0.012). The other indicators—SDANN (p = 0.700), RMSSD (p = 0.435), PNN50 (p = 0.098), TP (p = 0.950), VLF (p = 0.797), LF (p = 0.531), HF (p = 0.597), and LF/HF (p = 0.413)—did not reach statistical significance (Fig. 3, Supplementary Table 4). Fig. 3Violin plots showing the distribution of SDNN values across carotid artery subgroups in patients with T2D. The standard deviation of the NN interval (SDNN), a heart rate variability (HRV) parameter, was significantly different among the normal carotid artery (NCA), carotid intima-media thickening (CAT), and carotid plaque (CAP) groups. *p < 0.05 Violin plots showing the distribution of SDNN values across carotid artery subgroups in patients with T2D. The standard deviation of the NN interval (SDNN), a heart rate variability (HRV) parameter, was significantly different among the normal carotid artery (NCA), carotid intima-media thickening (CAT), and carotid plaque (CAP) groups. *p < 0.05 Correlation analysis of carotid atherosclerosis in T2D patients with other indicators We merged the carotid artery media thickness group and the carotid plaque group into a single carotid atherosclerosis group and then analyzed its correlation with various variables in T2D patients. Spearman correlation analysis was applied to nonnormally distributed and ordinal data. The results indicated that carotid atherosclerosis was positively correlated with the following indicators (r > 0.5): the SII (r = 0.658, p < 0.001), NLR (r = 0.601, p < 0.001), and PLR (r = 0.533, p < 0.001). Additionally, there was a negative correlation with L (r =-0.395, p < 0.001), the LMR (r =-0.302, p < 0.001), and the SDNN (r =-0.175, p = 0.007) (Supplementary Table 5). Logistic regression analysis of risk factors for carotid atherosclerosis in T2D patients We conducted a logistic regression analysis with carotid atherosclerosis as the dependent variable, including factors such as age, CHD, cerebral infarction, insulin, SDNN, and inflammatory indicators (N, L, PLT, NLR, PLR, LMR, SII, hs-CRP), D-D, HbA1c, and creatinine. The results indicated that hs-CRP, the LMR, the SDNN, and age are significant risk factors for carotid atherosclerosis in patients with T2D (p < 0.05). The odds ratios (ORs) and 95% confidence intervals (CIs) for these factors are as follows: hs-CRP, 3.000 (1.305–6.894); LMR, 1.502 (1.127–2.002); SDNN, 1.028 (1.006–1.050); and age, 1.088 (1.027–1.151) (Supplementary Table 6). While we acknowledge the potential for residual confounding from not adjusting for specific medications, this was necessitated by practical constraints. Retrospective data on drug dosage and duration were too heterogeneous to model accurately. Instead, we incorporated proxy adjustments for the conditions these medications treat (hypertension and CHD history). Notably, these proxy variables were well-balanced across groups (Supplementary Table 1; p > 0.05 for most), reducing the likelihood that medication differences are a major source of confounding. ROC curve for diagnosing carotid atherosclerosis in T2D patients We used carotid atherosclerotic arteries as the study group (merged CAT and CAP groups) and normal carotid arteries as the control group (NCA group). ROC curves were used to assess the diagnostic effectiveness of the risk factors SDNN, LMR, and hs-CRP, both individually and in combination, for detecting carotid atherosclerosis in patients with T2D. The areas under the curve (AUCs) for the SDNN, LMR, and hs-CRP were 0.638 (95% CI: 0.561–0.716), 0.620 (95% CI: 0.534–0.706), and 0.750 (95% CI: 0.688–0.812), respectively. The AUC for SDNN + LMR + hs-CRP was 0.941 (95% CI: 0.912–0.969). Table 1 summarizes the diagnostic parameters of the ROC curve for carotid atherosclerosis in T2D patients. These findings suggest that the combined use of the SDNN, LMR, and hs-CRP for diagnosing carotid atherosclerosis in T2D patients is more effective than the use of any of these factors alone (Fig. 4). Multicollinearity diagnostics confirmed the suitability of combining these variables, with all VIF values close to 1 (range: 1.012–1.026), Tolerance values > 0.97, and a maximum Condition Index of 6.84, all well within acceptable thresholds indicating no significant multicollinearity. Fig. 4ROC curve for diagnosing carotid atherosclerosis in T2D patients. hs-CRP: high-sensitivity C-reactive protein; SDNN: standard deviation of normal R-R intervals; LMR: lymphocyte-to-monocyte ratio ROC curve for diagnosing carotid atherosclerosis in T2D patients. hs-CRP: high-sensitivity C-reactive protein; SDNN: standard deviation of normal R-R intervals; LMR: lymphocyte-to-monocyte ratio Table 1Parameters of the ROC curve for diagnosing carotid atherosclerosis in T2D patientshs-CRPSDNNLMRSDNN + LMR + hs-CRPAUC (95%CI)0.750(0.688–0.812)0.638(0.561–0.716)0.620(0.534–0.706)0.941(0.912–0.969)S.E0.0320.0390.0440.015Sensitivity0.5240.0760.0650.838Specificity0.9790.9360.9790.936Cutoff value1.555172.028.6830.799p value0.0017< 0.0001< 0.0001< 0.0001 Table 1Parameters of the ROC curve for diagnosing carotid atherosclerosis in T2D patientshs-CRPSDNNLMRSDNN + LMR + hs-CRPAUC (95%CI)0.750(0.688–0.812)0.638(0.561–0.716)0.620(0.534–0.706)0.941(0.912–0.969)S.E0.0320.0390.0440.015Sensitivity0.5240.0760.0650.838Specificity0.9790.9360.9790.936Cutoff value1.555172.028.6830.799p value0.0017< 0.0001< 0.0001< 0.0001 Parameters of the ROC curve for diagnosing carotid atherosclerosis in T2D patients 0.750 (0.688–0.812) 0.638 (0.561–0.716) 0.620 (0.534–0.706) 0.941 (0.912–0.969) Analysis of inflammatory data between the low-HRV group and non-low-HRV group among T2D patients in the NCA group, CAT group and CAP group The HRV indicators related to carotid atherosclerosis in T2D patients were divided into a low HRV group and a non-low HRV group on the basis of the < 25th percentile. In the normal carotid group, there was a significant difference in the NLR (p = 0.008) between the low HRV group and the non-low HRV group. In the carotid intima–media thickening group, there were significant differences (p < 0.001) in indicators such as L, the NLR, the PLR, the LMR, and the SII between the low heart rate variability group and the non-low heart rate variability group. In the carotid plaque group, significant differences (p < 0.001) were observed between the low heart rate variability group and the non-low heart rate variability group in L, NLR, PLR, LMR, SII, and hs-CRP (p = 0.002) (Table 2). Table 2Comparison of inflammatory indicators between the low HRV group and non-low HRV group among T2D patients in the NCA group, CAT group and CAP groupThe normal carotid arteries (NCA) groupIndicatorLow HRV groupNon-low HRV groupZ/t/χ2 p WBC (109/L)6.55 [5.49,7.43]7.11 [5.85,8.69]-0.760.446N (109/L)4.26 ± 0.613.94 ± 1.171.370.248M (109/L)0.41 [0.29,0.62]0.53 [0.38,0.65]-0.990.321L (109/L)1.65 ± 0.352.44 ± 0.772.980.091PLT (109/L)195.00 ± 26.80197.40 ± 42.431.050.310NLR (109/L)2.64 [2.32,2.93]1.65 [1.27,1.91]-2.670.008PLR123.65 [86.00,165.08]81.50 [66.54,105.66]-1.830.067LMR4.56 [2.92,4.66]4.47 [3.78,6.03]-0.650.517SII519.33 ± 135.52]339.15 ± 139.81]0.0620.805hs-CRP (mg/L)0.50 [0.50,0.50]0.50 [0.50,1.75]-1.070.284The carotid intima-media thickening (CAT) groupIndicatorLow HRV groupNon-low HRV groupZ/t/χ2 p WBC (109/L)6.55 [5.49,7.43]7.11 [5.85,8.69]-1.530.125N (109/L)5.02 ± 0.754.86 ± 1.51-0.300.762M (109/L)0.50 [0.44,0.61]0.40 [0.35,0.55]-2.020.43L (109/L)1.21 [1.05,1.87]2.75 [2.01,4.10]-4.17< 0.001PLT (109/L)180.17 ± 52.59205.84.40 ± 60.290.6530.423NLR (109/L)4.10 [2.62,4.80]1.64 [1.26,2.05]-4.56< 0.001PLR142.02 [114.85,175.08]67.96 [50.35,1,2.38]-3.49< 0.001LMR2.52 [1.96,3.20]6.27 [4.72,8.82]-4.44< 0.001SII761.54 [631.62,831.07]307.19 [254.51,441.50]-3.74< 0.001hs-CRP (mg/L)0.50 [0.50,0.50]0.50 [0.50,0.50]-0.630.531The carotid plaque (CAP) groupIndicatorLow HRV groupNon-low HRV groupZ/t/χ2 p WBC (109/L)7.11 ± 1.666.45 ± 1.710.000.988N (109/L)6.89 ± 1.595.44 ± 1.371.330.252M (109/L)0.48 [0.36,0.59]0.41 [0.37,0.57]-0.9500.342L (109/L)1.28 [0.98,1.48]1.84 [1.52,2.12]-6.05< 0.001PLT (109/L)242.00 [209.25,277.00]235.00 [207.00,268.00]-0.990.320NLR (109/L)5.24 [4.25,7.44]2.97 [2.37,3.56]-7.67< 0.001PLR192.91 [158.00,253.25]133.62 [102.79,160.69]-5.99< 0.001LMR2.62 [1.98,3.12]3.88 [3.24,5.01]-6.09< 0.001SII1348.12 [1058.65,1670.79]693.91 [535.50,942.29]-7.48< 0.001hs-CRP (mg/L)4.92 [1.43,6.21]2.02 [0.05,4.43]-3.160.002WBC: white blood cell; N: neutrophil; M: monocyte; L: lymphocyte; PLT: platelet; NLR: neutrophil-to-lymphocyte ratio; PLR: platelet-to-lymphocyte ratio; LMR: lymphocyte-to-monocyte ratio; SII: systemic immune-inflammation index; hs-CRP: high-sensitivity C-reactive protein. Skewed variables are denoted by medians [interquartile ranges], and normally distributed variables are denoted by means ± standard deviations Table 2Comparison of inflammatory indicators between the low HRV group and non-low HRV group among T2D patients in the NCA group, CAT group and CAP groupThe normal carotid arteries (NCA) groupIndicatorLow HRV groupNon-low HRV groupZ/t/χ2 p WBC (109/L)6.55 [5.49,7.43]7.11 [5.85,8.69]-0.760.446N (109/L)4.26 ± 0.613.94 ± 1.171.370.248M (109/L)0.41 [0.29,0.62]0.53 [0.38,0.65]-0.990.321L (109/L)1.65 ± 0.352.44 ± 0.772.980.091PLT (109/L)195.00 ± 26.80197.40 ± 42.431.050.310NLR (109/L)2.64 [2.32,2.93]1.65 [1.27,1.91]-2.670.008PLR123.65 [86.00,165.08]81.50 [66.54,105.66]-1.830.067LMR4.56 [2.92,4.66]4.47 [3.78,6.03]-0.650.517SII519.33 ± 135.52]339.15 ± 139.81]0.0620.805hs-CRP (mg/L)0.50 [0.50,0.50]0.50 [0.50,1.75]-1.070.284The carotid intima-media thickening (CAT) groupIndicatorLow HRV groupNon-low HRV groupZ/t/χ2 p WBC (109/L)6.55 [5.49,7.43]7.11 [5.85,8.69]-1.530.125N (109/L)5.02 ± 0.754.86 ± 1.51-0.300.762M (109/L)0.50 [0.44,0.61]0.40 [0.35,0.55]-2.020.43L (109/L)1.21 [1.05,1.87]2.75 [2.01,4.10]-4.17< 0.001PLT (109/L)180.17 ± 52.59205.84.40 ± 60.290.6530.423NLR (109/L)4.10 [2.62,4.80]1.64 [1.26,2.05]-4.56< 0.001PLR142.02 [114.85,175.08]67.96 [50.35,1,2.38]-3.49< 0.001LMR2.52 [1.96,3.20]6.27 [4.72,8.82]-4.44< 0.001SII761.54 [631.62,831.07]307.19 [254.51,441.50]-3.74< 0.001hs-CRP (mg/L)0.50 [0.50,0.50]0.50 [0.50,0.50]-0.630.531The carotid plaque (CAP) groupIndicatorLow HRV groupNon-low HRV groupZ/t/χ2 p WBC (109/L)7.11 ± 1.666.45 ± 1.710.000.988N (109/L)6.89 ± 1.595.44 ± 1.371.330.252M (109/L)0.48 [0.36,0.59]0.41 [0.37,0.57]-0.9500.342L (109/L)1.28 [0.98,1.48]1.84 [1.52,2.12]-6.05< 0.001PLT (109/L)242.00 [209.25,277.00]235.00 [207.00,268.00]-0.990.320NLR (109/L)5.24 [4.25,7.44]2.97 [2.37,3.56]-7.67< 0.001PLR192.91 [158.00,253.25]133.62 [102.79,160.69]-5.99< 0.001LMR2.62 [1.98,3.12]3.88 [3.24,5.01]-6.09< 0.001SII1348.12 [1058.65,1670.79]693.91 [535.50,942.29]-7.48< 0.001hs-CRP (mg/L)4.92 [1.43,6.21]2.02 [0.05,4.43]-3.160.002WBC: white blood cell; N: neutrophil; M: monocyte; L: lymphocyte; PLT: platelet; NLR: neutrophil-to-lymphocyte ratio; PLR: platelet-to-lymphocyte ratio; LMR: lymphocyte-to-monocyte ratio; SII: systemic immune-inflammation index; hs-CRP: high-sensitivity C-reactive protein. Skewed variables are denoted by medians [interquartile ranges], and normally distributed variables are denoted by means ± standard deviations Comparison of inflammatory indicators between the low HRV group and non-low HRV group among T2D patients in the NCA group, CAT group and CAP group WBC: white blood cell; N: neutrophil; M: monocyte; L: lymphocyte; PLT: platelet; NLR: neutrophil-to-lymphocyte ratio; PLR: platelet-to-lymphocyte ratio; LMR: lymphocyte-to-monocyte ratio; SII: systemic immune-inflammation index; hs-CRP: high-sensitivity C-reactive protein. Skewed variables are denoted by medians [interquartile ranges], and normally distributed variables are denoted by means ± standard deviations MR results The results of the MR analysis revealed a significant positive correlation between the gene-predicted heart rate variability in three of the four groups and carotid atherosclerosis (Supplementary Table 7, Supplementary Fig. 1). Furthermore, the direction of the β estimates reflecting the causal relationships among these traits was consistent across the five MR methods, enhancing the robustness of our findings. The analysis revealed no heterogeneity among the data, as the Cochran test was not statistically significant (p > 0.05). Both the MR-PRESSO global test and MR‒Egger intercept test p values were greater than 0.05, indicating that there was no significant horizontal pleiotropy bias. The leave-one-out plot revealed that no single SNP affected the MR results (Supplementary Fig. 2). Forest plots (Supplementary Fig. 3) and funnel plots (Supplementary Fig. 4) also confirmed the stability of the results. These findings consistently support the idea of a significant causal relationship between genetically predicted outcomes. We further conducted reverse MR analyses and Steiger directionality tests on cIMT and HRV. No significant reverse causation was found, indicating that cIMT does not have a causal effect on HRV. Causal relationship analysis between inflammatory factors and cIMT MR analysis revealed that three inflammatory factors exhibit a significant causal relationship with cIMT. CCL11 has a slight risk-reducing effect on cIMT, whereas TNFSF10 and TSLP may increase the risk of cIMT (Supplementary Table 8). Causal relationship analysis between inflammatory factors and HRV When inflammatory factors were used as exposure factors and three indicators of heart rate variability (HRV) were used as outcomes for MR analysis, a significant causal relationship between inflammatory factors and HRV was found (Supplementary Table 9). CXCL10, CD244, and TGFA are protective factors for HRV, whereas IL5, FGF19, and CCL19 are risk factors for HRV. We further conducted reverse MR analysis of inflammatory factors and HRV and found no significant reverse causal relationship, indicating that HRV does not have a causal effect on inflammation. Discussion Our study identifies autonomic dysfunction, reflected by reduced heart rate variability (SDNN), and systemic inflammation, characterized by alterations in the lymphocyte-to-monocyte ratio (LMR) and high-sensitivity C-reactive protein (hs-CRP), as independent predictors of carotid atherosclerosis in T2D patients. The ROC model integrating these biomarkers demonstrated exceptional discriminative capacity (AUC = 0.941), supporting its potential as a cost-effective tool for identifying high-risk individuals requiring intensified vascular surveillance. The rationale for integrating these specific markers into a single model is underpinned by their complementary biological plausibility representing key pathways in diabetic atherogenesis: autonomic dysfunction (reflected by reduced SDNN) promotes systemic inflammation (e.g., elevated hs-CRP), while the LMR serves as an indicator of immune balance. This neuro-immune triad aligns with established evidence that sympathetic overactivity enhances pro-inflammatory pathways, thereby accelerating endothelial injury. Statistically, the combined model’s superior discriminatory power, alongside the absence of significant multicollinearity as confirmed by VIF diagnostics, justifies their simultaneous inclusion to assess their unique contributions to predicting carotid atherosclerosis. While the sample size is limited, the statistical significance of our findings remains robust. To address concerns regarding causality and unmeasured confounding, we further employed Mendelian randomization (MR) analysis—a method that leverages genetic instruments to infer causal relationships—using large-scale GWAS data. This approach complements our observational results and strengthens the inference of causality. Autonomic dysfunction, indicated by reduced heart rate variability (HRV), contributes to carotid structural changes in type 2 diabetes primarily through a neuro-immune mechanism. This involves an imbalance characterized by sympathetic overactivation and parasympathetic withdrawal, which amplifies systemic inflammation via increased release of pro-inflammatory cytokines such as IL-6 and TNF-α [28]. This inflammatory state promotes endothelial dysfunction, oxidative stress, and increased plaque vulnerability. Genetic evidence from Mendelian randomization analysis supports a causal role of HRV reduction in increased carotid intima-media thickness, with specific inflammatory mediators like CXCL10 and CCL11 implicating Th1-driven immune dysregulation as an intermediary. This pathway is consistent with the inflammatory reflex theory, wherein autonomic dysfunction disrupts the cholinergic anti-inflammatory pathway [8]. Notably, LMR emerged as a novel inflammatory indicator of carotid atherosclerosis. Reduced lymphocyte counts alongside elevated monocytes suggest enhanced myeloid differentiation into plaque foam cells—a hallmark of atherogenesis [29]. This positions LMR as a potential surrogate for monocyte-driven plaque instability, complementing conventional biomarkers like hs-CRP. Despite the cross-sectional design and limited sample size, the consistency between LMR’s predictive performance and its biological plausibility—coupled with genetic evidence from MR—supports its potential utility as a cost-effective inflammatory indicator in high-risk populations. Mendelian randomization analyses provided genetic validation of our observations. Bidirectional MR confirmed a causal relationship between reduced HRV and carotid atherosclerosis, independent of traditional metabolic confounders. Chemokines implicated in vascular remodeling (e.g., CCL11, TNFSF10) [30] were causally linked to plaque progression, supporting Th1-mediated immune dysregulation. Furthermore, inflammatory mediators such as CXCL10 showed strong causal associations with HRV reduction (Supplementary Table 9), indicating persistent neuro-immune crosstalk beyond glycemic control. Though based on general-population GWAS data, the consistency of these MR results with our T2D cohort suggests that (1) inflammatory-HRV interactions represent fundamental biological mechanisms operative across metabolic states, and (2) heightened atherosclerosis risk in T2D may arise from hyperglycemia-driven epigenetic modifications that amplify these baseline pathways [15]. Several limitations must be acknowledged. Our study is retrospective and single-center, with a modest sample size that may affect generalizability. Biomarkers were assessed at a single time point, limiting insights into temporal dynamics. Although we employed MR to strengthen causal inference, the use of general-population GWAS data—rather than diabetes-specific datasets—may not fully capture T2D-specific epigenetic modifications. Future longitudinal studies with larger, multi-ethnic cohorts and diabetes-focused omics data are needed to validate and extend our findings. Conclusions This study identifies autonomic dysfunction, indicated by reduced SDNN, and systemic inflammation, characterized by changes in the lymphocyte-to-monocyte ratio (LMR) and high-sensitivity C-reactive protein (hs-CRP), as strong and independent predictors of carotid atherosclerosis in patients with type 2 diabetes (T2D). When combined into a diagnostic model, these factors demonstrate remarkable discriminative power, with an area under the curve (AUC) of 0.941, showcasing their potential for effective and cost-efficient risk stratification in clinical settings. Additionally, our findings suggest a harmful neuroimmune cycle where an imbalance between sympathetic and parasympathetic activity increases inflammation, leading to greater endothelial damage and plaque instability. This relationship is further supported by Mendelian randomization analysis. Notably, the study highlights the emerging role of LMR as a marker for monocyte-driven plaque instability, along with pathways mediated by chemokines such as CCL11 and TNFSF10, which reflect ongoing Th1-driven immune dysregulation in diabetic vascular disease. These underlying mechanisms appear to be fundamentally preserved but are exacerbated by epigenetic changes induced by hyperglycemia in T2D. Supplementary Information Below is the link to the electronic supplementary material. Supplementary Material 1 Supplementary Material 1