Cross-Sectional Association of Lipoprotein(a) Levels With Vascular Complications and Related Parameters in a Japanese Type 2 Diabetes Cohort Authors: Tsutomu Hirano, Mayumi Uetani, Kento Nakata, Natsuko Suzuki, Ema Aoki, Takeshi Hirashima, Yasuki Ito Published: 2026 ABSTRACT Aims: To investigate the association between Lipoprotein(a) (Lp(a)) levels and vascular complications, as well as related parameters, in a cohort of Japanese patients with type 2 diabetes. Methods: This cross-sectional analysis was conducted using data from a cohort of patients with 1465 type 2 diabetes. Lp(a) was measured by a turbidimetric immunoassay. Participants were categorized according to Lp(a) levels: GI (<30), GII (30–49), and GIII (≥ 50 mg/dL). Small dense low-density lipoprotein (LDL)-cholesterol (sdLDL-C) and LDL-triglycerides were measured using homogeneous assays. Results: The median [interquartile range] plasma Lp(a) concentration was 12.7 [6.1–24.6] mg/dL (21.8 [6–50.5] nmol/L). The prevalence rates for GI, GII, and GIII were 80, 12, and 8%, respectively, while low Lp(a) levels (<5 mg/dL) were present in 18%. These prevalence rates were comparable to those reported in the general Japanese population. Higher Lp(a) grades were associated with the prevalence of coronary artery disease (CAD), peripheral artery disease including borderline cases, severe diabetic retinopathy, and chronic kidney disease (CKD), but not with stroke. Lp(a) levels were higher in older individuals, female, kidney dysfunction, and statin users, whereas lower levels were observed in those with higher glycemia, current drinkers, and liver dysfunction. Patients with CAD or statin users had lower levels of LDL-C, sdLDL-C, LDL-triglycerides, and apolipoprotein B but higher Lp(a) concentrations compared to the control group. Conclusions: Lp(a) levels were associated with the prevalence of diabetic complications, particularly CKD and CAD. Lp(a) remains relevant associated factor with CAD in cross-sectional studies influenced by ongoing lipid-lowering therapies. Introduction Lipoprotein(a) (Lp(a)) is a lipoprotein particle structurally similar to low-density lipoprotein (LDL), consisting of an LDL particle in which apolipoprotein(a) is covalently bound to apolipoprotein (apo) B-100 1) . LDL-cholesterol (C) is a primary causal risk factor for atherosclerotic cardiovascular disease (ASCVD) 2) , and aggressive LDL-C lowering has become feasible with the widespread use of potent statins and proprotein convertase subtilisin/kexin type 9 (PCSK9) inhibitors 3) . Lp(a) is widely recognized as an atherogenic lipoprotein 4 - 6) ; however, because its plasma concentration is largely genetically determined and effective pharmacological interventions were previously unavailable, it had long been overlooked in routine clinical practice 4 - 6) . Interest in Lp(a) has increased substantially with the recent development of Lepodisiran, a small interfering RNA targeting hepatic Lp(a) synthesis 7) . Plasma Lp(a) concentrations are primarily regulated by genetic factors, particularly copy number variation in the LPA gene 8) , and show marked differences across ethnicities. Lp(a) levels in the Japanese population (median 12 mg/dL; interquartile range 5–26 mg/dL) are comparable to those in non-Hispanic White (12, 5–32 mg/dL) and Chinese populations (11, 4–22 mg/dL), but lower than those observed in Hispanic (19, 8–43 mg/dL) and African American populations (39, 19–69 mg/dL) 4) . Despite the relatively lower Lp(a) levels in Japanese individuals, Yoshida et al. 9) demonstrated in a systematic review of 26 studies that elevated Lp(a) is a significant risk factor for ASCVD in the Japanese population. Patients with diabetes have a markedly higher prevalence of ASCVD; however, whether Lp(a) levels are higher in diabetic than in non-diabetic populations remains controversial 10) . Notably, several studies have paradoxically demonstrated that low Lp(a) levels are associated with an increased risk of incident type 2 diabetes 11 - 13) . Nevertheless, data on the distribution of Lp(a) levels in large cohorts of Japanese patients with diabetes remain limited 9) , and a few studies have investigated the association between Lp(a) concentrations and diabetic complications 14 , 15) . The ViNA cohort was initiated in October 2019 and includes regular clinical assessments, systematic evaluation of diabetic complications, and long-term outcome analyses, with Lp(a) measurement prespecified at baseline 16) . This cross-sectional analysis aimed to clarify the clinical significance of Lp(a) measurement in Japanese patients with type 2 diabetes by examining the distribution of baseline Lp(a) levels and their association with diabetic complications. Methods Subjects Ebina General Hospital is conducting the “ViNA Cohort” study to investigate diabetic complications and prognosis in patients with diabetes 16) . Eligible participants were adults aged 30–89 years who had been receiving continuous treatment for diabetes for at least one year. Exclusion criteria included active malignancy, severe liver, endocrine, or respiratory disease, and ongoing hemodialysis. In the present study, patients with type 1 diabetes were excluded, and only those with type 2 diabetes were enrolled. Baseline blood and urine samples collected between October 2019 and June 2021 were analyzed. Measurement of Lp(a) was pre-specified, along with unusual lipid measurements, including small dense low-density lipoprotein-cholesterol (sdLDL-C), LDL-triglycerides (TG), and apoB, apoCIII, and apoE. The final study population comprised 1,465 patients with type 2 diabetes (943 male and 522 female). Measurements Blood samples were collected in the morning after an overnight fast. Plasma C-peptide levels were measured using an enzyme-linked immunosorbent assay (ELISA). High-sensitivity C-reactive protein (hs-CRP) and brain natriuretic peptide (BNP) were determined using commercially available assay kits. Urinary albumin excretion was assessed as the urinary albumin-to-creatinine ratio (UACR). Plasma Lp(a) concentrations were measured using a turbidimetric immunoassay with the N-assay TIA Lp(a) kit (Nittobo, Tokyo, Japan). The repeatability, analytical precision, and coefficients of variation of this assay have been previously reported by Miida et al. 17) , and its performance has been shown to be comparable to that of other commercially available Lp(a) assay kits used in Japan. Participants were categorized according to Lp(a) concentration into three groups: GI (<30 mg/dL), GII (30–49 mg/dL), and GIII (≥ 50 mg/dL), in accordance with the European Atherosclerosis Society (EAS) consensus statement on Lp(a), which defines <30 mg/dL (<75 nmol/L) as low risk and ≥ 50 mg/dL (≥ 125 nmol/L) as high risk 5) . Lp(a) concentrations expressed in nmol/L were converted from mg/dL using Miida’s equation specific to the Nittobo assay: Lp(a) (nmol/L) = 2.40 × [Lp(a) (mg/dL)] − 8.64 17) . Plasma apoB concentrations were measured by immunoturbidimetric assay and converted from mg/dL to nmol/L by multiplying by 20 6) . This conversion enables understanding of the contribution of Lp(a)-associated apo B to total apoB 6) . Diagnosis of Diabetic Complications Diabetic retinopathy was diagnosed by certified ophthalmologists. Severe diabetic retinopathy was defined as pre-proliferative or proliferative retinopathy. Chronic kidney disease (CKD) was defined as an estimated glomerular filtration rate (eGFR) of <60 mL/min/1.73 m². Coronary artery disease (CAD), including myocardial infarction, stable angina, and unstable angina, was diagnosed by cardiologists. All patients with CAD underwent coronary angiography followed by percutaneous coronary intervention. Peripheral artery disease (PAD), including suspected cases, was diagnosed based on an ankle–brachial index (ABI) of <1.0 18) . Stroke was diagnosed by specialists in neurology or neurosurgery and included lacunar infarction, atherosclerotic cerebral thrombosis, cardioembolic stroke, and intracerebral hemorrhage. Asymptomatic cerebral infarctions detected incidentally by computed tomography (CT) or magnetic resonance imaging (MRI) were excluded. Vascular and Fat Assessments Brachial–ankle pulse wave velocity (PWV) was measured using a volume plethysmographic device (PWV/ABI; OMRON Health Care Co., Ltd., Kyoto, Japan), which simultaneously recorded PWV and ABI. Carotid ultrasonography was performed by experienced sonographers using B-mode imaging (Fujifilm, Tokyo, Japan). Maximum intima–media thickness (max-IMT) and mean-IMT were assessed according to the method described by Katakami et al. 19) . Liver steatosis was diagnosed by radiologists based on comprehensive assessment of CT images, including reduced visualization of intrahepatic bile ducts and increased liver–spleen contrast. Visceral fat area (VFA) and subcutaneous fat area (SFA) were quantified using a CT-based fat scan program (Fujifilm, Tokyo, Japan). Among the 1,465 participants, PWV/ABI data were available for 1,317 subjects, IMT data for 1,315 subjects, and SFA/VFA/liver steatosis data for 1,278 subjects; thus, data were missing for 148, 150, and 187 subjects, respectively. The study complied with the principal of the Declaration of Helsinki. The study was detailed to all subjects who consented to participate, and a written informed consent form was obtained from all participants prior to the study. This study was approved by the Ethics Committee of Ebina General Hospital (no115,2019). Statistics Continuous variables are expressed as the mean±standard deviation (SD) or as the median with interquartile range (IQR), as appropriate, and categorical variables are presented as numbers and percentages. Differences between two groups were assessed using Student’s t-test or the Mann–Whitney U test. Categorical variables were compared using the chi-square test. Non-normally distributed variables were logarithmically (Log) transformed prior to group comparisons or regression analyses. Trends across Lp(a) grade groups were evaluated using the Cochran–Armitage trend test for categorical variables and the Jonckheere–Terpstra trend test for continuous variables. Correlations between continuous variables were assessed using Pearson’s correlation analysis. Multivariate linear regression analysis was used to examine independent associations between Lp(a) levels and variables that were significant in univariate analyses, adjusting for age, sex, and eGFR. Multivariate logistic regression analysis was performed to evaluate the independent associations between Lp(a)grades and diabetic complications after adjustment for age and sex. Results are presented as odds ratios (OR) with 95% confidence intervals (CIs). A p-value less than 0.05 was considered statistically significant. Analyses were performed using JMP software version 17 (SAS Institute, Cary, NC, USA). Results Fig.1 shows the distribution of plasma Lp(a) concentrations among the study participants. The median Lp(a) level [interquartile range] was 12.7 [6.1–24.6] mg/dL, corresponding to 21.8 [6.0–50.5] nmol/L, while the mean (SD) value was 20 (22) mg/dL. The prevalence of Lp(a) categories GI, GII, and GIII was 80%, 12%, and 8%, respectively. Very low Lp(a) levels (<5 mg/dL) were observed in 17.6% of participants (n = 258). Distribution of plasma levels of Lipoprotein(a) in 1465 patients with type 2 diabetes Table 1 , upper panel summarizes the clinical characteristics of participants stratified by Lp(a) grade (GI, GII, and GIII). With increasing Lp(a) grade, the proportion of female participants and mean age increased, whereas diabetes duration and body mass index (BMI) were comparable across groups. eGFR decreased significantly as Lp(a) grade increased. The proportion of current alcohol drinkers declined across higher Lp(a) grades. The prevalence of CAD was 9.8% in the GI group, 11.7% in the GII group, and 19.3% in the GIII group, indicating approximately a twofold higher prevalence in the GIII group compared with the GI group. A total of 117 cases of stroke were identified, comprising 57 cases of lacunar infarction, 32 cases of atherosclerotic cerebral thrombosis, 10 cases of cardioembolic stroke, and 18 cases of intracerebral hemorrhage. In contrast to CAD, the prevalence of stroke did not show a significant difference between Lp(a) grades. This finding remained consistent even when atherosclerotic cerebral thrombosis was separated from total stroke for analysis. The prevalence of PAD, including borderline cases defined by ABI <1.0, was 8.3% in the GI group, 8.3% in the GII group, and 17.8% in the GIII group, indicating approximately a twofold higher prevalence in the GIII group compared with the GI and GII groups. The prevalence of PAD, defined as an ABI <0.9 18) , was 3.8% in the GI group, 5.1% in the GII group, and 7.9% in the GIII group, with no significant difference observed between groups (p for trend = 0.188). The prevalence of CKD increased progressively with higher Lp(a) grades. While the overall prevalence of diabetic retinopathy, including simple retinopathy, was comparable across the groups, the prevalence of severe diabetic retinopathy was significantly higher in the GII and GIII groups than in the GI group. However, when adjusting for CKD in logistic regression analysis, the association between Lp(a) grade and severe diabetic retinopathy disappeared. In contrast to vascular complications, the prevalence of liver steatosis decreased progressively with increasing Lp(a) grades. More than half of the participants were receiving statin therapy, whereas only approximately 7% were treated with fibrates. The proportion of statin users increased across higher Lp(a) grades, whereas fibrate use declined. About 20% of the patients were using insulin, and among oral antidiabetic agents, dipeptidyl peptidase-4 inhibitors and metformin were the most frequently prescribed. The overall use of antidiabetic medications did not differ substantially among the Lp(a) grade groups. Table 1.Clinical characteristics in participants with type 2 diabetes stratified by Lipoprotein (a) gradeTotal subjectsLp(a) GI(<30mg/dL)Lp(a) GII(30-49mg/dL)Lp(a) GIII(>50mg/dL) trend-p Lp(a)mg/dL (median[IQR])12.7 [6.1-24.6]9.6 [5.1-16.5]37.2 [33.6-42.2]74.6 [56.2-93.4]<0.0001Lp(a) nmol/L (median[IQR])21.8 [6-50.5]14.4 [3.6-30.9]80.6 [72-92.6]170 [126-215]<0.0001Number (%)14651167 (79%)179 (12%)119 (8%)<0.0001Gender (%Female)36344052<0.0001Age (mean± SD)66.2±11.265±1167±1167±100.03Duration of diabetes14.1±8.614.1±0.2514.4±0.6413.8±0.790.83BMI (mean± SD)25.2±4.225.3±4.225.2±4.325.2±4.50.77% prevelenceCurrent Smoker2021.619.515.90.13Current Drinker4345.136.331.9<0.0001CAD10.89.811.719.3<0.0001Stroke (total)87.810.66.70.82Atherosclerotic cerebral thrombosis2.12.31.12.50.51 PAD (ABI<0.9, n = 1317) 4.33.85.137.930.18 PAD (ABI<1.0, n = 1317) 98.38.317.8<0.0001CKD (eGFR<60)27.425.234.637.8<0.0001Diabetic Retinopathy (total)26.525.428.135.30.23Severe Diabetic Retinopathy10.1914.314.60.02 Liver Steatosis (n = 1278) 36.738.73027.6<0.0001Statins54.751.561.475.6<0.0001Ezetimibe7.37.27.28.40.67Fibrates6.87.73.92.50.01Omega3-fatty acids43.94.44.20.79Insulin19.519.717.820.10.87Sulfonylurea27.127.724.525.20.36Metformin48.447.750.252.90.23SGLT2i25.825.820.633.60.36DPP-4i66.365.868.768.70.44GLP-1RA44.22.74.20.65Pioglitazone5.35.35.550.96Logistic analysisReference*Odds 95%CIp *Odds 95%CIp age and gender adjustedLp(a) GILp(a) GIILp(a) GIIICAD11.210.73-2.000.442.431.46-4.04<0.001 PAD (ABI<1.0, n = 1317) 10.920.49-1.700.792.11.19-3.700.01Severe Diabetic Retinopathy11.671.04-2.680.031.710.98-2.980.06CKD11.511.05-2.150.021.841.20-2.80<0.01age, gender, and CKD adjustedCAD11.150.69-1.910.572.271.35-3.8<0.01 PAD (ABI<1.0, n = 1317) 10.860.46-1.590.631.961.10-3.470.02Severe Diabetic Retinopathy11.580.98-2.540.061.550.88-2.730.12age, gender, CKD, and statin-use adjustedCAD10.980.58-1.660.951.741.02-2.950.04 PAD (ABI<1.0, n = 1317) 10.830.45-1.550.571.861.04-3.310.03CAD=coronary heart disease, PAD=peripheral artery disease, and CKD=chronic kidney disease*Odss Ratio: Lp(a) I as reference Clinical characteristics in participants with type 2 diabetes stratified by Lipoprotein (a) grade CAD=coronary heart disease, PAD=peripheral artery disease, and CKD=chronic kidney disease *Odss Ratio: Lp(a) I as reference Table 1 , lower panel shows the logistic regression analysis for Lp(a) grade and diabetic complications after adjusting for significant confounding factors. After adjustment for age and gender, individuals in the Lp(a) GIII group had significantly higher odds ratios for CAD (OR 2.4), PAD including borderline cases (OR 2.1), and CKD (OR 1.8) compared with those in the Lp(a) GI group. Compared with Lp(a) GI, the Lp(a) GII group showed a significantly increased odds ratio for severe diabetic retinopathy (OR 1.6). Although the Lp(a) GIII group also exhibited a higher odds ratio for severe diabetic retinopathy (OR 1.7; 95% CI 0.98–2.98), this association did not reach statistical significance. After adjusting for age, gender, and CKD, the Lp(a) GIII group still had significantly higher odds ratios for CAD (OR 2.27) and PAD including borderline cases (OR 1.96), while no significant difference was observed for severe retinopathy (95% CI 0.88–2.73). The significantly higher ORs for CAD and PAD persisted after adjusting for age, gender, CKD, and statin use. Table 2 presents clinical and biochemical parameters stratified by Lp(a) grade (GI, GII, and GIII). Glycemic indices, including fasting plasma glucose, HbA1c, and fasting C-peptide levels, were comparable across the Lp(a) groups. hs-CRP did not differ significantly among groups, whereas BNP levels increased progressively with higher Lp(a) grades. Plasma albumin levels and eGFR decreased significantly with increasing Lp(a) grades, while UACR showed no differences. Markers of liver dysfunction, including AST, ALT, and γ-GTP, declined with increasing Lp(a) grades. Plasma LDL-C, non–HDL-C, LDL-TG, and apo B levels increased with Lp(a) grades, while HDL-C, VLDL-C, sdLDL-C, fasting TG, apoCIII, and apoE levels were similar among the groups. With regard to body fat distribution, SFA and VFA did not differ significantly across Lp(a) grades. PWV, ABI, and mean-IMT were also comparable among groups; however, max-IMT increased significantly with higher Lp(a) grades. Table 2.Clinical and biochemical parameters stratified by Lipoprotein (a) gradeTotal subjectsLp(a) G1 (<30mg/dL)Lp(a) GII (30-49mg/dL)Lp(a) GIII (>50mg/dL) trend-p HbA1c %7.3±0.87.3±0.87.2±0.87.3±0.70.13Fasting Glucose mg/dL149±40150±37145±51148±510.30C-peptide ng/mL1.38 [0.91-2.07]1.36 [0.91-2]1.52 [0.92-2.3]1.4 [0.9-2.27]0.63hs-CRP mg/dL0.059 [0.027-0.136]0.058 [0.02-0.13]0.066 [0.031-0.135]0.057 [0.023-0.159]0.09BNP pg/mL11.9 [5.8-25.1]11.3 5.8-24.313.1 [5.8-31]14.4 [5.8-28]0.01Albumin g/dL4.35±0.344.37 0.334.26 0.3434.25 0.41<0.0001 eGFR mL/min/1.73m2 71±2072±2066±1965±21<0.0001UACR mg/g16.3 [6.5-58.3]15.9 [6.4-56.4]16.7 [5.7-64.2]19.4 [7.8-59.3]0.17AST U/L22 [18-27]22 [18-28]22 [18-26]21 [17-25]<0.001ALT U/L26±1821 [15-31]20 [14-26]19 [15-26]<0.001γGTP U/L25 [17-42]25 [18-43]24 [18-41]19 [15-34]<0.001plasma lipidsTC mg/dL180±30176±30183±33184±330.02LDL-C mg/dL101±24100±24105±25104±26<0.001HDL-C mg/dL54±1454±1454±1353±120.82nonHDL-C mg/dL126±28125±27129±29130±300.01VLDL-C mg/dL24±1024±1024±1026±120.31sdLDL-C mg/dL31±1431±1432±1330±130.47LDL-TG mg/dL15.3±5.715.1±516.3±9.116.6±5.7<0.001TG mg/dL107 [77-152]109 [76-154]104 [80-138]105 [77-154]0.90apoB mg/dL87±1885±1891±1892±19<0.0001apoCIII mg/dL10.4 [8.3-13]10.2 [8.3-12.9]10.6 [8.6-12.7]10.8 [8.9-13.7]0.07apoE mg/dL4±1.14±1.14.1±1.23.9±0.90.61Vascular and fat assessments SFA cm2 (n = 1278) 157±82156±81161±80165±890.20 VFA cm2 (n = 1278) 155±75156±75148±71151±760.23 ABI (n = 1317) 1.15±0.11.15±0.11.15±0.091.13±0.110.06 PWV (n = 1317) 1930±4741922±4701970±4641959±5190.40 max-IMT mm (n = 1315) 1.96±0.931.87±0.912.04±0.972.09±1.10.01 mean-IMT mm (n = 1315) 0.89±0.260.89±0.260.88±0.270.89±0.220.97Values are represented as mean±standard deviation or median [interquartile range].Visceral fat area (VFA); Subcutaneous fat area (SFA); Brachial–ankle pulse wave velocity (PWV);Ankle–brachial index (ABI); Maximum intima–media thickness (max-IMT) Clinical and biochemical parameters stratified by Lipoprotein (a) grade Values are represented as mean±standard deviation or median [interquartile range]. Visceral fat area (VFA); Subcutaneous fat area (SFA); Brachial–ankle pulse wave velocity (PWV);Ankle–brachial index (ABI); Maximum intima–media thickness (max-IMT) Table 3 summarizes the correlations between plasma Lp(a) levels and various clinical parameters. Lp(a) levels were significantly positively associated with age, but showed no association with BMI or duration of diabetes. Fasting plasma glucose and HbA1c were inversely correlated with Lp(a) levels. Lp(a) was not associated with fasting C-peptide or hs-CRP, but showed a positive correlation with BNP. Lp(a) levels were inversely correlated with eGFR, whereas no significant association was observed with UACR. In addition, Lp(a) was inversely associated with markers of liver dysfunctions. Regarding lipid parameters, Lp(a) was positively correlated with total-cholesterol, LDL-C, non–HDL-C, and apoB, but inversely correlated with TG. No significant associations were observed between Lp(a) and VLDL-C, sdLDL-C, LDL-TG, apoCIII, or apoE. In vascular and fat assessments, Lp(a) was positively associated with max-IMT and inversely associated with ABI, but showed no significant associations with SFA, VFA, PWV, or mean-IMT. Given the significant associations of Lp(a) with age, sex, and eGFR, multivariate regression analyses were performed with adjustment for these confounding factors. After adjustment, Lp(a) remained independently associated with glycemic parameters, markers of liver dysfunction, LDL-C, non–HDL-C, TG, apoB, and max-IMT. Table 3.Correlations between plasma Lipoprotein (a) levels and various clinical parametersVariablesSimple Regressions Log Lp(a)Multiple Regressions Age, gender, and eGFR adjustedMultiple Regressions Age, gender, eGFR, and statin-use adjustedrp β [95%CI]p β [95%CI]p Age0.116<0.0001BMI0.0190.463Duration of diabetes0.0170.492HbA1c-0.0740.004-0.081[-0.143--0.018]0.011-0.088[-0.150--0.025]0.005Fasting Glucose-0.059<0.001-0.021[-0.003--0.000]0.001-0.002[-0.003--0.000]0.001Log C-peptide0.0470.069Log hs-CRP0.0470.069Log BNP0.0770.003-0.011[-0.055-0.076]0.745Albumin-0.16<0.0001-0.396[-0.554--0.237]<0.0001-0.419[-0.577--0.260]<0.0001eGFR-0.137<0.0001Log UACR0.0330.308AST-0.17<0.0001-0.011[-0.015--0.008]<0.0001-0.011[-0.015--0.008]<0.0001ALT-0.173<0.0001-0.007[-0.010--0.005]<0.0001-0.008[-0.011--0.005]<0.0001Log[γGTP]-0.173<0.0001-0.211[-0.289--0.132]<0.0001-0.21[-0.289--0.132]<0.0001plasma lipidsTC0.0850.0010.003[0.000-0.004]0.0040.003[0.001-0.005]<0.0001LDL-C0.0950.0000.004[0.002-0.007]<0.00010.006[0.003-0.008]<0.0001HDL-C-0.0110.482nonHDL-C0.0870.0010.003[0.001-0.005]0.0010.004[0.002-0.006]<0.0001VLDL-C0.0090.727sdLDL-C-0.0590.069LDL-TG0.0340.201Log TG-0.0610.019-0.129[-0.238--0.018]0.022-0.141[-0.251--0.031]0.011apoB0.136<0.00010.008[0.005-0.01]<0.00010.009[0.006-0.012]<0.0001Log apoCIII-0.0050.837apoE0.0020.95Vascular and fat assessments SFA (n = 1278) 0.0090.75 VFA (n = 1278) 0.0340.227 ABI (n = 1317) -0.0680.013-0.309[-0.848--0.229]0.26 PWV (n = 1317) 0.0390.154 max-IMT (n = 1315) 0.1<0.0010.094[0.03-0.158]0.0040.089[0.025-0.153]0.004 mean-IMT (n = 1315) 0.010.719Correlations (r) betweenLog(Lp(a)) and variables are assessed using Pearson’s correlation analysis.Multivariate linear regression analysis is used to examine independent associations between Log(Lp(a)) and variables that are significant in univariate analyses, adjusting for age, gender, eGFR, and statin use Correlations between plasma Lipoprotein (a) levels and various clinical parameters Correlations (r) betweenLog(Lp(a)) and variables are assessed using Pearson’s correlation analysis. Multivariate linear regression analysis is used to examine independent associations between Log(Lp(a)) and variables that are significant in univariate analyses, adjusting for age, gender, eGFR, and statin use eGFR and statin use were strong confounding factors for Lp(a) levels, so these factors were incorporated into the multiple regression analysis along with age and sex. After adjusting for eGFR, BNP and ABI lost their significant association with Lp(a), but the other parameters maintained a significant association with Lp(a). The results were similar when further adjusting for statin use. Table 4 compares clinical and biochemical parameters between participants with and without CAD. Participants with CAD were older and more male than those without CAD. They exhibited higher HbA1c, BNP, and UACR, but lower eGFR, while BMI was comparable between the groups. As the majority of CAD patients were receiving statin therapy, often at higher doses, plasma levels of LDL-C, sdLDL-C, non-HDL-C, and apoB were significantly lower in the CAD group than in the non-CAD group. In contrast, plasma Lp(a) concentrations were approximately 1.2-fold higher, and the molar ratio of Lp(a) to apoB was 1.6-fold higher in participants with CAD compared with those without CAD. Consistent with the presence of advanced atherosclerosis, mean- and max-IMT were significantly greater, and ABI was significantly lower, in the CAD group. PWV, SFA, and VFA did not differ significantly between the groups. Table 4.Clinical and biochemical parameters between participants with and without CAD CAD (n = 159 ) Non-CAD (n = 1306) p Age y69±1065±11<0.0001Gender (%male)76%62%<0.01 BMI kg/m2 25±325±40.98LP(a) mg/dL16.1 [7.2-33.6]12.5 [6-23.8]0.02Lp(a) nmol/L30 [8.6-72]21 5.7-48.4]0.02HbA1c %7.51±0.917.37±0.850.02Fasting Glucose mg/dL152±41149±390.42BNP pg/mL25.4 [11.7-56.8]10.8 [5.8-22.5]<0.0001 eGFR mL/min/1.73m2 62±2172±19<0.0001UACR mg/g26.9 [7.9-143.6]15.2 [6.3-53.1]<0.01plasma lipids (mg/dL)TC159±27182±30<0.0001LDL-C85±21103±24<0.0001HDL-C48±1254±14<0.0001nonHDL-C111±25128±27<0.0001VLDL-C25±1224±100.26sdLDL-C27±1132±14<0.0001LDL-TG14.5±3.915.4±5.90.06TG110 [81-156]107 [77-151]0.23apoB (mg/dL)78±1688±18<0.0001apoB (nmol/L)1565±3211763±372<0.0001Lp(a)/apoB (mol/mol)0.019 [0.006-0.0448]0.012 [0.0034-0.0288]<0.0001apoCIII10.4 [8.4-13.2]10.4 [8.3-12.9]0.94apoE3.7±1.34±1.1<0.01Other variablesABI1.12±0.131.15±0.1<0.0001PWV1952±5991928±4570.59mean-IMT mm1±0.330.88±0.25<0.0001max-IMT mm2.41±0.961.85±0.91<0.0001 SFA cm2 150±71158±830.28 VFA cm2 159±68154±750.49Stain use84%51%<0.0001Fibrate use3%7%0.11Values are represented as mean±standard deviation or median [interquartile range]. Differences between two groups were assessed using Student’s t-test or the Mann–Whitney U test. Categorical variables were compared using the chi-square test.Non-normally distributed variables were logarithmically transformed prior to group comparisons. Clinical and biochemical parameters between participants with and without CAD Values are represented as mean±standard deviation or median [interquartile range]. Differences between two groups were assessed using Student’s t-test or the Mann–Whitney U test. Categorical variables were compared using the chi-square test. Non-normally distributed variables were logarithmically transformed prior to group comparisons. Table 5 compares clinical and biochemical parameters between participants with and without statin therapy. Statin users exhibited approximately 1.3-fold higher plasma Lp(a) levels than non-statin users. Compared with non-statin users, statin users had a higher HbA1c, and UACR, but a lower proportion of males and a lower eGFR. As expected, plasma levels of total-cholesterol, LDL-C, sdLDL-C, LDL-TG, non–HDL-C, and apoB were significantly lower in statin users than in non-statin users. In contrast, the molar ratio of Lp(a) to apoB was approximately 1.5-fold higher in statin users. Regarding vascular and fat composition parameters, ABI was lower and SFA was greater in statin users, whereas IMT, PWV, and VFA were comparable between statin and non-statin users. Among patients receiving statin therapy, those using fibrates were rare. Table 5.Clinical and biochemical parameters between participants with and without statin use Statin use (n = 802 ) Non-statin use (n = 663) p Age y66±1066±110.9Gender (%male)57%72%<0.0001 BMI kg/m2 25.6±4.124,8±4.3<0.0001LP(a) mg/dL14.6 [6.5-29.9]11.3 [5.6-20.8]<0.0001Lp(a) nmol/L26.4 [6.9-63.2]18.4 [4.8-41.2]<0.0001HbA1c7.42±0.837.31±0.880.02Fasting Glucose mg/dL149±39150±410.73BNP pg/mL11.9 [5.8-24.4]12 [5.8-25.9]0.67 eGFR mL/min/1.73m2 70±2072±200.04UACR mg/g18.7 [7-67.6]14.6 [5.7-51.6]0.02plasma lipids (mg/dL)TC172±29189±29<0.0001LDL-C95±22109±24<0.0001HDL-C53±1355±15<0.01nonHDL-C119±26134±27<0.0001VLDL-C24.5±10.924.5±10.70.98sdLDL-C30±1333±15<0.0001LDL-TG15±515.7±6.40.01TG108 [83-153]104 [71-150]0.01apoB (mg/dL)84±1890±18<0.0001apoB (nmol/L)1682±3611812±375<0.0001Lp(a)/apoB (mol/mol)0.016 [0.004-0.0370.010 [0.002-0.024<0.0001apoCIII10.5 [8.6-13.2]10.2 [8.1-12.7]0.01apoE3.9±1.14.2±1.1<0.0001Other variablesABI1.14±0.1071.16±0.103<0.0001PWV1911±47551954±4710.1mean-IMT mm0.906±0.2760.879±0.2460.05max-IMT mm1.931±0.9291.898±0.9470.51 SFA cm2 164±81149±81<0.01 VFA cm2 158±74151±750.13Stain use100%0%Fibrate use2%12%<0.0001Values are represented as mean±standard deviation or median [interquartile range]. Differences between two groups were assessed using Student’s t-test or the Mann–Whitney U test. Clinical and biochemical parameters between participants with and without statin use Values are represented as mean±standard deviation or median [interquartile range]. Differences between two groups were assessed using Student’s t-test or the Mann–Whitney U test. Discussion In individuals with type 2 diabetes, a condition that frequently coexists with ASCVD, previous studies have paradoxically reported lower rather than higher Lp(a) concentrations 4 , 10 - 13) . Our results demonstrated that the median plasma Lp(a) level in this diabetes cohort was 12.7 [6.1–24.6] mg/dL, with a mean value of 20±22 mg/dL. These values are nearly identical to those reported in the general Japanese population, including the Jichi Medical School Cohort (median 12 [5–26] mg/dL) 20) , Abe and Noma‘s report 21) (median 11mg/dL), the Suita Study (mean 20±16 mg/dL) 22) , and a large hospital-based insurance claims database in Japan comprising individuals at high cardiovascular risk (median 14 [8–25] mg/dL) 23) . Furthermore, the prevalence of elevated Lp(a) levels (≥ 30 mg/dL) in our cohort (20%) was comparable to that reported in the Suita Study and the hospital-based database (approximately 17%) 22 , 23) . Individuals with very low Lp(a) concentrations (below 3–5 mg/dL) have been reported to have a higher risk of diabetes 4 , 5) . In the present study, 18% of participants had very low Lp(a) levels (<5 mg/dL), a proportion comparable to that reported in general population studies, where the prevalence of Lp(a) <5 mg/dL is approximately 15–20% 22 , 23) . Although a weak inverse correlation was observed between plasma Lp(a) levels and blood glucose or HbA1c, overall Lp(a) concentrations in our diabetic cohort were not lower than those reported in the general Japanese population. However, since vascular complications and statin use were associated with elevated Lp(a) levels, diabetic patients without vascular complications or statin use may have lower Lp(a) levels than the general population. Supplemental Fig.1 shows the distribution of Lp(a) levels according to the presence or absence of diabetes complications (CKD, PAD, stroke, diabetic retinopathy, and CKD) and statin use. The median Lp(a) level in the group without diabetic complications and not using statins decreased to 9.9 [4.9–17.5] mg/dL, which appears somewhat lower than values in the general Japanese population. Therefore, it cannot be ruled out that lipoprotein(a) levels may be low in diabetes itself. Distribution of plasma levels of Lipoprotein(a) in type 2 diabetes patients without /with diabetes complications or statin use Lp(a) was inversely correlated with eGFR, and accordingly, Lp(a) levels were higher in patients with CKD than in those without CKD, as has been well documented previously 24) . It is also well established that patients with nephrotic syndrome exhibit markedly elevated Lp(a) levels 25) . In the present study, however, no significant association was observed between urinary albumin–creatinine ratio (UACR) and Lp(a) levels. In contrast, plasma albumin concentrations showed a significant inverse correlation with Lp(a). These findings suggest that hypoalbuminemia itself may be a key determinant of elevated Lp(a) levels, whereas albuminuria per se may not influence Lp(a) unless it leads to a reduction in circulating albumin. The kidney plays an important role in Lp(a) clearance, likely through glomerular filtration followed by uptake and degradation in the proximal tubules 24) . Thus, reduced renal catabolism of Lp(a), rather than increased hepatic production, may be a major mechanism underlying elevated plasma Lp(a) concentrations in patients with CKD who do not have massive albuminuria. Recently, Hiraishi et al. 26) reported that an inverse association between Lp(a) and eGFR was observed in non–statin users but not in statin users. In line with this report, our analysis showed that in the statin-treated group, the correlation between Lp(a) and eGFR was weaker compared to the untreated group (r = 0.111 vs. 0.158), but the significant negative correlation between Lp(a) and eGFR persisted (p = 0.0015). There have been conflicting reports regarding whether Lp(a) is a risk factor for diabetic retinopathy 27) In this study, the prevalence of diabetic retinopathy, including simple retinopathy, did not show significant differences across Lp(a) grades. However, the prevalence of severe diabetic retinopathy was significantly higher in participants with higher Lp(a) grades. Our findings may be supported by a recent meta-analysis showing that Lp(a) concentrations are higher in patients with diabetic retinopathy and even higher in those with proliferative diabetic retinopathy 28) . However, in the CKD-adjusted logistic analysis, the relationship between Lp(a) grade and severe diabetic retinopathy was lost. Given that CKD and retinopathy both arise from common metabolic abnormalities in diabetes, and that CKD itself is a factor worsening retinopathy, the effect of Lp(a) on retinopathy is interpreted as being strongly influenced by CKD. Previous studies have reported that individuals with habitual alcohol consumption have lower Lp(a) levels 20 , 29) . In line with these observations, the proportion of current drinkers decreased progressively with increasing Lp(a) grades in the present study. Correspondingly, markers of liver dysfunction as well as the prevalence of liver steatosis, were more common in participants with lower Lp(a) grades. Sporadic reports have also described reduced Lp(a) concentrations in individuals with non-alcoholic fatty liver disease 30 , 31) . Lp(a) might be involved in hepatic lipid transport and clearance, potentially contributing to reduced lipid accumulation in the liver 32) . In the present study, higher Lp(a) grades were significantly associated with the prevalence of CAD but not with stroke, although several studies revealed significant association between Lp(a) and stroke 4 , 33) . According to the recent systematic review by Yoshida for Japanese populations 9) , most studies investigate its association with CAD and no paper clearly demonstrate a link with stroke. Accumulating evidence suggests that the contribution of Lp(a) differs by vascular bed, exerting a stronger effect on CAD than on cerebrovascular disease 4 , 5) . Following CAD, PAD has been reported to be associated with Lp(a) 4 , 5 , 9) . As a representative example, the European Prospective Investigation into Cancer and Nutrition (EPIC)-Norfolk cohort study which included a large population of apparently healthy individuals, demonstrated that Lp(a) was significantly associated with PAD and CAD outcomes, but not with stroke 34) . In our study, PAD was diagnosed using ABI, and the definition included borderline cases (ABI <1.0). Defining PAD as ABI <0.9 yielded a low prevalence rate among the subjects (4.3%), and many subjects had ABI values between 0.9 and 1.0. Defining PAD as ABI <1.0 yielded a prevalence of 9%, double that of ABI <0.9. Therefore, the borderline PAD cases were included. Although this approach limits the ability to draw definitive conclusions, the association between Lp(a) and PAD is considered stronger than its association with stroke. A number of large prospective cohort studies and Mendelian randomization analyses have established Lp(a) as an independent risk factor for CAD, irrespective of LDL-C and apoB levels 5 , 6 , 11) . In the present study, participants with Lp(a) concentrations ≥ 50 mg/dL had a significantly higher prevalence of CAD, with an odds ratio of 2.43 (95% CI: 1.46–4.04), compared with those with Lp(a) ≤ 30 mg/dL. Current clinical guidelines recommend intensive LDL-C lowering to ≤ 70 mg/dL for secondary prevention in patients with CAD, with high-intensity statin therapy as the first-line treatment 3) . Consequently, LDL-C levels are generally lower in CAD patients than in non-CAD controls. Notably, Lp(a) concentrations are not reduced by statin therapy; rather, statins have been reported to increase Lp(a) levels 35) . Supporting this observation, incubation of cultured hepatocytes with atorvastatin has been shown to increase LPA mRNA expression and apo (a) protein production 35) The present results demonstrated that CAD patients had significantly lower levels of LDL-C, sdLDL-C, and apoB, but higher levels of Lp(a). Similar patterns were observed when comparing statin users with non-statin users. Each Lp(a) particle contains one molecule of apo B, similar to LDL. Therefore, the relative contribution of Lp(a)-derived apo B to total apo B is particularly useful, especially when LDL-C and apo B levels are adequately controlled 6) . This study suggests that Lp(a) may contribute as an atherogenic burden under conditions of potent LDL reduction, as evidenced by approximately 1.5-fold higher Lp(a)/apoB molar ratios in CAD patients and statin users compared to the control group. sdLDL-C and LDL-TG are recognized as residual risk factors for CAD beyond LDL-C 36) . However, the relationship between these LDL subspecies and Lp(a) has not been fully elucidated. In the present study, no significant correlations were observed between Lp(a) and sdLDL-C or LDL-TG, supporting the concept that Lp(a) represents a distinct lipoprotein entity rather than an LDL subspecies. While sdLDL-C is an established therapeutic target for secondary prevention in CAD patients with adequately controlled LDL-C levels 37) , it remains to be determined whether Lp(a) will assume a similarly important role as a modifiable residual risk factor. The results of ongoing and future intervention trials targeting Lp(a) are therefore eagerly anticipated. This study has several limitations. First, it was a cross-sectional analysis based on baseline data from a cohort study; therefore, causal relationships between Lp(a) levels and diabetic complications cannot be established. Second, PAD was assessed using the ABI and included borderline cases, which may not fully represent clinically confirmed PAD events. Third, because Lp(a) levels are influenced by renal function, caution is required when interpreting the causal relationship between elevated Lp(a) levels and increased CKD prevalence. Fourth, a substantial proportion of patients with CAD were receiving intensive statin therapy, which may be attributed to higher Lp(a) concentrations. Analysis of statin types and dosages will be necessary going forward. Conclusion In this cross-sectional analysis of 1,465 individuals from a type 2 diabetes cohort, the distribution of plasma Lp(a) concentrations was largely comparable to that reported in other Japanese populations. Higher Lp(a) grades were significantly associated with the prevalence of CAD, CKD, PAD including borderline cases, and severe diabetic retinopathy. In particular, Lp(a) is a significant associated factor for CAD in patients with type 2 diabetes, especially under statin therapy. Disclosure Tsutomu Hirano receives advisor fee from Denka Co, and lecture fee from Kowa Co. Yasuki Ito is an employ of Denka Co. Ethical Consideration This study was approved by the Ethics Committee of Ebina General Hospital (no 115, 2019) Approval date; October 1, 2019.Informed Consent: The study was detailed to all subjects who consented to participate, and a written informed consent form was obtained from all participants prior to the study.