Association of circulating oxidized LDL-C/LDL-C ratio with severity of coronary atherosclerosis and other emerging biomarkers of cardiovascular disease in patients with type 2 diabetes


 Background

Atherosclerosis cardiovascular diseases (ASCVD) are the leading cause of mortality and morbidity in patients with diabetes. But there are still no simpler and more effective biomarkers which can predict the severity of coronary atherosclerosis in diabetic patients. Oxidized low-density lipoprotein (oxLDL) is a novel CVD biomarker that also increase in diabetic patients. The aim of present study was to evaluate the associations of circulating oxLDL/LDL-C ratio with severity of coronary atherosclerosis and other emerging biomarkers of cardiovascular disease such as fibroblast growth factor 21 (FGF21), interleukin 33 (IL33) and vascular cell adhesion molecule-1 (VCAM-1) in patients with type 2 diabetes.
Methods

152 patients with type 2 diabetes were recruited in our study. Plasma levels in oxLDL, FGF21, IL33 and VCAM-1 were measured by ELISA. oxLDL/LDL-C ratio was calculated. Coronary computed tomographic angiography (CCTA) was performed in all patients. Patients were allocated into four groups according to CCTA findings and Gensini score: normal, mild, moderate and severe coronary atherosclerosis group. Association between the oxLDL/LDL-C ratio and the severity of coronary atherosclerosis were evaluated using logistic regression models and receiver operator characteristic (ROC) curve analyses.
Results

Correlation analysis showed that oxLDL/LDL-C ratio was positively associated with severity of coronary atherosclerosis (p < 0.05). After adjusted for age, duration of diabetes, the positive correlation between severity of coronary atherosclerosis and oxLDL/LDL-C ratio still existed (OR 2.03, 95% CI 1.31- 3.14, p < 0.01). OxLDL/LDL-C ratio in severe group was significantly higher than which in the other three groups respectively (p< 0.001). The cut-off value of oxLDL/LDL-C ratio in predicting severe coronary atherosclerosis was more than 0.091 with specificity of 66.5% and sensitivity of 75%. Only IL33 correlated positively with oxLDL/LDL-C ratio (r = 0.274, p<0.01). However, VCAM-1 had a similar trend with oxLDL/LDL-C ratio in these four groups.
Conclusions

OxLDL/LDL-C ratio is considered as a potential biomarker in diabetic patients for early recognition and intervention of severe coronary atherosclerosis, and will be more effective if tested IL33 and VCAM-1 at the same time.

general cardiovascular risk, with a reduction in serum levels used clinically as a treatment target.
However, there is no su cient clinical evidence that LDL-C level can predict the severity of coronary artery disease to date. Atherosclerosis is a complex process, and there is still scope for more speci c biomarkers with pathological relevance to improve the clinical risk prediction of cardiovascular events especially in patients with diabetes [4] . Native LDL-C, however, cannot induce cholesterol accumulation in monocytes/macrophages which increase, rather, is due to the uptake of one or more modi ed forms of LDL-C [5] . The major modi cation that can induce cholesterol accumulation is oxidized LDL-C [6] . It is hypothesized that under conditions of oxidative stress, oxLDL formation predominantly occurs within the extracellular space of the vascular wall. Patients with diabetes exhibit a decreased plasma antioxidant activity as well as increased levels of lipid hydroperoxides and F2-isoprotanes. The latter two are recognized in vivo markers of oxidative stress [7] . Hyperglycemia-induced oxidative stress modi es LDL-C into an oxLDL form [8] . Moreover, elevated oxLDL titres may prove to be more causally associated with cardiovascular events than LDL-C, due to its central role in atherosclerotic plaque biology, and thus may have potential for improving risk prediction beyond LDL-C in patients with diabetes. The present observations suggest that oxLDL and ox-LDL/LDL-C are better biomarkers than popular lipid pro le [total cholesterol (TC), LDL-C, high density lipoprotein-cholesterol (HDL-C)] for discriminating between patients with CAD and healthy subjects [9] .
There has been evidences that oxLDL is increased in subclinical atherosclerosis [10] and is often a stronger predictor of acute coronary artery disease (CAD) than standard lipid measures or other conventional risk factors [11] . OxLDL levels are reportedly able to distinguish patients with CAD from healthy cohorts [12] , and serve as a predictor of future myocardial infarction in patients with unstable CAD [13] . OxLDL level is also associated with type 2 diabetes mellitus (T2DM) [14] . Previous studies have examined the ratios of oxLDL/LDL-C and demonstrated that this metric is more informative than oxLDL alone. Data from these study revealed that the correlationship between CAD and relative degree of LDL-C oxidation is stronger than with the level of oxidized LDL-C in vivo [15][16][17] .Most of evidences suggeste that oxLDL or oxLDL/LDL-C ratio can predict future cardiovascular events. However it remains unclear whether it can predict the severity of coronary atherosclerosis. The association between circulating oxLDL and the severity of coronary atherosclerosis yielded con icting results. In several studies, oxLDL levels were not signi cantly related to the extent of coronary heart disease (CHD) [18][19] , but in other studies the association was not found [20] . Especially there was no report about the relationship in patients with diabetes. Since cardiovascular events are irreversible, it will be more meaningful to recognize severe coronary atherosclerosis earlier through some effective biomarkers.
Since studies have demonstrated that the severity of CAD assessed by Gensini score is associated with increased cardiovascular events [21] . The present study aimed to assess the relationship between oxLDL/LDL-C ratio with severity of coronary atherosclerosis through Gensini scores calculated on the basis of coronary computed tomographic angiography (CCTA) in patients with diabetes. Recent advances in CT technology improve signi cantly the diagnostic accuracy of CCTA, which leads to substantial increases in its non-invasive use for patients with suspected CAD [22] . It was reported that the presence and severity of CAD visualized by CCTA predict death or MI across different large ethnicities [23] .
Current evidences show that the calculation of SYNTAX score derived from CCTA is accurate with respect to the one derived from ICA assessment [24] . Many studies calculated Gensini score derived from CCTA and proved the accuracy and its impacts [25][26][27] .
Moreover in present study we detected some emerging biomarkers such as FGF21 and IL33, which were reported to be related with diabetes and CAD in some contradictory results, and conventional biomarker such as VCAM-1. FGF21 is a multifunctional protein with major secretion and expression in the adipose tissues and liver with control over energy, glucose, and lipid metabolism [28] . There are higher levels in FGF21 in patients with AS, according to current preclinical and clinical reports [29][30] . IL-33 exerts its cellular functions by binding a receptor complex composed of suppression of tumorigenity (ST2L) and IL-1R accessory protein [31] . According to current knowledge, IL-33 seems to be released during necrotic cell death, which is thought to be associated with tissue damage. For these properties, IL-33 was proposed to act as an "alarmin" [32] . VCAM-1 is the transmembrane glycoproteins that are responsible for transmigrating leukocytes into the vascular intima. Elevated expression and activity of VCAM-1 is indicative of in ammation, endothelial dysfunction, and atherosclerosis [33] . The present study tried to investigate the relationship of these biomarkers with oxLDL/LDL-C ratio and the severity of diabetic coronary atherosclerosis.

Study participants
A total of 152 patients with type 2 diabetes at the Shanghai East Hospital, Tongji university, were included in our study. Patients were diagnosed according to the 1999 World Health Organization criteria. Patients with type 1 diabetes mellitus or speci c types of diabetes mellitus, acute complications of diabetes, liver or renal dysfunction, serious cardiac arrhythmias or acute cardiac insu ciency, diabetic foot ulcers and history of cerebral infarction, tumor and psychosis were excluded. Written informed consents were obtained from all participants. The study was approved by the Human Research and Ethics Committee of Shanghai East Hospital and adhered to the tenets of the Declaration of Helsinki. The study was registered in the Chinese clinical trial registry (ChiCTR-2100047148).
Data collection and physical assessment Information on sex, age, weight, height, duration of diabetes, and history of smoking, coronary artery disease, hypertension and cerebral apoplexy were obtained using a questionnaire. The body mass index (BMI) was calculated as the weight (kg) divided by the square of height (m). Blood pressure was measured using a sphygmomanometer during a physical examination.

Biochemical assessment
Venous blood samples were collected from all patients after overnight fasting. The levels of TC, LDL-C, HDL-C, triglycerides (TG), fasting plasma glucose (FPG), glycosylated hemoglobin (HbA1c) and serum creatinine were measured by suitable laboratory techniques including HPLC for HbA1c. Serum samples stored at -80℃were used for the determination of oxLDL levels using ELISA assays kit supplied by Mercodia (Uppsala, Sweden). Serum FGF21, IL33 and VCAM-1 were determined by ELISA kits supplied by Abcam (Cambridge, MA, USA).

Coronary computed tomographic angiography and Gensini score calculation
A 64-multi-detector computed tomography scanner (Aquilion ONE TSX-301C, TOSHIBA was used. Imaging range was as follows: 1 inch above the left main ostium down to 1 inch below the bottom of the heart. Cardiac data were reconstructed from 5 to 95% of R-R interval and with 10% intervals. Coronary vessels were analyzed. Volume renderings and curved multiplanar reformations were reported. Imaging of coronary arteries was established with the use of 30-40 contiguous 3 mm slices (mid-diastole) with ECG triggering and breath holding for 15 s.
In this study, severity of coronary atherosclerosis was assessed by Gensini score, which is based on the percentage of luminal narrowing. The Gensini score was computed by assessing the severity score of each coronary stenosis according to the degree of luminal narrowing and its geographic importance. Reduction in the lumen diameter and appearance of concentric lesion or eccentric plaques were evaluated (reduction of 25%, 50%, 75%, 90%, 99%, and complete occlusion were given Gensini scores of 1, 2, 4, 8, 16, and 32, respectively). Each vascular segment was assigned a multiplier in accordance with the functional signi cance of the myocardial area supplied by that segment: the left main coronary artery ×5; the proximal segment of left anterior descending artery (LAD) ×2.5; the proximal segment circum ex artery ×2.5; the mid segment of the LAD ×1.5; the right coronary artery, the distal segment of LAD, the posterolateral artery, and the obtuse marginal artery ×1; and others ×0.5 [34] . All patients were allocated into four groups according to CCTA ndings and Gensini score: normal group (score=0), mild coronary atherosclerosis group (0<scores≤3), moderate coronary atherosclerosis group (3.01≤scores≤32.67) and severe coronary atherosclerosis group (32.68≤scores≤180).

Statistical methods
For statistical analysis, SPSS 21.0 was used. Data were expressed as mean ± standard deviation (SD) for continuous variables and percentages (%) for categorical variables. Student's t-test was used to compare continuous variables. Normality of continuous variables was determined and con rmed using Kolmogorov-Smirnov test. Differences among the groups were analysed using analysis of variance for measurement data, as well as the chi squared test for categorical values. Pearson correlation coe cient and Spearman's correlation coe cient were used to express the correlations between oxLDL/LDL-C ratio and different variables. Multiple or binary logistic regression analysis was performed to evaluate the odds ratio (OR) and associated factors. We calculated the odds ratio (OR, 95% CI) in two logistic regression models: a non adjusted model and a multivariable model adjusted for all variables, with p < 0.05 in the non adjusted model. A ROC curve was employed to nd the cut off of oxLDL/LDL-C ratio to indicate the occurrence of severe coronary atherosclerosis. All p values were two tailed, and p < 0.05 was considered to be statistically.

Results
General data of the patients The clinical data of the four groups were collected and compared. There was gradual increase in age form normal group to severe group with statistical difference between severe group and the other three groups. Duration of diabetes has the same trend as age with statistical difference between normal group and severe group. BMI in mild group was the highest among these four groups, and has statistical difference compared with moderate group. There were statistical differences in HbA1c levels between groups. The lowest LDL-C level was unexpectedly appeared in severe group with statistical difference compared with the other three groups. It is noted that the rate of statin-use in baseline was also the highest in severe group which may be the reason for the lowest LDL-C level. (Table 1) Comparison of oxLDL/LDL-C ratio and other interested biomarkers levels in different coronary atherosclerosis groups OxLDL/LDL-C ratio in severe group was signi cantly higher than which in the other three groups respectively (0.111±0.052 vs. 0.083±0.040, 0.086±0.036, 0.085±0.048 respectively, p< 0.05) ( Figure  1).When oxLDL/LDL-C ratio was strati ed into two groups according to the median (≤0.0819 were assigned to the low-ratio group; >0.0819 were assigned to the high-ratio group), the Gensini scores were signi cantly higher in high-ratio group than in low-ratio group (25.34±5.72 vs 12.43±2.21, P=0.038) (Figure 1). The rate of patients with severe coronary atherosclerosis in high-ratio group was signi cantly higher than that in low-ratio group, P=0.039) (Figure 2). We also found that the patients in severe group had higher VCAM-1level than those in the other three  (Figure 3). There were no differences in the level of IL33, FGF21 among these four groups.

Correlation of variables with severity of coronary atherosclerosis in diabetic patients
Spearman's correlation analysis showed age, and oxLDL/LDL-C ratio were positively correlated with severity of coronary atherosclerosis(both p < 0.05).
Moreover, taking low oxLDL/LDL-C ratio group as the referent, we evaluated the correlation between oxLDL/LDL-C ratio and severity of coronary atherosclerosis in two statistical models by ordinal logistic regression ( Table 2). In non-adjusted model, a signi cant positive correlation was observed between severity of coronary atherosclerosis and high oxLDL/LDL-C ratio (OR 1.83, 95% CI 1.21-2.77, p<0.01).

Correlation between oxLDL/LDL-C ratio and other variable
No correlation was found between oxLDL/LDL-C ratio and age, duration of diabetes, use of statin, HbA1c, BMI, LDL-C, FGF21, and VCAM-1 levels. Only IL33 showed positive correlation with oxLDL/LDL-C ratio tested by Spearman analysis (r = 0.339, p<0.001). When all patients were divided in two groups: low oxLDL/LDL-C ratio group and high oxLDL/LDL-C ratio group, IL33 levels were signi cantly higher in the high oxLDL/LDL-C ratio group compared with the low group (5.34±0.25 vs. 5.19±0.22, p=0.001).
We divided the patients into two groups based on whether they had used statins. The results indicated that level of serum LDL-C and oxLDL were both decreased signi cantly in statin-user compared with nonstatin-user ( Comparison of oxLDL/LDL-C ratio and other biomarkers between diabetic patients with and without signi cant coronary lesions Lesions were considered signi cant if the coronary stenosis was ≥50% diameter and insigni cant if the stenosis was <50%. There were signi cant differences in oxLDL/LDL-C ratio, IL33 and VCAM-1 levels between diabetic patients with and without signi cant coronary lesions (p < 0.05) ( Table 3).

Discussion
Although the invasive conventional coronary angiography (ICA) is the golden standard to evaluate the severity of coronary lesions, its use in the clinical arena of T2DM patients is limited. This is an invasive technique that can rarely be used to evaluate the minor or moderate coronary lesions with no obvious clinical symptoms [35] . In order to avoid the selection bias toward subjects with severe coronary lesions, we used CCTA instead of ICA in our study on the basis of the evidences for accuracy of CCTA. In this study, wefound that oxLDL/LDL-C ratio observably associated with severity of diabetic coronary lesions and signi cantly increased in patients with severe coronary lesions. This suggested that oxLDL/LDL-C ratio may be a potential biomarker for screening severe coronary atherosclerosis in patients with T2DM.
OxLDL plays an important role in atherogenesis by promoting an in ammatory environment and lipid deposition in the arterial wall [36] . Therefore, elevated levels of circulating oxLDL or oxLDL/LDL-C ratio have been demonstrated in large prospective cohorts, powered for clinical endpoints, to confer an adverse CVD prognosis. The Bruneck Study showed a signi cant association between oxidation-speci c epitopes biomarkers including oxidized phospholipids of ApoB-100 and the development of adverse CVD events over 15-year follow up [37] . A recent meta-analysis supported these ndings, reporting that elevated serum oxLDL is associated with an increased risk of CVD events [38] . In a previous prospective nested casecontrol study of initially healthy men, plasma oxLDL concentrations were signi cantly higher in men who subsequently experienced a coronary event than in the matched controls [11] .Moreover, oxLDL is signi cantly associated with diabetes and its complications. There are signi cant higher levels of oxLDL in patients with diabetes compared with non-diabetics [39] . A signi cant association of plasma oxLDL/LDL-C with CHD was found in males with diabetes in a 10-year prospective study [40] . Most of studies were focus on the association of oxLDL and/or oxLDL/LDL-C with the cardiovascular events. None of studies demonstrated the correlation between oxLDL/LDL-C ratio and the severity of coronary atherosclerosis in patients with diabetes. Faviou E et al reported that serum oxLDL level was associated with coronary artery disease as well as with the severity of the clinical presentation. However, this study didn't estimate the coronary artery lesions by ICA or CCTA [41] . To this extent, our study for the rst time demonstrated that oxLDL/LDL-C ratio could re ect the severity of coronary atherosclerosis evaluated by CCTA in patients with diabetes. This biomarker may provide a tool to identify the severe coronary artery lesions earlier in order to prevent cardiovascular events.
As for the in uencing factors of severity of coronary lesions, the results of our study showed a positive correlation between severity of coronary lesions and age, as well as no correlation with duration of diabetes, HbA1c and LDL-C levels. When divided all patients into four groups according to the Gensini scores, the severe group had the longest duration of diabetes and lowest LDL-C levels which may be caused by the highest rate of statin-use. Our study indicated that statins-use decreased both LDL-C and oxLDL levels, however, oxLDL/LDL-C ratio was not affected.
More importantly, this study demonstrated the predicting and screening value of oxLDL/LDL-C ratio for severe coronary atherosclerosis in patients with type 2 diabetes. The present study showed a positive correlation between severity of coronary lesions and oxLDL/LDL-C ratio even adjusted for age, duration and other associated variables. The risk for developing more severe coronary atherosclerosis in high oxLDL/LDL-C ratio group was nearly two-times higher than low oxLDL/LDL-C ratio group. After dividing all patients with T2DM into four groups from normal to severe according to Gensini scores, we found that there was the highest oxLDL/LDL-C ratio in severe group which had a signi cant difference compared with other three groups. Gensini scores presenting the severity of coronary lesions were in the high oxLDL/LDL-C ratio group than which in the low group. The cut-off for oxLDL/LDL-C ratio to discriminate severe coronary atherosclerosis from normal to moderate lesions was 0.091 for oxLDL/LDL-C ratio, yielding acceptable sensitivity and speci city for predicting severe coronary atherosclerosis in this study.
Though hyperglycemia can induce oxidative stress and modify LDL-C into pro-atherogenic oxLDL [8] [42] . There was no correlation between oxLDL/LDL-C ratio and levels in HbA1c or LDL-C in our study which was con ict with the previous report [43] . This discrepancy in the results could be due to the poorly glycemic control and the effect of statin therapy at baseline. It is noted that with the lowest LDL-C levels and highest rate of statin-use, severe group still had the highest oxLDL/LDL-C ratio. These results indicated that statin treatment may be useless in decreasing circulating oxLDL/LDL-C ratio. Studies investigating association between circulating oxLDL and the statin treatment yielded con icting results.
In several studies, the levels of circulating oxLDL were signi cantly decreased by treatment with statins in patients with type 2 diabetes [44] and CHD [45] . In contrast, Kudret Keskin et al showed that statin treatment had no effect on serum oxLDL level [46] .Our study provided new evidence to provethat statin treatment can decrease both LDL-C and oxLDL levels.
Besides oxLDL/LDL-C ratio, other interested biomarkers such as FGF21, IL33 and VCAM-1 which were reported to be associated with diabetes or CAD were also evaluated in the present study. None of them were indicated to be correlated with the severity of coronary atherosclerosis in patients with T2DM. Only VCAM-1 had the similar trend with oxLDL/LDL-C ratio that had the highest levels in the patients of severe group with statistical signi cance. As a classical cell adhesion molecules, VCAM-1 combined with oxLDL/LDL-C ratio may be more effective than oxLDL/LDL-C ratio alone to predict and screen severe coronary atherosclerosis. IL33 was positively correlated with oxLDL/LDL-C ratio. Moreover, IL33 levels were higher in the high oxLDL/LDL-C ratio group than in the low group. Previous study had demonstrated that oxLDL can stimulated human umbilical vein endothelial cells and vascular smooth muscle cells to release large amounts of thymic stromal lymphopoietin (TSLP). TSLP-IL33 signaling pathway might interact with each other in Th2 cell-mediated in ammatory responses to play its athero-protective role [47] . It may be the explanation for the correlation between oxLDL/LDL-C ratio and IL33. We proposed that IL33 may be an alarmin of oxLDL/LDL-C ratio in patients with T2DM. We also found that all of the three biomarkers including oxLDL/LDL-C ratio, IL33 and VCAM-1 were higher in the diabetic patients with signi cant coronary lesions than in the patients with non-signi cant lesions. These results further supported the value of these three biomarkers in predicting and screening severe coronary atherosclerosis in patients with T2DM.
Some limitations of this study should also be considered. Firstly, the number of the samples was relatively small and recruited patients were all inpatients with poor glycemic control. Finally, it was a cross-sectional study, lacking the long-term outcome of coronary atherosclerosis and oxLDL/LDL-C ratio changing in patients with T2DM. It needs further follow-up study to identify the detailed association between the oxLDL/LDL-C ratio and the development of coronary atherosclerosis in type 2 diabetes population.

Conclusion
In conclusion, our study demonstrated for the rst time that oxLDL/LDL-C ratio is signi cantly associated with the severity of coronary atherosclerosis and predict severe coronary lesions in patients with type 2 diabetes. It may be more effective to test the combination of oxLDL/LDL-C ratio, VCAM-1 and IL33.
OxLDL/LDL-C ratio is considered as a potential biomarker in diabetic patients for early recognition and intervention of severe coronary atherosclerosis.

Declarations
Ethics approval and consent to participate The study protocol was approved by the Human Research and Ethics Committee of Shanghai East Hospital, and the study was conducted according to the principles of the Helsinki Declaration II. All patients provided written informed consent.

Consent for publication
Not applicable.

Availability of data and materials
The datasets used and analyzed during the current study are available from the corresponding author on reasonable request.  Rate of severe coronary atherosclerosis in low and high oxLDL/LDL-C ratio groups of diabetic patients.
*p<0.05 compared with low oxLDL/LDL-C ratio group.  ROC curve for oxLDL/LDL-C ratio in prediction by severe coronary atherosclerosis.

Figure 5
Serum level of LDL-C, oxLDL and oxLDL/LDL-C in statin-using or non-statin-using group. *p<0.05 compared with non-statin-using group.