Association between remnant cholesterol and arterial stiffness: A secondary analysis based on a cross‐sectional study

Abstract The relationship between conventional lipid parameters and arterial stiffness (AS) has been verified by previous studies. However, it remains unknown whether non‐conventional lipid parameters have certain predictive effect on AS represented by brachial‐ankle pulse wave velocity (baPWV). Therefore, the study was to explore the relationship between remnant cholesterol (RC) and other non‐conventional lipid parameters and AS in the general population free from cardiovascular disease. The study included 912 participants aged 24–84 years from a medical health checkup center of Murakami Memorial Hospital. Logistic regression analysis and receiver operating characteristic (ROC) curves were used to examine the association between non‐conventional lipid parameters and AS. The results showed that compared with non‐AS group, the AS group had higher RC, non‐high‐density lipoprotein cholesterol (Non‐HDL‐C), atherogenic index of plasma (AIP), lipoprotein combine index (LCI), atherosclerosis index (AI), triglycerides/HDL‐C (TG/HDL‐C), Castelli's risk index I (CRI‐I) and Castelli's risk index II (CRI‐II). Then, the authors divided participants into two groups by the optimal cutoff point of 23.6 of RC determined by Youden index. The baPWV was significantly higher in higher RC group compared with lower RC group, and RC was positively correlated with baPWV. Multivariate Logistic regression analysis showed that, regarding lower RC as reference, higher RC was independently associated with higher risk of AS, independent of other risk factors (OR = 1.794, 95% CI: 1.267‐2.539, p = .001). The area under the curve of AS predicted by RC was higher than that of other non‐conventional lipid parameters (almost all p < .05). The findings indicated that increased RC was a significant predictor of AS.

factors of baPWV. Previous studies have shown that high blood pressure and advanced age were the major risk factors for AS. 8 Moreover, lipid metabolism is also one of the important factors leading to AS. Several studies have shown that conventional lipid parameters, including cholesterol and triglycerides, were related to AS. [9][10][11] However, as far as we know, the relationship between non-conventional lipid parameters and AS has not been reported yet, such as remnant cholesterol (RC).
Dyslipidemia is a recognized cause of CVD, which has been confirmed by many genetic and epidemiological studies. 12,13 Decreasing plasma low density lipoprotein cholesterol (LDL-C) is a key method for the prevention of CVD, which is strongly recommended in current guidelines. 14,15 However, patients with significantly reduced LDL-C still have a significant risk of cardiovascular residue. 16 Recently, emerging evidence shows that RC may be responsible for the abovementioned residual CVD risk. 17,18 RC represents the cholesterol composition of rich in triglyceride lipoproteins, which consists of chylomicron remnants, intermediate density lipoprotein and very low-density lipoprotein. 17 Several studies have discussed the association between RC and CVD, metabolic diseases and major adverse cardiovascular outcomes in different people. 16,[19][20][21][22][23][24] Although experimental studies have demonstrated that RC promotes the formation and development of atherosclerosis through adhesion and pro-inflammatory effects, which leads to the occurrence of related diseases, 25,26 data about the predictive implications of RC for AS derived from baPWV in the general population with different models is currently lacking.
Therefore, it is of great clinical significance to identify the relationship between RC and AS if we are to develop new therapeutic targets and to tailor risk reduction strategies that match individual risk level.
Based on this, the present study was conceived with the purpose of: (1) recognizing the potential connection between RC and AS; (2) exploring whether the predictive value of RC for AS is affected by different subgroups; (3) determining whether the predictive value of RC for AS is higher than that of other non-conventional lipid parameters.

Study population
The present study is a secondary analysis based on a cross-sectional study, which is to improve public health by early assessment of poten- the details of which have been described elsewhere. 28  where and will not be repeated here. [27][28][29] According to the original research design, 27   LDL-C/HDL-C. 37 In this study, we divided participants into two groups by the optimal cutoff point of RC determined by ROC curve analysis: lower RC (≤ 23.6; n = 376) and higher RC (> 23.6; n = 536).

RESULTS
The study included 912 participants (mean age: 51.1 ± 9.6 years; 64.9% men). Baseline characteristics of the total population and participants with or without AS were showed in Table 1. Participants in AS group had significantly higher RC compared with those non-AS group. Participants with AS had higher age and blood-pressure parameters, and higher prevalence of fatty liver. As for laboratory parameters, participants with AS showed higher levels of fasting glucose, uric acid, AST, ALT, GGT, TG, TC and LDL-C, while lower levels of eGFR. In terms of other non-conventional lipid parameters, those with AS had higher levels of non-HDL-C, AIP, LCI, AI, TG/HDL-C, CRI-I and CRI-II ( Figure 1).
ROC curve analysis indicated that the AUC of RC for predicting AS was 0.629 (95% CI 0.597-0.661, p < 001). The RC of 23.6 was used as the optimal cutoff point for predicting AS. In this study, we further divided participants into two groups by the optimal cutoff point of RC: lower RC (≤ 23.6; n = 376) and higher RC (> 23.6; n = 536). Baseline characteristics of participants stratified by the RC of 23.6 were displayed in Table 2. Compared with participants in lower RC group, those with higher RC appeared to be older, display higher levels of bloodpressure parameters, BMI and baPWV, and higher percentage of male, alcohol consumption and fatty liver. Laboratory indices including fasting glucose, uric acid, AST, ALT, GGT, TG, TC and LDL-C were significantly higher in participants with higher RC, while HDL-C and eGFR levels were comparatively lower. With regard to non-conventional lipid parameters, participants with higher RC also exhibited higher non-HDL-C, AIP, LCI, AI, TG/HDL-C, CRI-I and CRI-II compared to participants with lower RC.
The Pearson or Spearman's rank correlation analysis was used to examine the correlations between the RC and other variables. The RC was positively related to age, alcohol consumption, fatty liver, BMI, blood-pressure parameters, baPWV, TG, TC, LDL-C, fasting glucose, uric acid, AST, ALT and GGT, while negatively related to sex, HDL-C and eGFR (Table 3).
In Logistic regression analysis, four models (crude model and Model  Figure 3).

DISCUSSION
As far as we know, the present study was the first report on the relationship between RC and AS. In the present study, we retrospectively explored the predictive importance of RC for AS. The main findings were as follows: (1) the RC was related to various risk factors of CVD; (2) compared with participants in lower RC group, those with higher RC showed higher levels of baPWV; (3) the higher RC (per 1-unit increase or regarding lower RC as reference) was a strong independent risk predictor of AS in our population, although after adjusting for possible interference factors; (4) the discriminant ability of RC for predicting AS was significantly higher than that of other non-conventional lipid parameters. These findings demonstrate that RC is independently associated with the risk of AS in the general population free from cardiovascular disease.
AS has been widely considered as an important risk factor for CVD, and it is very common in patients with CVD. Some studies have shown that AS is significantly associated with preclinical disease, development of coronary heart disease and poor prognosis. 3,38,39 Therefore, for these patients, the quantitative evaluation of the degree of AS has important clinical significance for primary and secondary prevention of CVD. Percutaneous angiography has been recognized as the gold standard approach for the diagnosis of AS. However, the procedure is not only invasive, but also time-consuming and costly, which makes it relatively difficult to apply in clinical practice. Therefore, a non-invasive, convenient and cheap index to evaluate AS came into being, namely baPWV. A large number of studies have shown that the baPWV, as a reliable alternative quantitative marker of AS, is related to the traditional CVD risk factors including dyslipidemia and has been proved to be a encouraging biomarker for the occurrence and development of CVD. 2,6,40 However, the occurrence and development of AS is inseparable from glucose and lipid metabolism and insulin resistance, of which lipid metabolism is the most important, including conventional lipid parameters and non-conventional lipid parameters. 1 As we all know, the one that has attracted the most attention is LDL-C.
Raised LDL-C is currently widely used as the primary intervention target, which is attribute to its characteristics as a major risk factor for CVD. 15 However, when LDL-C is reduced below the recommended level, the risk of cardiovascular events is still high, that is, the recognized residual risk, which has attracted the active attention of the majority of scholars. 26 With the increase of attention, a growing number of studies have found that RC is responsible for the residual risk, which is mainly due to the ability of RC to converge various atherogenic effects, including the activation of monocytes, the upregulation of pro-inflammatory factors and the increase of prothrombotic factors production. 17,18,25,26 From this point of view, RC seems to have the same atherogenic effect as LDL-C, but the pathophysiological mechanism is not clear. Even so, the idea that higher RC tends to be associated with higher prevalence of AS can also be reasonably explained, as shown in this study.
In addition to basic research, many clinical studies also showed that higher RC was associated with an increased risk of CVD, independent of other traditional risk factors. Except for observational studies, genetic studies have strongly confirmed that higher RC is a fatal TA B L E 2 Baseline characteristics of participants stratified by the optimal cutoff point of RC   The Pearson's correlation coefficients were evaluated to detect the variables associated with the RC, including age, body mass index, systolic blood pressure, diastolic blood pressure, pulse pressure, mean blood pressure, baPWV, total cholesterol, low-density lipoprotein cholesterol, high-density lipoprotein cholesterol, uric acid and estimated glomerular filtration rate; The Spearman's rank correlation coefficients were evaluated to detect the variables associated with the RC, including sex, smoking status, alcohol consumption, regular exercise, fatty liver, triglycerides, fasting glucose, aspartate transaminase, alanine aminotransferase and γ-glutamyltranspeptidase. The RC was regarded as a continuous variable in this analysis. Abbreviations: RC, remnant cholesterol; baPWV, brachial-ankle pulse wave velocity. factor in CVD. 12 And previous studies have shown that RC can be used not only as a prognostic index of CVD, 16 but also as a predictor of CVD. 12 It is also reported that RC is inextricably linked with other diseases, including aortic valve stenosis, 22 stroke, non-alcoholic fatty liver disease, 20 diabetic complications 41 and renal dysfunction. 42 Moreover, RC is also a good predictor for subclinical CVD. As found by Lin A and coworkers and Elshazly MB and coworkers, RC is not only related to the burden of coronary atherosclerotic plaque, 43 but also to the progression of coronary atherosclerosis. 43 17 However, these indicators are not routinely measured in clinical practice, which makes them not suitable for popularization and application. Gven this, RC, which is derived from the commonly used clinical indicators (TC, LDL-C and HDL-C), has been proposed and widely used, and the RC in this study was calculated according to this accepted formula. 18 Although some studies have suggested that non-HDL-C can be used as an alternative index of RC, 45 and previous studies have shown that non-HDL-C can be used as an independent predictor of CVD, 33,36 the non-HDL-C level obtained in this study is not only much higher than the calculated RC, but also the predictive value of AS is lower than RC. Therefore, the calculated RC is sufficient to meet the needs of clinical management and scientific research. In addition to RC and non-HDL-C, previous studies have also shown that AIP, AI, LCI, CRI-I, CRI-II and TG/HDL-C are associated with cardiovascular disease, 11,46-49 but we found that the predictive performance of these non-conventional lipid parameters is lower than that of RC. In view of this, RC seems to be the most important atherogenic indicator besides LDL-C. Nowadays, with the development of precision and targeted therapy, accurate measurement of each pathogenic component is very important for disease management. A large clinical trial has reported that intensive lipid-lowering therapy in patients with higher RC can reduce the risk of CVD. 50 Taken together, these data suggest that RC may be both a prognostic marker and a potential target for future therapeutic interventions.
Innovatively, our findings added to the evidence of RC and CVD from clinical to subclinical diseases. Moreover, we comprehensively integrated different forms of non-conventional lipid parameters for the first time and compared their predictive value for AS. Therefore, this study provided additional information that the estimation of RC and other non-conventional lipid parameters may be of clinical significance in primary prevention to identify people at risk of CVD. In spite of this, several limitations still existed in this study. Firstly, the present study was a secondary analysis based on a cross-sectional study, so it was impossible to obtain all the original data. Secondly, the present study was also a cross-sectional study, which could not identify the causal relationship between RC and AS. Moreover, in Logistic regression analysis, we only controlled for several meaningful confounding factors, but there might be other confounding factors not included in our study, such as inflammatory indicators and medication history including antidyslipidemic and anti-hypertensive medications. Besides, dichotomizing baPWV may lead to several problems, such as the loss of information, the increase of potential confounding factors, the decrease of statistical ability and the bias of results. Furthermore, in this study, defining baPWV > 1400 cm/s as an indicator of higher AS may overestimate the prevalence of AS. Finally, The data of this study only came from the general population of a single ethnic group, so the findings may not be applicable more populations broadly.

CONCLUSIONS
Taken together, this is the first study to demonstrate that higher RC was a independent predictive factor for participants with AS, which might enrich the research field of predictors of AS, introduce an available indicator for risk management of AS and provide new insights on the necessity of monitoring RC in participants with impaired lipid metabolism for CAD risk assessment. Further prospective and randomized studies need to be conducted to determine whether interventions for RC have a positive effect on improving AS.