Hemoglobin A1c Levels Affect Visit-to-visit Variability of Lipid Proles in Patients Undergoing Elective Percutaneous Coronary Intervention: A Retrospective Study

Both hemoglobin A1c (HbA1c) levels and visit-to-visit variability of lipid proles are risk factors for cardiovascular disease (CVD). We conducted a retrospective cohort study to explore the relationship between HbA1c and lipid variability. We retrospectively collected baseline and follow-up data on patients who underwent elective percutaneous coronary intervention (PCI) from 2009 to April 2019. Univariate and multivariate linear regression analyses were performed to assess the association between HbA1c and lipid variability. Subgroup analyses employed multivariate linear analyses. ndings study follows: (1) HbA1c level was a potential risk factor for the variability of LDL-C, non-HDL-C, TC and TG. The results were consistent when SD, CV, and VIM were used to represent lipid variability, respectively; (2) Subgroup analysis demonstrated that the effect HbA1c on the variability of LDL-C, non-HDL-C, TC, and TG remained similar in several relevant subgroups. In patients without diabetes, the signicant positive correlations between HbA1c and lipids variability remained, including LDL-C, non-HDL-C, TC and TG.

HbA1c 6.5% group. Multivariate linear regression indicated that HbA1c level was a potential risk factor for the variability of LDL-C, non-HDL-C, TC and TG, which was independent of the mean values of lipids.
Subgroup analyses demonstrated that the relationship between HbA1c and the variability of LDL-C, non-HDL-C, TC, and TG did not importantly vary across several subgroups. These results remained consistent when lipid variability was represented by the standard deviation (SD), coe cient of variation (CV) and variability independent of the mean (VIM), respectively.

Conclusion
HbA1c is a potential risk factor for the variability of LDL-C, non-HDL-C, TC and TG in patients undergoing elective PCI. Background Cardiovascular disease (CVD) remains the leading cause of death globally and is one of the most common complications of diabetes mellitus (DM) [1,2]. Hemoglobin A1c (HbA1c) is usually regarded as an indicator representing the average blood glucose levels over the past 2-3 months and is often employed to assess glycemic control [3]. Several epidemiological studies [4,5] have demonstrated associations between elevated HbA1c levels and adverse CVD outcomes in patients with type 2 DM.
Moreover, even within the normal reference range, increased blood glucose levels are associated with an increased risk of CVD [6].
Traditionally, mean values of indicators or exposures are often employed to estimate the prognostic risk for CVD. For example, higher low-density lipoprotein cholesterol (LDL-C) levels and lower high-density lipoprotein cholesterol (HDL-C) levels are associated with an increased risk of CVD [7,8]. However, if there is wide variability in the exposure, the mean value will no longer be reliable. In recent years, more attention is being focused on the variability of these exposures. With particular emphasis on lipid pro les, the visitto-visit variability of LDL-C, HDL-C, non-high-density lipoprotein cholesterol (non-HDL-C), total cholesterol (TC) and triglyceride (TG) has each been demonstrated to be a potential risk factor for CVD [9][10][11][12][13]; these associations are independent of the mean levels of the exposures and traditional cardiovascular risk factors. Based on intravascular ultrasound examination, elevated variability of LDL-C has been shown to independently promote the progression of atherosclerosis, which is the crucial underlying pathology of CVD [14] . In general, lipid variability, an independent risk factor for CVD, is attracting considerable attention.
Both HbA1c and the variability of blood lipids are risk factors for CVD. However, whether there is any relationship between HbA1c and the variability of lipids is still unclear. Therefore, the purpose of this study is to investigate the relationship between HbA1c and lipid variability in patients undergoing elective percutaneous coronary intervention (PCI).

Study subjects
The study population comprised all consecutive patients who attended the Sir Run Run Shaw Hospital, Zhejiang University in China, from January 2009 to April 2019. Eligibility criteria for inclusion into the study were: (1) patients who underwent elective PCI, (2) su cient clinical information such as HbA1c and lipid pro les were available at baseline and follow-ups, (3) patients were followed up at 1, 3, 6, 9, and 12 months following PCI according to the prescribed follow-up procedures. Patients with acute myocardial infarction (MI), active cardiopulmonary diseases, heart failure, severe liver and/or renal insu ciency, cancer, acute or chronic infection and other serious diseases were excluded from the study. A total of 4,445 patients were nally enrolled in the study. All PCI procedures were performed by experienced interventional cardiologists using the recommended guidelines [15]. Baseline and follow-up measurements of exposures were ascertained from fasting venous blood samples (at least 8 hrs overnight). Levels of HbA1c, LDL-C, HDL-C, non-HDL-C, TC and TG were determined at each follow-up.
Standard follow-up procedures and examinations were employed for all patients who had undergone elective PCI. The Ethics Committee of Sir Run Run Shaw Hospital of Zhejiang University approved the study (20200803-34).

Assessment of variability in lipid pro les
Baseline and follow-up values of lipids were measured by a blood chemistry analyzer (Hitachi 747; Hitachi, Tokyo, Japan). The variability of lipids re ected the degree of uctuation of individual blood lipid levels during the 1-year follow-up period. To assess the lipid variability more comprehensively, the following 3 indicators were used:(1) standard deviation (SD); (2) coe cient of variation (CV): The CV of blood lipids was de ned as follows, CV=(SD/mean)×100 (%); and (3) variability independent of the mean (VIM): VIM was calculated as (SD/mean β )×100 (%), in which β was derived from the tting of the curve and was the regression coe cient based on the natural logarithm of SD and the natural logarithm of the mean [16].

De nitions
The level of HbA1c was de ned as the mean value during a 1-year follow-up period and was calculated using values of HbA1c measured at follow-up. Diabetes mellitus was de ned as a fasting serum glucose≥126 mg/dL, a history of diabetes or the current use of antidiabetic medications. Hypertension was de ned as blood pressure≥140/90 mmHg, a documented history of hypertension or on antihypertensive medications. Regular statin therapy was de ned as atorvastatin≤20mg or rosuvastatin≤10mg per day. Intensive statin therapy was de ned as atorvastatin≥40mg or rosuvastatin≥20mg per day. Particular attention was paid to patients' compliance with statin use. Heart failure was de ned by ejection fraction (EF) < 40% or N-terminal pro B-type natriuretic peptide (NT-pro BNP) > 2000 pg/ml without renal failure. Body mass index (BMI) was calculated as weight in kilograms, divided by height in meters squared. Glomerular ltration rate (GFR) was estimated using the Japanese Society of Nephrology equation as follows: estimated GFR (eGFR) (mL/min/1.73 m 2 ) = 194×serum creatinine -1.094 ×age -0.287 (×0.739 for women) [17].

Statistical analysis
Continuous variables were presented as median (interquartile range) or mean (standard deviation).
Categorical variables were represented as frequency (%). Baseline characteristics were compared between patients with HbA1c 6.5% and HbA1c≥6.5% using chi-square tests for categorical variables and nonparametric tests for continuous variables. Univariate and multivariate linear regression analyses were used to evaluate the relationship between HbA1c and lipid variability. Subgroup analyses employed multivariate linear regression analyses. A value of P<0.05 (2 sided

Results of the multivariate linear regression in subgroups
In multivariate subgroup analyses, the relationship between HbA1c and the variability of LDL-C, non-HDL-C, TC, and TG did not signi cantly vary across several subgroups such as non-diabetes, diabetes, atorvastatin, rosuvastatin, regular statin, intensive statin and statin-ezetimibe combined therapy. Similarly, when CV and VIM were employed to represent lipid variability, respectively, subgroup analysis results were consistent with results when using SD. (see Table 6).

Discussion
The main ndings of this study are summarized as follows: (1) HbA1c level was a potential risk factor for the variability of LDL-C, non-HDL-C, TC and TG. The results were consistent when SD, CV, and VIM were used to represent lipid variability, respectively; (2) Subgroup analysis demonstrated that the effect HbA1c on the variability of LDL-C, non-HDL-C, TC, and TG remained similar in several relevant subgroups. In patients without diabetes, the signi cant positive correlations between HbA1c and lipids variability remained, including LDL-C, non-HDL-C, TC and TG.
Elevated visit-to-visit variability of blood lipids is associated with different adverse outcomes or organ dysfunction, including CVD, obstructive sleep apnea, renal function decline and cognitive decline [18][19][20]. Based on the Treating to New Targets (TNT) trial, the visit-to-visit variability of LDL-C, HDL-C and TG was demonstrated to be an independent predictor of cardiovascular events [21,22]. Moreover, Lee et al. [12] con rmed that the visit-to-visit variability of non-HDL-C was associated with major adverse cardiovascular and cerebrovascular events (MACCE) in patients who had undergone PCI. Furthermore, a study based on 3.6 million people in the general population showed that the variability of TC was associated with the risk of all-cause mortality, MI and stroke [23]. A wealth of data suggests that the visitto-visit variability of lipid pro les is independently associated with the risk of CVD. However, the underlying mechanisms for the association are still unclear. Similarly, HbA1c has been identi ed as a risk factor for CVD and con rmed by many epidemiological studies [4,5,24,25]. The current study has demonstrated the potential role of HbA1c as a risk factor for the variability of LDL-C, non-HDL-C, TC, and TG. To further explore the relationship between HbA1c and blood lipid variability, we postulate the following underlying mechanisms based on ndings from previous studies.
Insulin resistance (IR) is considered one of the underlying mechanisms, which play an essential role in glucose and lipid metabolism. In this study, IR may be ubiquitous among subjects. A quarter of the patients had a history of type 2 DM, which is very likely to be accompanied by IR [26]. Besides, within the normal range of glucose tolerance, patients with CVD have more IR compared to those without CVD, which suggests that patients without type 2 DM in this study may also have IR [27]. IR not only affects HbA1c levels by reducing glucose transport into cells and glycogen synthesis, but is also closely related to dyslipidemia [28]. Although the existing evidence is not su cient to ultimately demonstrate the effect of IR on lipid variability, the evidence for the role of IR in lipid metabolism appears to suggest this possibility. IR leads to dyslipidemia via the following ways: (1) increased TG, (2) decreased HDL-C, and (3) the appearance of small dense LDL particles [29]. Moreover, hyperinsulinemia, which is the response of insulin resistance, stimulates the synthesis and secretion of very LDL and promotes LDL-C transport into arterial smooth muscle cells [30,31]. In general, given the high prevalence of IR in our patients and its effects on glucose and lipid metabolism, IR could be the potential mechanism behind our ndings [32].
Based on analyses of baseline characteristics, the variability of LDL-C, non-HDL-C, TC and TG showed signi cant differences between categories of HbA1c. However, this nding was not observed for the variability of HDL-C. Unlike other blood lipids, HDL-C is often called "good cholesterol," which exerts multiple bene cial functions within the cardiovascular system [33]. There is a negative correlation between HbA1c and HDL-C levels. Reduced HDL-C levels often accompany elevated HbA1c [34]. To some extent, it is possible that this negative correlation weakened the in uence of HbA1c on the variability of HDL-C, hence the null ndings.
Both HbA1c and blood lipid variability are risk factors for adverse CVD outcomes. Based on the data of patients undergoing elective PCI, this study clari ed the relationship between HbA1c and blood lipid variability. First, insights on the relationships promote the understanding of the role of HbA1c and lipid variability in CVD development. Second, these ndings provide new perspectives for further understanding the relationship between glucose metabolism and lipid metabolism, which are closely related in vivo.

Limitations
There are a number of limitations to consider. First, being a single-center retrospective study, it is limited by inherent biases. Second, we had no data on antidiabetic medications, which may affect lipid metabolism or improve insulin sensitivity. Third, this study examined the effect of long-term blood glucose levels on lipid variability. Indicators of short-term blood glucose levels or glucose tolerance such as random glucose, fasting plasma glucose and 2h post-OGTT glucose, were not included in this study,

Conclusion
HbA1c is a potential risk factor for the variability of LDL-C, non-HDLC, TC, and TG in patients undergoing elective PCI.