A Nonlinear Association Between Gamma-Glutamyltransferase to High-Density Lipoprotein Cholesterol Ratio and Arterial Stiffness in Japanese Population: A Secondary Analysis Based on a Cross-Sectional Study

Introduction

Brachial-ankle pulse wave velocity (baPWV) is a reliable indicator for quantitatively assessing the degree of arterial stiffness.¹ As an independent risk factor for cardiovascular events, baPWV is widely used in health check-ups and epidemiological investigations.² Accumulating evidence suggests that higher baPWV was associated with an increased risk of future cardiovascular and cerebrovascular disease,^1,3,4 hypertension⁵ and diabetes.⁶ Considering the clinical significance of baPWV, it is of vital importance to investigate the reliable biomarkers to identify high-risk populations.

Gamma-glutamyltransferase (GGT) is an indicator widely used in clinical practice for evaluating the degree of liver function impairment. Evidences suggested a close relationship between GGT and insulin resistance (IR), and an increased GGT was an independent predictor for the risk of Type 2 diabetes.^7–9 Moreover, elevated GGT has been shown to associate with the chronic inflammatory reaction derived from hypersensitive C-reactive protein (hs-CRP).¹⁰ In addition, higher GGT was shown as an independent predictor for cardiovascular mortality,¹¹ stroke¹² and metabolic syndrome.¹³ A lipid indicator, high-density lipoprotein cholesterol (HDL-C) could stimulate glucose uptake and improve β-cell function, while a lower HDL-C suggesting decreased β-cell function and IR.^14,15 Furthermore, lower HDL-C was also a manifestations of metabolic syndrome¹⁶ and a risk factor for diabetes.^17,18

In recent years, accumulating studies focused on the the predictive value of GGT to HDL-C ratio (GTHR) in diabetes,¹⁹ nonalcoholic fatty liver disease,^20,21 metabolic syndrome²¹ and cardiovascular disease mortality.²² Moreover, diverse relationship including the linear, nonlinear or J-shaped relationships were shown.^19–22 Although the exact pathogenesis is still undetermined, it is well established that IR was closely associated with AS.²³ Considering the association between GGT, HDL-C, and IR, we speculated that GTHR may play a role in the development of AS. Therefore, we aimed to investigate the relationship between GTHR and AS, to better screen the high risk patients, therefore providing an early lifestyle intervention in these patients.

Methods

Study Population and Data Collection

The data in the present study were obtained from a open access database “DATADRYAD”. Researchers can download the raw data freely from the official website (www.Datadryad.org). In this study, we re-analyzed the data from a previously published study titled “Association between serum γ-glutamyltranspeptidase and atherosclerosis: a population-based cross-sectional study (10.5061/dryad.m484p).” It is a health check-up program carried out by Takuya Fukuda et al at Murakami Memorial Hospital, Gifu, Japan between March 2004 and December 2012, which aims to improve public health by early screening the patients at high risk of chronic diseases and identifying the potential risk factors. The inclusion and exclusion criteria and the study flowchart are shown in Figure 1. A total of 912 individuals were included in this study. Written informed consent were obtained from all the participants and the study was consistent with the principles of Declaration of Helsinki.

Figure 1 The flowchart of this study.

Assessment of baPWV, GTHR and Other Indicators

The indicators included in this study are consistent with the original data file. The demographic characteristics including age, sex, body mass index (BMI) were recorded. Comorbidities including blood pressure, smoking, drinking, exercise situation and fatty liver condition were re-analyzed. The laboratory indicators including alanine aminotransferase (ALT), aspartate transaminase (AST), γ-glutamyltranspeptidase (GGT), fasting glucose, uric acid, estimated glomerular filtration rate (eGFR), total cholesterol (TC), triglyceride (TG), low-density lipoprotein cholesterol (LDL-C) and high-density lipoprotein cholesterol (HDL-C) were compared. The details of blood biochemical testing and the data collection have been well described in the original study, therefore we did not repeat it. The automatic waveform analyzer (Colin Medical Technology, Komaki, Japan) was used to measure baPWV to assess the degree of AS quantitatively. The method used and the notes for attention were described in detail in the original study. In this study, AS was determined when baPWV >1400 cm/s according to the previous study.

Statistical Analysis

Continuous variables were shown as mean ±standard deviation or median (quartile), which were compared using one-way ANOVA or Kruskal–Wallis H-test, respectively. Categorical data were displayed in percentage, which were compared by chi-squared tests. We used the univariate linear regression model to assess the relationship between LogGTHR and baPWV. The multivariate models including the unadjusted, minimally adjusted and fully adjusted analyses were displayed in this study. The indicators adjusted followed the rules: when added the parameter to this model, changed the matched odds ratio by at least 10%.¹⁷ Furthermore, we also used a generalized additive model to identify the nonlinear relationship. If nonlinear relationship was determined, a two-piecewise linear regression model was used to evaluate the threshold effect of LogGTHR on baPWV in terms of the smoothing plot. Recursive method calculates automatically the inflection point, where the maximum model likelihood will be used. The stratified linear regression models were used to perform subgroup analyses. The modification and interaction of subgroup were inspected by the likelihood ration test. All of the data analysis were carried with the R packages. A two-side P-value<0.05 was considered statistically significant.

Results

Baseline and Clinical Characteristics

A total of 912 participants were included in this study. Patients were divided to three groups according to the the tertiles of the LogGTHR. We discovered that patients in the high LogGTHRG group had a higher incidence of male, fatty liver disease, current smoking, ex-smoking, AS, an increased level of BMI, systolic and diastolic blood pressure, AST, ALT, γGTP, fasting glucose, uric acid, TC, TG, LDL-C, GTHR, alcohol consumption, baPWV, while a decreased level of HDL-C, eGFR (P<0.05) (Table 1). Other indicators including age and ABI were comparable among the three groups (P>0.05) (Table 1).

Table 1 Baseline Characteristics of Participants According to LogGTHR Tertiles

Univariate Analysis

Univariate analysis showed that male, age, SBP, DBP, AST, ALT, fasting glucose, γGGT, uric acid, TC, TG, HDL-C, LDL-C and fatty liver were positively associated with AS (P<0.05) (Table 2). We also found that eGFR was negatively associated with higher AS (P<0.05) (Table 2). Parameters including BMI, smoking status, alcohol consumption, exercise and ABI were not correlated with AS (P>0.05) (Table 2).

Table 2 The Results of Univariate Analysis

The Relationship Between GTHR and baPWVs

The univariate linear regression model was performed to investigate the relationship between LogGTHR and baPWV. We found that LogGTHR was positively correlated with baPWV in the nonadjusted model (β= 162.94, 95%CI: (confidence interval) 110.67 to 215.21, P<0.0001) (Table 3). We also discovered a significant change in minimally adjusted model (β= 131.93, 95%CI: 79.56 to 184.30, P<0.0001). However, no statistical significance were found in full adjusted models (β= 24.18, 95%CI: −35.99 to 84.34, P=0.4312) (Table 3). In addition, when LogGTHR was considered as categorical variable (tertiles), the same trend was also observed in nonadjusted and minimally adjusted model (trend was P<0.001) (Table 3).

Table 3 Relationship Between GTRH and baPWV in Different Models

The Analyses of Nonlinear Relationship

Since LogGTHR was continuous data, therefore the nonlinear analysis was performed. We discovered that there is a nonlinear relationship between LogGTHR and baPWV (β= 30.08, 95%CI: −33.12 to 93.29, P=0.351) after adjusting for age, sex, BMI, SBP, DBP, AST, ALT, GGT, uric acid, fasting glucose, TC, HDL-C, LDL-C, eGFR, smoking, and exercise status, fatty liver, alcohol consumption, and ABI (Figure 2). The inflection point was 1.43. When K<1.43 (80.37% of the participants), the relationship was positively correlated (β= 165.15, 95%CI: 62.72 to 267.58, P=0.0016), while when ≥1.43, there was a negative association between LogGTHR and baPWV (β= −106.23, 95%CI:-209.25 to −3.21, P=0.0436). (Table 4).

Table 4 The Results of Two-piecewise Linear Regression Model

Figure 2 The relationship between GTHR and baPWV. A nonlinear relationship between them was detected after adjusting for age, sex, BMI, SBP, DBP, AST, ALT, uric acid, fasting glucose, TC, TG, LDL, eGFR, smoking, and exercise status, fatty liver, alcohol consumption and ABI.

Subgroup Analyses

The subgroup analysis were shown in Table 5. We found that the test for interactions were significant for sex and fatty liver disease (P<0.05). Meanwhile while, no statistical significance were found in age, current smoking, exercise status, alcohol consumption, BMI and uric acid (P>0.05) (Table 5). We adjusted the menopausal status in female patients in the subgroup analysis, and found that the effect size of GTHR on baPWV was not altered (P>0.05) (Table 5).

Table 5 Effect Size of GTHR on baPWV in Subgroups

Discussion

In the present study, we found GTHR was not associated with baPWV in the fully adjusted model. However, after adjusting for confounders, there is a nonlinear relationship between GTHR and baPWV. The inflection point was 21.06. When K<21.06, GTHR was positively correlated with baPWV, while when ≥21.06, the association was not statistically different. To the best of our knowledge, this is the first study to explore the potential relationship between the GTHR and AS.

AS was suggested as an early sign for the evaluation of vascular damage, which is strongly associated with poor prognosis in patients with cardiovascular disease.²⁴ In addition, patients with AS are more predisposed to develop hypertension, diabetes, obesity, and metabolic syndrome.^25–28 Although AS is the result of joint action of various factors, IR may play a role in this process via low grade inflammation and oxidative responses, causing endothelial damage and AS formation.^29,30 Traditionally, the hyperinsulinemic euglycemic clamp is the gold standard for evaluating the status of IR.³¹ Due to the high expenses, the complex and time-consuming assessment process, it is not an ideal method for the diagnosis of IR in routine clinical practice.³¹ Recently, some novel and simple indicators have been reported to be surrogates of IR, such as triglyceride-glucose index (TyG). Compared with the hyperinsulinemic-euglycemic clamp, the TyG index has been widely used in all individuals regardless of their insulin treatment status.³² Since then, the association between IR and AS have been widely discussed. IR derived from HOMA index,³³ TyG³⁴ and TyG-BMI³⁵ have been associated with the presence of AS.

In the present study, we first reported the nonlinear relationship between LogGTHR and AS. We discovered that LogGTHR was positively related with baPWV when less than 1.43, while this association was negative when LogGTHR was >1.43. Evidence suggested that GGT and HDL-C were reliable parameters for evaluating IR.^7–9,14,15 A higher GTHR means a higher GGT or a lower HDL-C, therefore, elevated GTHR is associated with IR. In a more recent cross-sectional study, which included 1665 Type 2 diabetes patients, the authors suggested that GTHR had a superior predictive value than TyG for the presence of metabolic syndrome.³⁶ This indicated that GTHR was also a reliable indicator for evaluating IR, therefore, could be used a risk factor for AS. The association between GTHR, a surrogate marker of IR, and AS is also supported by the pathophysiological mechanism. Patients with IR had a disorders of glucose and lipid metabolism, causing vascular endothelial dysfunction and oxidative stress, therefore bringing in the vascular remodeling and arterial stiffness.³⁷ In addition, chronic inflammation was often found in the phase of IR, which may lead to the damage of vascular endothelial dysfunction, resulting in the development of AS.³⁸ In the present study, we suggested that when less than 1.43, logGTHR was positively associated and IR was an important pathogenesis in this association. However, this association was negative when LogGTHR was more than 1.43. An increased GTHR affects the arterial stiffness up to a certain point beyond which factors such as hepatic dysfunction and medication effects counterbalance the association, reflecting the saturation. The exact nature of this correlations could not determined by the present study. However, considering GTHR as an indicator for IR, patients with high risk of AS could be early identified when LogGTHR was <1.43.

Our study has several limitations: (1) This is a cross-sectional study; therefore, causal relationships could not be determined. (2) Since the data were acquired from the published sources, variables not included in the dataset could not be adjusted, which may have a influence on the results. (3) In our study, we used baPWV >1400 cm/s for the determination of AS, so the results should be interpreted in this context. (4) The participants were healthy check-up individuals from Japan, so it is unclear whether these results could apply to other populations.

Conclusion

There is a nonlinear relationship between LogGTHR and baPWV in the present study. LogGTHR was positively related with baPWV when <1.43, while this association was negative when LogGTHR was >1.43. Large scale, multicenter, prospective cohort studies are still needed to confirm our conclusions.