Serum alkaline phosphatase is associated with arterial stiffness and 10‐year cardiovascular disease risk in a Chinese population

Serum alkaline phosphatase (ALP) has been recognized as a biomarker of cardiovascular disease (CVD) risk, recently. This study aimed to explore the association of ALP with arterial stiffness and 10‐year CVD risk.


| INTRODUCTION
With rapid economic growth and the ageing of the population, the prevalence of cardiovascular disease (CVD) has rapidly and substantially increased in China. 1 CVD is a leading cause of death in China, being the cause of 40% of deaths in the Chinese population, 2 representing a major public health concern and severe economic burden. Despite improved understanding and treatment of its risk factors (eg diabetes and dyslipidaemia), the prevalence of CVD has rapidly increased. Hence, investigation of additional modifiable risk factors is urgently needed.
Alkaline phosphatase is an orthophosphate monoester phosphohydrolase and commonly measured in clinical practice as a marker of hepatic or bony disease. Recently, a growing body of evidences has demonstrated that high ALP level, even within the normal range, is significantly correlated with risk of CVD. 3,4 Furthermore, population-based studies have suggested that elevated ALP level is a risk factor for all-cause and cardiovascular mortality in people with or without kidney disease. 5,6 Extensive evidence has indicated that brachial-ankle pulse wave velocity (baPWV) is a reliable marker of subclinical atherosclerosis and may be a predictor of cardiovascular events. 7,8 The Framingham risk score (FRS) is the most applicable method to evaluate CVD risk and to predict the cumulative 10-year CVD risk. 9 Although the association between ALP levels and CVD risk has been widely investigated, there are limited data on the association of ALP levels with baPWV and FRS. Therefore, the aim of the current study was to explore the association of ALP levels with baPWV and FRS in a general Chinese population.

| Study participants
This study represented a retrospective analysis of 12 539 participants (6915 men and 5624 women, age 20-79 years) who underwent a health examination at the Health Promotion Center of the First Affiliated Hospital of Nanjing Medical University, between September 2017 and December 2019. We excluded participants with known hepatic, biliary or bone disease, the previous history of cardiovascular events, arteriosclerosis obliterans, cerebrovascular accident, malignancy, systemic acute or chronic inflammatory diseases and reduced renal function with GFR < 60ml/(min × 1.73m 2 ). The study was approved by the Human Research Ethics Committee of the First Affiliated Hospital of Nanjing Medical University.

| Data collection
Each participant completed a standard questionnaire to selfreport their smoking habit, history of acute and chronic illnesses and drug use. Weight, height and blood pressure were measured in accordance with international standards. The body mass index (BMI) = weight (kg)/height (m) 2 . All blood samples were obtained from the antecubital vein after 10-hour overnight fast. Routine biochemical analyses including lipid profile, fating blood glucose, renal function and serum ALP level were measured by enzymatic methods (Chemistry Analyzer Hitachi 7020). Glycated haemoglobin A1c (HbA1c) values were measured by high-performance liquid chromatography.

| BaPWV Measurement
VP-1000 automated PWV/ABI analyser was used to measure baPWV. Briefly, the participants were examined in the supine position after resting for 10-15 minutes. BaPWV was measured as previously described. 10 We collected data including ΔTba and La-Lb and calculated baPWV. ΔTba was expressed in the time interval between the brachial and ankle waveforms. La-Lb was expressed in the distance between the brachium and the ankle estimated automatically according to the subject's height. baPWV was calculated as (La-Lb)/ ΔTba. The average value of left and right baPWV was used for analysis.

| Framingham 10-year risk estimation
Framingham risk score is a simplified and common tool for the assessment of risk level of CVD over 10 years. FRS was calculated based on the six coronary risk factors including age, sex, total cholesterol (TC), high-density lipoproteincholesterol (HDL-C), systolic blood pressure (SBP) (treated or untreated), smoking habits in men and women, separately. 11 Absolute CVD risk percentage over 10 years was classified as low risk (< 10%), intermediate risk (10%-20%) and high risk (> 20%).

| Statistical analysis
Continuous variables were expressed as mean ± SD, and categorical variables were expressed as percentages (numbers). One-way ANOVA was used to compare means, and the Pearson chi-square test was used to compare proportions. Age-adjusted baPWV means and standard errors were calculated using analysis of covariance (ANCOVA) according to serum ALP quartiles. Pearson's correlation was used to analyse the bivariate correlation between baPWV, Framingham risk score and clinical variables. The binary logistic regression analysis was performed to investigate the independent association of serum ALP level with high baPWV (0 = normal baPWV, 1 = high baPWV) and the 10-year CVD risk (0 = low risk, 1 = intermediate or high risk). Receiver operator characteristic (ROC) analyses were performed to calculate area under the ROC curve (AUROC) of serum ALP level for incident high baPWV. Dose-response association of serum ALP level with baPWV and elevated baPWV was conducted using generalized additive model (GAM) and a fitted smoothing curve (penalized spline method). All data were analysed using SPSS18.0 statistical software and Empower (R). The tests were considered significant when the P-value was < .05.

| Baseline characteristics of the study population
The study included 12 539 participants (6915 men and 5624 women) with a mean age of 49.31 ± 9.91 years. Demographic and clinical characteristics of the study population are listed in Table 1. Participants were stratified into four groups based on their ALP levels. Participants in the upper quartiles of ALP levels were more likely to older than those in the lower quartiles of ALP levels. The proportion of smoking, BMI, SPB, DBP, WBCC, neutrophil count, FBG, HbA1c, TC, TG, LDL-C, uric acid, ALT, AST, GGT and Framingham risk score were higher, while HDL-C was lower in the upper quartiles of ALP levels than those in the lower quartiles of ALP levels for both sexes. For men, the mean values of age, SPB, DBP, WBCC, neutrophil count, FBG, HbA1c, TG, uric acid, ALT, AST, GGT and Framingham risk score were higher, while HDL-C was lower in the upper quartiles of ALP levels than those in the lower quartiles of ALP levels. For women, the mean values of age, BMI, SPB, DBP, FBG, HbA1c, TC, TG, LDL-C, uric acid, ALT, AST, GGT and Framingham risk score were higher, while HDL-C was lower in the upper quartiles of ALP levels than those in the lower quartiles of ALP levels.

| BaPWV and the 10-year CVD risk compared across the quartiles of ALP levels
Age-adjusted mean baPWV significantly increased with increasing quartiles of ALP levels in the overall population (1233.99 ± 207. 23 Figure 1). Figure 2 showed the association between serum ALP level and FRS categories among participants. As shown in this figure, the prevalence of the high CVD risk was significantly higher in the high serum ALP level group than in the low-level group, while the prevalence of the low CVD risk was significantly higher in the low serum ALP level group than in the high-level group.

Framingham risk score and clinical variables
Pearson's correlation analysis showed that baPWV was positively correlated with age, BMI, SPB, DBP, WBCC, neutrophil count, FBG, HbA1c, TC, TG, LDL-C, uric acid, creatinine, ALT, AST, GGT and ALP ( Figure S1), while negatively correlated with HDL-C. The relationships between Framingham risk score and clinical variables, including ALP, were similar to the associations between baPWV and clinical variables ( Table 2).
Overall, there were significant positive associations of serum ALP level with baPWV and elevated baPWV ( Figure 3A,B). Per one unit increment in serum ALP level, baPWV is changed in 0.998 m/s (95% CI 0.821-1.138, P < .001) according to the estimation from regression coefficients indication, and the odds ratios (OR) of the risk of elevated baPWV were 1.134 (95% CI 1.103-1.165, P < .001).

| Binary logistic regression analysis
showing the independent contribution of serum ALP level to the 10-year CVD risk As Framingham risk score was calculated by age, sex, smoking status, SBP, TC and HDL-C, we only adjusted for BMI, DBP, WBCC, neutrophil count, HbA1c, FBG, TG, LDL-C and uric acid in the binary logistic regression. Serum ALP level was independently associated with the

| Performance of serum ALP level for diagnosing subjects with elevated baPWV
We further conducted a ROC curve analysis to assess the diagnostic value of serum ALP level (Figure 4 and Figure 5). The area under the Receiver operator characteristic (AUROC) curve to analyse the ability of serum ALP level to identify individuals at risk for elevated baPWV was 0.740 (95% CI 0.726-0.754, P < .001) in women and the optimal cut-off point for serum ALP level was 84 U/L (sensitivity: 71.2%, specificity: 63.4%). In contrast, the AUROC of serum ALP level [0.575 (95% CI 0.559-0.590), P < .001] in men was relatively smaller than that in women, which indicated that serum ALP level might serve as a simple and effective tool for elevated baPWV risk assessment only in women.

| DISCUSSION
Here for the first time, we demonstrated the association of serum ALP level with arterial stiffness and 10-year CVD risk in a lager Chinese population. A major finding of our study was that serum ALP level was positively associated with baPWV as a marker of arterial stiffness and 10-year CVD risk evaluated by FRS, independent of classical cardiovascular risk factors in both women and men. Furthermore, gender differences in the effect of serum ALP level on identifying individuals at risk for elevated baPWV were reported in the current study. Alkaline phosphatase (ALP) is mainly expressed in bone, liver and kidney. Besides the role in increasing ALP level for hepatic or bony disease, elevated ALP is also significantly associated with other diseases including metabolic syndrome, diabetes and nonalcoholic fatty liver disease, [12][13][14] all of which are risk factors for CVD. 15,16 In this regard, there is a growing interest in the relationship of ALP level with CVD. A meta-analysis of prospective cohort studies showed that each standard deviation increment in the baseline ALP was associated with 8% greater risk of CVD. 17 A prospective population-based study found that ALP level >179 IU/L was associated with over 30% higher risk of CVD, independent of several established traditional CVD risk factors. 18 Two prospective cohort studies conducted in a German and Iranian population, respectively, have demonstrated that elevated ALP is independently associated with the risk of all-cause mortality in patients with coronary heart disease. 18,19 Moreover, medicines such as extra-terminal (BET) protein inhibitor apabetalone lowing serum ALP level are paralleled by a reduction in risk of cardiovascular events. 20 In view of the robust evidence presented in the review, Haarhaus et al 21 proposed that ALP was an evolving treatment target for CVD and metabolic syndrome. Thus, ALP is not only a marker of cardiovascular risk but also a novel treatment target for CVD. BaPWV is a simple, noninvasive method, which correlates well with arterial stiffness, and it is also a useful tool for identifying a subgroup in the population that are at increased risk F I G U R E 1 Age-adjusted BaPWV compared across the ALP level quartiles for all participants, men and women F I G U R E 2 Percentages of 10-CVD risk categorise compared across the quartiles of the ALP levels for cardiovascular events. 22 However, there are limited data about the association between serum ALP and baPWV, especially in a large Chinese population. To the best of our knowledge, there is only one study about the correlation between serum ALP and baPWV in Korean adults, 23 but the sample of this study was relatively small and did not calculate the cut-off value of serum ALP for high baPWV. In the present study, which was conducted in a relatively large number of Chinese adults, we found that serum ALP level was positively correlated with baPWV, which was consistent with the study conducted in Korean adults. 23 In addition, we also found that elevated ALP level was an independent risk factor for elevated baPWV for both men and women even adjustment for traditional CVD risk factors, demonstrating that serum ALP level can be an important biomarker of subclinical atherosclerosis. Furthermore, we observed significant gender differences in the predicted value of serum ALP level for high baPWV. By ROC curve analysis, the AUROC of serum ALP level to identify individuals at risk for elevated baPWV was 0.740 (95% CI 0.726-0.754) in women, larger than that in men (0.575, 95% CI 0.559-0.590). This result implied that serum ALP level  predictability of elevated baPWV could be better in women than in men. Consistent with our findings, there have been several previous studies indicating the gender difference in cardiovascular risk-predictive value of screening tools. Nakagomi A et al 24 found that all insulin surrogate markers including triglyceride glucose (TyG) index were also associated with increased baPWV and the associations were stronger in women than in men. Rhee TM et al 25 showed that baPWV was positively correlated with four cardiovascular different risk scores and the correlation was stronger in women than in men. A systematic review and metaanalysis also found that serum uric acid was positively association with baPWV in women, but not in men. 26 Along with these findings, our result highlights the evidence to support more powerful predicted value of serum ALP level for high baPWV in women than in men. The pathophysiology of the gender difference in high baPWV prediction by serum ALP has not yet been clearly understood and needs further study. Although underlying mechanisms of the association of ALP with baPWV are not properly known, some causes were suggested. First, ALP plays an important role in the process of vascular calcification. 27 We speculated that ALP mediates its deleterious prognostic impact through vascular calcification considered to be a marker of subclinical atherosclerosis.
Second, chronic low-grade inflammation plays a key role in initiating and developing of atherosclerotic process. Several experimental and clinical studies suggested that ALP was significantly correlated with low-grade inflammation factors such as serum hsCRP, TNF-a and IL-1β. 28,29 Here, we speculated that the potential impacts of ALP on baPWV may involve mechanisms relating to inflammatory pathways. In the present study, we found that serum ALP was independently correlated with baPWV after adjustment for white blood cell count and neutrophil count. But we did not investigate levels of other inflammatory factors such as hsCRP, TNF-a and IL-1β, so the potential role of serum ALP in subclinical atherosclerosis via inflammation needs to be explored in our further research. Some other causes including metabolic syndrome, fibroblast growth factor-23 and vitamin D deficiency have also been suggested as mechanisms to explain ALP associated with arterial stiffness. On the other hand, elevated ALP levels could be caused by nonalcoholic fatty liver disease (NAFLD). It is well established that NAFLD is closely associated baPWV. 30 Further studies on precise mechanism of ALP responsible for subclinical atherosclerosis are warranted.
The Framingham risk score (FRS) is a simplified and has been widely used in different populations to estimate the 10year risk of CVD. 31 A limited number of studies have assessed the association between serum ALP level and FRS. In the present study, we found that the prevalence of the high CVD risk was increased along with the increased level of serum ALP. Meanwhile, the prevalence of the low CVD risk was decreased along with the increased level of serum ALP. Of note, we also found that serum ALP level was an independent risk factor of intermediate or high 10-year CVD risk after adjusting for confound factors. FRS was calculated based on the six coronary risk factors including age, sex, TC, HDL-C, SBP and smoking habits. Several studies showed that serum ALP level was significantly correlated with hypertension and lipid profile. 32,33 The association of serum ALP with hypertension and lipid profile has also been suggested as causes to explain 10-year CVD risk associated with serum ALP level.
This study has several limitations. Firstly, it is difficult to determine whether serum ALP causative effect on arterial stiffness because of the cross-sectional design. Thus, further prospective study is warranted to explore whether participants with high ALP will develop arterial stiffness in future. Secondly, we did not detect inflammatory factors such as hsCRP, TNF-a and IL-1β, so the potential role of ALP in arterial stiffness via inflammation could not be studied. Thirdly, the participants in the current study were enrolled in a single hospital for health examination, and thus, the generalizability of the results may be difficult.

| CONCLUSION
Serum ALP level is positively associated with arterial stiffness measured by baPWV and 10-year CVD risk evaluated by Framingham risk score in the general Chinese population. Serum ALP may perform better in identifying individuals at risk for subclinical atherosclerosis of healthy population, especially for women. Prospective and multi-centre studies are needed to confirm our results.

| Founding information
Financial support for this project was from The Natural Science Foundation of the Jiangsu Higher Education Institutions of China (20KJB320009).