Figures
Abstract
Background
It is demonstrated that elevated serum levels of alkaline phosphatase (ALP) and phosphate indicate a higher risks of cardiovascular disease (CVD) and total mortality in population with chronic kidney disease (CKD), but it remains unclear whether this association exists in people with normal or preserved renal function.
Method
Clinical trials were searched from Embase and PubMed from inception to 2013 December using the keywords “ALP”, “phosphate”, “CVD”, “mortality” and so on, and finally 24 trials with a total of 147634 patients were included in this study. Dose-response and semi-parametric meta-analyses were performed.
Results
A linear association of serum levels of ALP and phosphate with risks of coronary heart disease (CHD) events, CVD events and deaths was identified. The relative risk(RR)of ALP for CVD deaths was 1.02 (95% confidence interval [CI], 1.01–1.04). The RR of phosphate for CVD deaths and events was 1.05 (95% CI, 1.02–1.09) and 1.04 (95% CI: 1.03–1.06), respectively. A non-linear association of ALP and phosphate with total mortality was identified. Compared with the reference category of ALP and phosphate, the pooled RR of ALP for total mortality was 1.57 (95% CI, 1.27–1.95) for the high ALP group, while the RR of phosphate for total mortality was 1.33 (95% CI, 1.21–1.46) for the high phosphate group. It was observed in subgroup analysis that higher levels of serum ALP and phosphate seemed to indicate a higher mortality rate in diabetic patients and those having previous CVD. The higher total mortality rate was more obvious in the men and Asians with high ALP.
Conclusion
A non-linear relationship exists between serum levels of ALP and phosphate and risk of total mortality. There appears to be a positive association of serum levels of ALP/phosphate with total mortality in people with normal or preserved renal function, while the relationship between ALP and CVD is still ambiguous.
Citation: Li J-W, Xu C, Fan Y, Wang Y, Xiao Y-B (2014) Can Serum Levels of Alkaline Phosphatase and Phosphate Predict Cardiovascular Diseases and Total Mortality in Individuals with Preserved Renal Function? A Systemic Review and Meta-Analysis. PLoS ONE 9(7): e102276. https://doi.org/10.1371/journal.pone.0102276
Editor: Giuseppe Remuzzi, Mario Negri Institute for Pharmacological Research and Azienda Ospedaliera Ospedali Riuniti di Bergamo, Italy
Received: March 24, 2014; Accepted: June 16, 2014; Published: July 17, 2014
Copyright: © 2014 Li et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Data Availability: The authors confirm that all data underlying the findings are fully available without restriction. All relevant data are within the paper and its Supporting Information files.
Funding: The authors have no support or funding to report.
Competing interests: The authors have declared that no competing interests exist.
Introduction
Alkaline phosphatase (ALP) is an enzyme responsible for hydrolyzing phosphate esters and liberating inorganic phosphate. Serum ALP activity is elevated in hepatobiliary and bone diseases such as obstructive jaundice and bone cancer. Phosphorus exists in both organic and inorganic forms in our body. The phosphorus in inorganic form is often combined with calcium within the skeleton, while approximately 15% as a phosphate salt exists in the blood. Circulating ALP and phosphate concentrations often increase in end-stage renal disease, and are associated with higher cardiovascular and total mortality among hemodialysis patients [1], [2]. Previous meta-analyses show that elevated serum levels of phosphorus indicate a higher risk of mortality in population with chronic kidney disease (CKD) [3], [4], with the premise that the relation between the exposure and outcome being linear. However, some researchers have recently found that higher levels of serum ALP and phosphate indicate a higher risk of cardiovascular diseases (CVDs) and total mortality in people without CKD. To our knowledge, no meta-analysis has been conducted yet to study this issue in individuals with normal or preserved renal function, and the shape of the association remains uncertain. Therefore, we conduct a dose-response meta-analysis to quantify the association between serum levels of ALP and phosphate and the risk of CVDs and total mortality in these individuals.
Methods
Meta-analyses of observational studies were performed following the standard criteria [5].
Search strategy
PubMed and Embase were searched to identify all relevant articles from inception to 2013 December. Key words used for search included “alkaline phosphatase”, “phosphate”, “hypophosphatemia”, “hyperphosphatemia”, “cardiovascular diseases” and “mortality”. The search strategy was provided in a supplemental file (File S1). Furthermore, we manually searched references of retrieved studies for potentially relevant publications that were not identified in the database search.
Study selection
Studies were included in this meta-analysis if they meet the following criteria: the exposures were alkaline phosphatase or phosphate; the outcome was risks of CVD (including incidence of CHD events, CVD events and deaths) and total mortality. If one study did not consider stroke as CVD events but reported risks of stroke, then this study was also included in the analysis of CVD events.
Studies were excluded if they lacked HR or RR or numbers of cases and patients for each category (a continuous value per unit was acceptable); or the outcomes were about functional score and calcification. Additionally, reviews, animal studies, letters and conference abstracts without sufficient data were excluded. Whenever more than one article conducted the same study, we combined the results to get a more complete one and avoided the duplication of information.
Data extraction
The literature search and data extraction were undertaken independently by two authors (J. L. and C. X.) following a standard extraction process. Information on the following parameters was extracted and entered into a database: authors, study name, subjects, publication year, age, gender, country, follow-up years, study types (prospective/retrospective), adjusting variates, diabetes mellitus, patient types (CVD or not), number of cases and events and events and RR (or HR) for all-cause mortality and CVD in each category. The end points were CHD events, CVD deaths, CVD events and total mortality. We firstly manually excluded studies containing patients with only end-stage renal diseases or stages 3–5 CKD or on dialysis through title and abstract screening, but included studies of CKD after full-text reading if the results of participants with eGFR >60 were reported separately (only two were included). We did not exclude studies including CKD patients of general population, but most of these studies took CKD as a parameter for full adjustment. Thus we did our best to exclude those containing participants with eGFR<60. And we collected both partially and fully adjusted RR to compare the results, as it’s reported that adjustments, instead of age, may encumber the comparison between studies [6]. Newcastle-Ottawa quality assessment scale (NOS) [7] was used for quality assessment. When HR or RR was presented only as graph, we used the software g3data (version 1.51, www.frantz.fi/software/g3data.php) to extract the estimates. Any disagreements in the abstracted data were adjudicated by a third reviewer (Y. X).
Statistical analysis
The number of individuals encountering events, and the partially or fully adjusted HR or RR of ALP (10 UI/L) and phosphate (1 mg/dL) from each study were used. Here the HR was deemed equivalent to RR. Those articles reporting both CVD and total mortality were treated as separate reports. If the continuous RR per unit was unavailable to a study, we used the methods as previous described [in press, doi: 10.1007/s10072-014-1850-1]. In short, we first evaluated a potential linear association between ALP/phosphate levels and risk of outcomes, using restricted three knot cubic spine method (or four knots at percentiles 5%, 35%, 65%, and 95%, if results of the three knot ones show P<0.05). If linearity exists, we used the methods described by Nicola et al [8] to estimate RR of ALP (10 UI/L) and phosphate (1 mg/dL). This method assumes a linear relation between exposure and the logarithm of RR, and estimates the increase in log odds with per unit increase of exposure, taking appropriate account of the non-independence of the RR and transforming these RR into an overall RR as a continuous effect. We assigned the midpoint of the cut points of the class as the dose value, and for open-ended risk factor classes we used the methods described by Nienke et al [6] to get the dose value. The following methods were adopted to determine the event number in categories not reported in a study. If one study reported the total event number and gave RR in the three categories as 1, 1.2, 1.4, we assumed the event number of the third tertile is 1.4 times of that in the reference category, and then calculated the total event number in each category. If the number of events or participants was not given, we preferred to use continuous RR given directly in the study (also in condition which the unit step isn’t 10 UI/L of ALP and 1 mg/dL of phosphate). If P for linearity is less than 0.05, we first got the estimating values directly from the cubic spine estimation, and then used semi-parametric methods described by Frank [9]. In this method, the lowest and the highest groups of ALP or phosphate corresponded to the reference and highest categories, respectively. The middle group corresponded to either the second or the third (or the fourth, if exists) category in the original article, depending on the similarity of the ALP/phosphate in the third or fourth category to the middle group. We performed stratified analyses based on publication year, country, gender, age, diabetes, the NOS score, patient type, number of patients, follow-up duration and types of study in semi-parametric analysis. A similar stratified analysis was conducted in parametric analysis by testing the likelihood ratio and comparing the models derived from the interaction terms or not. We also assessed the publication bias using the Begg’s funnel plots, and the trim-and-fill method was applied to adjust the risk estimates when publication bias existed [10]. We chose a more significant results irrespective of the model used in the trim-and-fill method.
For all analyses, the statistical significance was set at a p value of 0.05, and CIs were calculated at the 95% level. If the results were homogeneous (I2<50%, P>0.05), fixed effects models (Inverse-variance method) were utilized, whereas if these results were heterogeneous (I2≥50%, P≤0.05), then random effect models (DerSimonian and Laird method) were used. Mantel-Haenszel method was not used because we could not get 2*2 cell counts or the mean and standard deviations but only RR and 95% CI from each study using GLST method. Analyses were performed using Stata software version 11 (StataCorp, College Station, Texas).
Results
Study characteristics
We identified 108 studies for further evaluation after reviewing titles, abstracts to exclude the duplicates or irrelevant studies, papers on CKD, case reports and reviews. Then 84 studies were excluded after full-text reading, because of which, 46 studies had no unqualified results, 1 had no full text, and 37 contained patients without exposure to serum ALP or phosphate. Among these excluded ones, Camille P. [11] and Ana Ludimila [12] studied the association between ALP/phosphate and vascular calcification; the research of Jinkwon and his colleagues [13] studied the functional outcomes of cerebrum with ALP and phosphate; Jason P. [14] focused on aortic valve calcification; the study of Gregory [15] was about phosphate and hypertension; and Bernard M.Y [16] investigated the association between alkaline phosphatase and peripheral arterial disease. Finally, a total of 24 trials [16]–[39] with 147634 individuals were identified from 2011 potentially relevant publications (Figure 1). Table 1 summarized the baseline characteristics of included trials. The individuals in studies of Sasiwarang [35] (2013), Edith [26] (2013), Matthew [36] (2010), Ravi D [19] (2007), Robert N. F [23] (2008), Bernard M.Y. C [16] (2009) Tobias E. L [27] (2010), K. M. Fleming [22] (2011), Per-Anton Westerberg [37] (2013), C. J. Bates [38] (2011), Bernard [40] (2009) and Julie R. D [20] (2013) were general population. The studies by Jun-Bean (2013) [29], Jan-P v K [34] (2010), Marcello (2009) [33], Michael [21] (2013), W.-S. [30] (2010) and Grandi [24] (2012) targeted at CVD patients. While CKD patients were the main subjects in Andrew [28] (2013) and Bryan K [41] (2005) studies. We chose patients with CKD stages 1–2 as study subjects in studies like these. The individuals in study of Jun-Bean [18] (2013) all had type II diabetes. Postmenopausal women with osteoporosis and male doctors were components in Yelena S [31] (2011) and Eric N [39] (2011) studies, respectively. The study by Marcello (2005) [33] contained the same subjects as what he chose in 2009 [33]. In his two studies, the former focused on phosphate while the latter was on ALP, which were both included in our analysis. Stephen J. [42] studied the the same objects as those in Robert N. [23] study, but we included the latter because it contained a more complete data. Vidya Raj [43] and Tonelli, M [32] studied the same study population, and we enrolled the latter because it consisted of more suitable subjects. Here we defined preserved renal function as eGFR>60 ml/min/1.73 m2. The adjusted information can be seen in Table 1.
Association between serum levels of ALP and phosphate and risk of CVD
As shown in Figure 2, the RR of ALP (10 UI/L)for CHD events was 1.04 (95% CI: 1.01–1.06, P = 0.436 for heterogeneity, I2 = 0.0%, partial adjustment); for CVD deaths and events was 1.02 (95% CI: 1.01–1.04, P = 0.172 for heterogeneity, I2 = 40%) and 1.05 (95% CI: 1.00–1.10, P = 0.126 for heterogeneity, I2 = 51.7%, partial adjustment), respectively; the RR of phosphate (1 mg/dL) for CHD events was 0.99 (95% CI: 0.96–1.01, P = 0.567 for heterogeneity, I2 = 0%); for CVD deaths and events was 1.05 (95% CI: 1.02–1.09, P = 0.349 for heterogeneity, I2 = 10.1%) and 1.04 (95% CI: 1.03–1.06, P = 0.120 for heterogeneity, I2 = 36.0%), respectively. For results of the linear relationship between ALP/Phosphate and CVD, please refer to Supplementary Material (File S2).
Association between serum levels of ALP and phosphate and risk of total mortality
We observed a non-linear association between serum levels of ALP/phosphate and risk of total mortality (P for non-linearity = 0.0002 and <0.0001, respectively) under full adjustment (Figure 3, 4). Similar trends were observed under partial adjustment (data not shown). The results reported here were under full adjustment. Compared with the reference ALP (51 UI/L), the RR directly from the cubic spline model for total mortality was 1.07 (95% CI, 1.05–1.09) for ALP = 79 UI/L, and 1.34 (95% CI, 1.26–1.42) for ALP = 124 UI/L, with substantial heterogeneity (Pr = 0.0001). Compared with the reference (Phosphate 2 mg/dL), the RR directly from the cubic spline model for total mortality was was 1.07 (95% CI, 1.04–1.09) for phosphate = 3.45 mg/dL, and 1.42 (95% CI, 1.29–1.56) for phosphate = 4.5 mg/dL, with little heterogeneity (Pr = 0.71). We also used semi-parametric methods to estimate the RR for ALP and phosphate. Compared with the reference category of ALP (<70 UI/L), the pooled RR for total mortality was 1.14 (95% CI, 1.09–1.20, P = 0.057, I2 = 48.9%) for the median ALP group (70–90 UI/L), and 1.57 (95% CI, 1.27–1.95, P<0.001, I2 = 90.1%) for the high ALP group (>90 UI/L). The RR of phosphate for total mortality was 1.08 (95% CI, 1.03–1.13, P = 0.38, I2 = 6.6%) for the median phosphate group (3–4 mg/dL), and 1.33 (95% CI, 1.21–1.46, P = 0.003, I2 = 57.5%) for the high phosphate group (>4 mg/dL) (Figure 4). We found an inflexion point at phosphate = 3.5 mg/dL.
Short-dashed line represents linear regression. Other line represents cubic spline regression.
Short-dashed line represents linear regression. Other line represents cubic spline regression.
Subgroup analyses
Subgroup analyses were conducted to determine the total mortality related to ALP and phosphate (Table 2, 3). The subgroup analyses were defined according to study type, study location, number of participants, duration of follow-up, publish publication year, gender, age, diabetes mellitus, the NOS score and patient types (CVD or not). The subgroup analysis of study type was unprocurable to ALP because only one study included was retrospective. A high level of ALP predicted a higher incidence of total mortality in the Asians than that in the Europeans and Americans (RR: 3.1, 95% CI: 2.2–4.3 for the Asians; RR: 1.3, 95% CI: 0.97–1.7 for the Europeans and RR: 1.4, 95% CI: 1.2–1.7 for the Americans; P<0.001); in male than that in female(RR: 2.0, 95% CI: 1.2–3.1 for male and RR: 1.4, 95% CI: 1.4–1.8 for female; P<0.001);in shorter follow-up than that in longer follow-up (RR: 2.0, 95% CI: 1.3–3.1 for duration<6 years and RR: 1.4, 95% CI: 1.1–1.8 for duration>6 years; P<0.001); in patients with serious diabetes than that in patients with mild diabetes (RR: 2.0, 95% CI: 1.2–3.1 for diabetes >11% and RR: 1.4, 95% CI: 1.1–1.5 for diabetes <11%; P<0.001); in CVD patients than that in non-CVD patients (RR: 2.4, 95% CI: 1.3–4.5 for CVD patients and RR: 1.4, 95% CI: 1.1–1.7 for non-CVD ones; P<0.001), in study with lower NOS score than that in study with higher NOS score (RR: 2.1, 95% CI: 1.4–3.0 for NOS<7 and RR: 1.3, 95% CI: 1.0–1.6 for NOS> = 7; P<0.001). A high level of phosphate also predicted a higher incidence of total mortality in patients with CVD (RR: 1.5, 95% CI: 1.3–1.8 for CVD patients and RR: 1.3, 95% CI: 1.2–1.4 for non-CVD ones; P<0.003), in patients with serious diabetes than that in patients with mild diabetes (RR: 1.5, 95% CI: 1.3–1.6 for diabetes >11% and RR: 1.2, 95% CI: 1.1–1.3 for diabetes <11%; P<0.001). Similar analyses were conducted using parametric methods. Publication year, gender, number of patients, duration of follow-up and diabetes mellitus had the same trends for ALP while others did not. No significant difference was found in any of the subgroups for phosphate.
Publication bias
Publication bias for associations between high levels of ALP/phosphate and total mortality was assessed by Begg’s tests (p = 0.007 and p = 0.010, respectively), which was confirmed using parametric methods (data not shown). The RR estimates altered (RR = 1.11, 95% CI: 1.07–1.17, 5 studies added for high ALP group and RR = 1.21, 95% CI: 1.09–1.33, 6 studies added for high phosphate group) after using the trim-and-fill method to adjust the potential publication bias.
Discussion
This is the first meta-analysis, to our knowledge, to assess the association of serum levels of ALP and phosphate with the risk of death and cardiovascular diseases in individuals without CKD 3∼5. Our research, which includes 24 clinical trials with 147634 patients, identifies a non-linear association between serum ALP/phosphate level and the risk of total mortality. Elevated serum levels of ALP and phosphate indicate a higher total mortality in people with normal or preserved renal function. The relationship between ALP and cardiovascular diseases is still ambiguous.
To date, the majority of studies about the relation of ALP and phosphate levels with total mortality have found a significant association between serum ALP level and total mortality in individuals without CKD3∼5. For example, Jun-Bean et al. [29] analyzed data from 1636 Korean coronary artery disease (CAD) patients after receiving percutaneous coronary intervention (PCI) with almost 2 years of follow-up, finding that the risk of total mortality increases in patients with a higher level of ALP. This study was carried out among Asians with CAD, and 37.3% of included subjects were diabetic patients. Our subgroup analysis shows that Asian ethnic origin, CVD and diabetes affect the prediction of ALP on total mortality, which may explain why this study obtains the most significant result. Yelena et al. [31] studied postmenopausal women with osteoporosis and found that phosphorus level was not associated with a higher risk of total mortality during 4 years of follow-up. Only a small group of subjects had diabetes in this study (<4%), and the subjects all suffered from osteoporosis. Osteoporosis alone is associated with fibroblast growth factor (FGF)-23, which may influence the level of phosphate [44]. These factors may explain why this study does not get significant result.
Our study shows that ALP is positively associated with CVD deaths irrespective of adjustment methods, while elevated ALP reflects a higher incidence of CHD and CVD events only after partial adjustment. The association between ALP and CVD deaths is weakened but still exists after full adjustment, which implies that higher levels of ALP play such roles other than adjustment factors, such as GFR, diabetes mellitus, CRP and hepatic dysfunction. On the other hand, the association between ALP and CHD and CVD events disappears after full adjustment. GFR, diabetes mellitus, hypertension and CRP are full adjustment factors, which may influence the effect of ALP on CHD and CVD events. We speculate that the true association between ALP and CHD/CVD events is likely to be modest; the difference in ALP measurements and covariate adjustment factors such as diabetes mellitus may result in changes in the strength and even the significance of the associations, thus leading to different conclusions. Higher serum levels of ALP have been associated with inflammation [35] and calcification [29]. ALP level is positively associated with CRP [40] and neutrophil to lymphocyte ratio (N/L) [45], and is elevated in vessels with medial calcification [46]. Thus inflammation and calcification could play a role in the association of ALP with CVD deaths.
We found that higher serum levels of phosphate increase the incidence of CVD events and deaths, irrespective of the adjusting method, but not that of CHD events. We compare our results with previous results regarding CKD [3], finding that although the predicting trend of CVD events is the same, the strength of association of phosphate with CVD events in non-CKD is weaker than that of CKD. Our previous work has found serum phosphate is not associated with risk of stroke, here we find serum phosphate is associated with CVD events, which imply that serum phosphate may be only associated with some kinds of CVD events such as coronary artery disease and heart failure but not stroke [in press, doi: 10.1007/s10072-014-1850-1]. Serum phosphate levels are significantly associated with age, female gender, diabetes mellitus, hypertension and hypercholesterolemia, but not with eGFR [47]. This is partially confirmed in another research, which shows that the relationship between serum phosphorus level and renal function is biphasic, with no association between phosphorus and renal function when eGFR is >60 and a linear relationship when eGFR is <60. [48] These previous results indicate that serum phosphate level may be an independent predictor of cardiovascular events and deaths in individuals without CKD 3∼5, which is confirmed in our meta-analysis. The serum level of phosphate is regulated by a balance between dietary intake, absorption from the gastrointestinal tract, storage in the skeleton, and urinary excretion [21]. High serum phosphate level is associated with more favorable profiles of cardiovascular risk factors [49] and calcification [50]. Indeed elevated phosphate level may directly stimulate smooth muscle cells to change phenotypically into osteoblast-like cells [20]. High serum level of phosphate can also cause degradation of the extracellular matrix, change of osteochondroge [51] and increase the production of reactive oxygen species [52] and stimulate an osteoblastic transcriptional program in the vasculature [53] in vascular smooth muscle. Serum phosphate level is a novel pre-clinical biomarker for cardiovascular calcification [54]. Thus calcification potentially establishes the association of ALP level with CVD deaths and events.
High serum levels of ALP and phosphate are associated with an increasing incidence of total mortality. We found a non-linear association of ALP and phosphate levels with total mortality, and this association exists even in the normal range (the medium group, Fig. 5), but the risk rises dramatically at a high level (Fig. 3, 4). The association may be partly caused by hypovitaminosis D, which is associated with elevated serum ALP and phosphate levels. Hypovitaminosis D is associated with greater plasma renin activity, inflammation, and higher blood pressure [21], and FGF-23 may play such roles in similar indirect ways, both of which are called phosphate-responsive hormones [55]. We found a turning point of serum phosphate level at about 3.5 mg/dL, beyond which the incidence of total mortality elevated enormously. People with a high serum phosphate level may need special care. We can not explain why some healthy people have higher serum levels of phosphate than the others, and genetic variants account for little of phosphate elevation [55], which needs to be explained by future study. Heterogeneity exists in both high ALP and phosphate group analyses, which we attribute to differences among studies in sample sizes, characteristics of the populations, measurement standards for ALP and phosphate, statistical adjustments and different grouping methods.
We found that ALP was not robust in subgroup analysis. Indeed serum ALP level is elevated in various conditions including liver and bone diseases, various kinds of cancers, chlorpropamide therapy, infectious mononucleosis, leukemoid response to infection, hormonal contraception, pregnancy, and hyperthyroidism. Different study populations may contribute to the diversity. Our research has found that elevated ALP level is more effective in predicting the risk of total mortality in Asian people and in male gender. It is reported that ALP activity is associated with male gender but not with age in adult group. [56] Both ALP and phosphate levels are more effective in predicting the risk of total mortality of people with diabetes and cardiovascular diseases. It is reported that 38% of diabetic patients have elevated serum alkaline phosphatase levels, mean fasting serum glucose level is positively associated with ALP, and bone ALP is the predominant type, which may be caused by diabetic bone diseases [57]. Indeed 9 out of 15 studies on the association of phosphate level with total mortality have shown an ascending incidence of diabetes in higher categories of phosphate (data now shown). Elevated ALP level is also observed in AMI patients [58]. We find that CVD and diabetes are the two main factors influencing heterogeneity of the association between phosphate level and total mortality, as heterogeneity in subgroup analysis by these two factors decreases apparently. The subgroup analysis of ALP by the NOS score shows different estimates between the groups. Patients with CVD and diabetes are more common in the NOS<7 group, which can partly explain the phenomenon.
Our findings raise the possibility that lowering serum ALP and phosphorus levels may decrease CVD or total mortality rates in the general population. Prior studies show that oral phosphorus binders fail to lower serum phosphorus concentration. A better choice to reduce it may be to limit the dietary phosphorus intake [59], which needs to be confirmed by future clinical trials.
It should be noted that severe publication bias exists in both analyses of ALP and phosphate. We have used the trim-and-fill method to adjust the risk estimates. We speculate that the publication bias may partly originate from heterogeneity among tests as we find that publication bias is mild in medium-level phosphate group (P for Begg’s test = 0.238), while it is serious in high-level phosphate group (P for Begg’s test = 0.008). The demarcation among reference, median and high categories differ dramatically among different studies, which may explain this phenomenon. Previous researchers have found that publication bias can be attributed to other factors such as heterogeneity. [60].
This meta-analysis has several limitations. Firstly, some studies do not provide the number of cases or participants in each category. Although we have used the continuous RR or estimated them as described in Methods, misestimation of RR is still inevitable. Secondly, mild to severe publication bias exists. We have tried various methods in subgroup analysis to explain why it exists; We carefully observed the funnel plots and suspected that small studies showing no beneficial effects were missing or heterogeneity among studies may partly explain the bias, since standard error of log RR may not reflect number of participants in each study. The trim-and-fill method is used to amend the results. Thirdly, factors of full adjustment vary among studies included. The author of an earlier meta-analysis of phosphate with the same research aim considered 5 adjusting covariates more suitable [3], which is similar to our partial adjustment group. And another research on methodology indicates that adjustment may hamper the estimation [6]. We find that adjustment can influence the results especially regarding the association between ALP and CVD. Both adjusted and un-adjusted results are presented if such phenomenon exists. Last, we does not conduct the analysis based on kidney function (subgroups: normal kidney function and CKD 1∼2), because our subjects mainly consist of community-based cohorts and some particular patients (e.g. population with stable CHD) without separately analyzing CKD patients. The number of studies on CKD 1∼2 is not enough to make a subgroup analysis. However, renal function serves as adjusting variants in most included studies, and has been adjusted in results of fully adjusted groups.
Supporting Information
File S2.
Results of the linear relationship between ALP/Phosphate and CVD.
https://doi.org/10.1371/journal.pone.0102276.s003
(DOCX)
Acknowledgments
We thank Prof. Hendriek C. Boshuizen (National Institute of Public Health and the Environment, Bilthoven, the Netherlands), Prof. Ya-Zhou Wu and Yan-qi Zhang (Department of Health Statistics, Third Military Medical University, Chongqing, PR China) for their kindly help in statistical methods.
Author Contributions
Conceived and designed the experiments: YBX. Performed the experiments: JWL CX. Analyzed the data: YF YW. Contributed reagents/materials/analysis tools: YF YW. Contributed to the writing of the manuscript: JWL.
References
- 1. Tanhauserova V, Kuricova K, Pacal L, Bartakova V, Rehorova J, et al. (2014) Genetic variability in enzymes of metabolic pathways conferring protection against non-enzymatic glycation versus diabetes-related morbidity and mortality. Clinical Chemistry and Laboratory Medicine 52: 77–83.
- 2. Regidor DL, Kovesdy CP, Mehrotra R, Rambod M, Jing J, et al. (2008) Serum alkaline phosphatase predicts mortality among maintenance hemodialysis patients. J Am Soc Nephrol 19: 2193–2203.
- 3. Palmer SC, Hayen A, Macaskill P, Pellegrini F, Craig JC, et al. (2011) Serum levels of phosphorus, parathyroid hormone, and calcium and risks of death and cardiovascular disease in individuals with chronic kidney disease: a systematic review and meta-analysis. JAMA 305: 1119–1127.
- 4. Natoli JL, Boer R, Nathanson BH, Miller RM, Chiroli S, et al. (2013) Is there an association between elevated or low serum levels of phosphorus, parathyroid hormone, and calcium and mortality in patients with end stage renal disease? A meta-analysis. BMC Nephrol 14: 88.
- 5. Stroup DF, Berlin JA, Morton SC, Olkin I, Williamson GD, et al. (2000) Meta-analysis of observational studies in epidemiology: a proposal for reporting. Meta-analysis Of Observational Studies in Epidemiology (MOOSE) group. JAMA 283: 2008–2012.
- 6. Hartemink N, Boshuizen HC, Nagelkerke NJ, Jacobs MA, van Houwelingen HC (2006) Combining risk estimates from observational studies with different exposure cutpoints: a meta-analysis on body mass index and diabetes type 2. Am J Epidemiol 163: 1042–1052.
- 7.
Wells G, Shea B, O’connell D, Peterson J, Welch V, et al.. (2000) The Newcastle-Ottawa Scale (NOS) for assessing the quality of nonrandomised studies in meta-analyses.
- 8. Orsini N, Bellocco R, Greenland S (2006) Generalized least squares for trend estimation of summarized dose-response data. Stata Journal 6: 40.
- 9.
Ding M, Bhupathiraju SN, Satija A, van Dam RM, Hu FB (2013) Long-Term Coffee Consumption and Risk of Cardiovascular Disease: A Systematic Review and a Dose-Response Meta-Analysis of Prospective Cohort Studies. Circulation.
- 10. Duval S, Tweedie R (2000) A nonparametric “trim and fill” method of accounting for publication bias in meta-analysis. Journal of the American Statistical Association 95: 89–98.
- 11. Figueiredo CP, Rajamannan NM, Lopes JB, Caparbo VF, Takayama L, et al. (2013) Serum phosphate and hip bone mineral density as additional factors for high vascular calcification scores in a community-dwelling: the Sao Paulo Ageing & Health Study (SPAH). Bone 52: 354–359.
- 12. Cancela AL, Santos RD, Titan SM, Goldenstein PT, Rochitte CE, et al. (2012) Phosphorus is associated with coronary artery disease in patients with preserved renal function. PLoS One 7: e36883.
- 13.
Kim J, Song TJ, Song D, Lee HS, Nam CM, et al.. (2013) Serum Alkaline Phosphatase and Phosphate in Cerebral Atherosclerosis and Functional Outcomes After Cerebral Infarction. Stroke.
- 14. Linefsky JP, O’Brien KD, Katz R, de Boer IH, Barasch E, et al. (2011) Association of serum phosphate levels with aortic valve sclerosis and annular calcification: the cardiovascular health study. J Am Coll Cardiol 58: 291–297.
- 15. Vyssoulis G, Karpanou E, Tzamou V, Kyvelou SM, Michaelidis A, et al. (2010) Serum phosphate in white-coat hypertensive patients: focus on dipping status and metabolic syndrome. Hypertens Res 33: 825–830.
- 16. Cheung BM, Ong KL, Wong LY (2009) Elevated serum alkaline phosphatase and peripheral arterial disease in the United States National Health and Nutrition Examination Survey 1999–2004. Int J Cardiol 135: 156–161.
- 17. Abramowitz M, Muntner P, Coco M, Southern W, Lotwin I, et al. (2010) Serum alkaline phosphatase and phosphate and risk of mortality and hospitalization. Clin J Am Soc Nephrol 5: 1064–1071.
- 18. Chonchol M, Dale R, Schrier RW, Estacio R (2009) Serum Phosphorus and Cardiovascular Mortality in Type 2 Diabetes. American Journal of Medicine 122: 380–386.
- 19. Dhingra R, Sullivan LM, Fox CS, Wang TJ, D’Agostino RB Sr, et al. (2007) Relations of serum phosphorus and calcium levels to the incidence of cardiovascular disease in the community. Archives of Internal Medicine 167: 879–885.
- 20. Dominguez JR, Kestenbaum B, Chonchol M, Block G, Laughlin GA, et al. (2013) Relationships between serum and urine phosphorus with all-cause and cardiovascular mortality: The osteoporotic fractures in men (MrOS) study. American Journal of Kidney Diseases 61: 555–563.
- 21. Ess M, Heitmair-Wietzorrek K, Frick M, Umlauf N, Ulmer H, et al. (2013) Serum phosphate and long-term outcome among patients with stable heart failure. J Card Fail 19: 25–30.
- 22. Fleming KM, West J, Aithal GP, Fletcher AE (2011) Abnormal liver tests in people aged 75 and above: prevalence and association with mortality. Aliment Pharmacol Ther 34: 324–334.
- 23. Foley RN, Collins AJ, Ishani A, Kalra PA (2008) Calcium-phosphate levels and cardiovascular disease in community-dwelling adults: the Atherosclerosis Risk in Communities (ARIC) Study. Am Heart J 156: 556–563.
- 24. Grandi NC, Brenner H, Hahmann H, Wusten B, Marz W, et al. (2012) Calcium, phosphate and the risk of cardiovascular events and all-cause mortality in a population with stable coronary heart disease. Heart 98: 926–933.
- 25. Kestenbaum B, Sampson JN, Rudser KD, Patterson DJ, Seliger SL, et al. (2005) Serum phosphate levels and mortality risk among people with chronic kidney disease. J Am Soc Nephrol 16: 520–528.
- 26.
Koehler EM, Sanna D, Hansen BE, van Rooij FJ, Heeringa J, et al.. (2013) Serum liver enzymes are associated with all-cause mortality in an elderly population. Liver Int.
- 27. Larsson TE, Olauson H, Hagstrom E, Ingelsson E, Arnlov J, et al. (2010) Conjoint effects of serum calcium and phosphate on risk of total, cardiovascular, and noncardiovascular mortality in the community. Arteriosclerosis, Thrombosis, and Vascular Biology 30: 333–339.
- 28. McGovern AP, de Lusignan S, van Vlymen J, Liyanage H, Tomson CR, et al. (2013) Serum Phosphate as a Risk Factor for Cardiovascular Events in People with and without Chronic Kidney Disease: A Large Community Based Cohort Study. PLoS One 8: e74996.
- 29. Park JB, Kang DY, Yang HM, Cho HJ, Park KW, et al. (2013) Serum alkaline phosphatase is a predictor of mortality, myocardial infarction, or stent thrombosis after implantation of coronary drug-eluting stent. Eur Heart J 34: 920–931.
- 30. Ryu WS, Lee SH, Kim CK, Kim BJ, Yoon BW (2010) Increased serum alkaline phosphatase as a predictor of long-term mortality after stroke. Neurology 75: 1995–2002.
- 31. Slinin Y, Blackwell T, Ishani A, Cummings SR, Ensrud KE (2011) Serum calcium, phosphorus and cardiovascular events in post-menopausal women. International Journal of Cardiology 149: 335–340.
- 32. Tonelli M, Curhan G, Pfeffer M, Sacks F, Thadhani R, et al. (2009) Relation between alkaline phosphatase, serum phosphate, and all-cause or cardiovascular mortality. Circulation 120: 1784–1792.
- 33. Tonelli M, Sacks F, Pfeffer M, Gao Z, Curhan G (2005) Relation between serum phosphate level and cardiovascular event rate in people with coronary disease. Circulation 112: 2627–2633.
- 34. van Kuijk JP, Flu WJ, Chonchol M, Valentijn TM, Verhagen HJM, et al. (2010) Elevated Preoperative Phosphorus Levels Are an Independent Risk Factor for Cardiovascular Mortality. American Journal of Nephrology 32: 163–168.
- 35. Wannamethee SG, Sattar N, Papcosta O, Lennon L, Whincup PH (2013) Alkaline phosphatase, serum phosphate, and incident cardiovascular disease and total mortality in older men. Arterioscler Thromb Vasc Biol 33: 1070–1076.
- 36. Webber M, Krishnan A, Thomas NG, Cheung BM (2010) Association between serum alkaline phosphatase and C-reactive protein in the United States National Health and Nutrition Examination Survey 2005–2006. Clin Chem Lab Med 48: 167–173.
- 37.
Westerberg PA, Tivesten A, Karlsson MK, Mellstrom D, Orwoll E, et al.. (2013) Fibroblast growth factor 23, mineral metabolism and mortality among elderly men (Swedish MrOs). BMC Nephrology 14.
- 38. Bates CJ, Hamer M, Mishra GD (2012) A study of relationships between bone-related vitamins and minerals, related risk markers, and subsequent mortality in older British people: the National Diet and Nutrition Survey of People Aged 65 Years and Over. Osteoporos Int 23: 457–466.
- 39. Taylor EN, Rimm EB, Stampfer MJ, Curhan GC (2011) Plasma fibroblast growth factor 23, parathyroid hormone, phosphorus, and risk of coronary heart disease. Am Heart J 161: 956–962.
- 40. Cheung BM, Ong KL, Cheung RV, Wong LY, Wat NM, et al. (2008) Association between plasma alkaline phosphatase and C-reactive protein in Hong Kong Chinese. Clin Chem Lab Med 46: 523–527.
- 41. Yousaf F, Charytan C (2014) Review of cinacalcet hydrochloride in the management of secondary hyperparathyroidism. Renal Failure 36: 131–138.
- 42. Onufrak SJ, Bellasi A, Cardarelli F, Vaccarino V, Muntner P, et al. (2009) Investigation of gender heterogeneity in the associations of serum phosphorus with incident coronary artery disease and all-cause mortality. American Journal of Epidemiology 169: 67–77.
- 43.
Krishnamurthy VR, Baird BC, Wei G, Greene T, Raphael K, et al.. (2011) Associations of serum alkaline phosphatase with metabolic syndrome and mortality. Am J Med 124: 566 e561–567.
- 44. Celik E, Guzel S, Abali R, Guzelant AY, Celik Guzel E, et al. (2013) The relationship between fibroblast growth factor 23 and osteoporosis in postmenopausal women. Minerva Med 104: 497–504.
- 45.
Liu X, Guo Q, Feng X, Wang J, Wu J, et al.. (2014) Alkaline Phosphatase and Mortality in Patients on Peritoneal Dialysis. Clin J Am Soc Nephrol.
- 46. Shanahan CM, Cary NR, Salisbury JR, Proudfoot D, Weissberg PL, et al. (1999) Medial localization of mineralization-regulating proteins in association with Monckeberg’s sclerosis: evidence for smooth muscle cell-mediated vascular calcification. Circulation 100: 2168–2176.
- 47. Onufrak SJ, Bellasi A, Shaw LJ, Herzog CA, Cardarelli F, et al. (2008) Phosphorus levels are associated with subclinical atherosclerosis in the general population. Atherosclerosis 199: 424–431.
- 48.
Aronson D, Kapeliovich M, Hammerman H, Dragu R (2013) The Relation between Serum Phosphorus Levels and Clinical Outcomes after Acute Myocardial Infarction. PLoS ONE 8.
- 49. Lippi G, Montagnana M, Salvagno GL, Targher G, Guidi GC (2009) Relationship between serum phosphate and cardiovascular risk factors in a large cohort of adult outpatients. Diabetes Res Clin Pract 84: e3–5.
- 50. Shin S, Kim KJ, Chang HJ, Cho I, Kim YJ, et al. (2012) Impact of serum calcium and phosphate on coronary atherosclerosis detected by cardiac computed tomography. Eur Heart J 33: 2873–2881.
- 51. Lau WL, Pai A, Moe SM, Giachelli CM (2011) Direct effects of phosphate on vascular cell function. Adv Chronic Kidney Dis 18: 105–112.
- 52. Chue CD, Townend JN, Steeds RP, Ferro CJ (2010) Arterial stiffness in chronic kidney disease: causes and consequences. Heart 96: 817–823.
- 53. Mathew S, Tustison KS, Sugatani T, Chaudhary LR, Rifas L, et al. (2008) The mechanism of phosphorus as a cardiovascular risk factor in CKD. J Am Soc Nephrol 19: 1092–1105.
- 54. Rajamannan NM (2011) Serum phosphate concentrations: a novel pre-clinical biomarker for cardiovascular calcification. J Am Coll Cardiol 58: 298–299.
- 55. Ellam TJ, Chico TJ (2012) Phosphate: the new cholesterol? The role of the phosphate axis in non-uremic vascular disease. Atherosclerosis 220: 310–318.
- 56. Eastman JR, Bixler D (1977) Serum alkaline phosphatase: normal values by sex and age. Clin Chem 23: 1769–1770.
- 57. Maxwell DB, Fisher EA, Ross-Clunis HA 3rd, Estep HL (1986) Serum alkaline phosphatase in diabetes mellitus. J Am Coll Nutr 5: 55–59.
- 58. Iqbal MP, Mehboobali N, Azam I, Tareen AK (2013) Association of alkaline phosphatase with acute myocardial infarction in a population with high prevalence of hypovitaminosis D. Clin Chim Acta. 425: 192–195.
- 59. Cupisti A, Gallieni M, Rizzo MA, Caria S, Meola M, et al. (2013) Phosphate control in dialysis. Int J Nephrol Renovasc Dis 6: 193–205.
- 60. Sterne JA, Egger M, Smith GD (2001) Systematic reviews in health care: investigating and dealing with publication and other biases in meta-analysis. BMJ: British Medical Journal 323: 101.