BACKGROUND: Fetuin-A is a hepatic secretory protein that inhibits arterial calcification in vitro. The association of fetuin-A with coronary arterial calcification (CAC) in the general population is uncertain.
METHODS: Among 2457 community-living individuals without cardiovascular disease (CVD), we measured serum fetuin-A concentrations by ELISA and evaluated the cross-sectional association of fetuin-A with CAC prevalence (any vs none) and severity; on follow-up 3.2 years (median) later, we evaluated the association of fetuin-A with CAC incidence and progression.
RESULTS: The mean age was 62 (SD 10) years, and the mean fetuin-A concentration was 0.48 (0.10) g/L. At baseline, 1200 individuals (49%) had CAC, and 272 individuals developed CAC during follow-up. At baseline, there was a threshold effect at the lowest fetuin-A quartile with CAC prevalence. In models adjusted for demographics, traditional cardiovascular disease (CVD) risk factors and kidney function, the lowest fetuin-A quartile had 7% (95% CI 1%–13%; P = 0.04) greater CAC prevalence compared with quartiles 2–4. Similar associations were observed with CAC severity at baseline, but the association was more linear. Each SD (0.10 g/L) lower fetuin-A was associated with a 12% (95% CI 3%–21%; P = 0.01) greater CAC severity in adjusted models. There was no significant association of fetuin-A with CAC incidence or progression.
CONCLUSIONS: Fetuin-A is inversely associated with CAC severity among community-living individuals without CVD. Whether fetuin-A concentrations are associated with incident CVD event in the general population requires future study.
Fetuin-A is a hepatic secretory protein that inhibits arterial calcification in vitro (1). In serum, it interacts with calcium and phosphorus, increasing their solubility and inhibiting precipitation (1, 2). Consistent with this function, fetuin-A knockout mice have soft-tissue calcification when challenged with diets enriched in vitamin D or phosphorus (3, 4). Because fetuin-A prevents arterial calcification, is found in relatively high concentrations in human blood (5), and is believed to exert its inhibitory effects on arterial calcification within the bloodstream, blood fetuin-A concentrations represent an attractive candidate marker of arterial calcification.
End-stage renal disease (ESRD)14 is characterized by extensive arterial calcification. In patients with ESRD, most (6–9) but not all (10, 11) studies have found that low fetuin-A is associated with greater severity of coronary artery calcification (CAC) or abdominal aortic calcification. Whether fetuin-A plays a role in calcification in the absence of advanced kidney disease is less clear. We previously reported that low serum fetuin-A concentrations were associated with cardiac valve calcification in a population of 970 individuals with prevalent cardiovascular disease (CVD), most with normal kidney function (12). In another study, among 382 older whites, low fetuin-A concentrations were associated with greater CAC measured on computed tomography (CT) approximately 5 years after fetuin-A measurement (13). Two small studies (n = 64 each) evaluated patients referred for angiography and reported no association between fetuin-A and CAC severity (14, 15).
Relationships of fetuin-A with vascular disease are made more complex by its additional actions on insulin resistance. Fetuin-A inhibits the insulin receptor and induces insulin resistance in vitro and in animal models (16–20). Whereas low fetuin-A is associated with arterial calcification and CVD events in ESRD (6–9), high fetuin-A has been associated with insulin resistance (21–23) and greater risk for diabetes (24, 25). Thus, fetuin-A may exert simultaneous competing effects on CVD, and these effects may differ by diabetes status.
We evaluated the association of serum fetuin-A concentrations with the prevalence, severity, and incidence of CAC in a large community-living population without clinical CVD. As a secondary aim, we evaluated the heterogeneity in these associations by sex, race, and diabetes status.
Between 2000 and 2002, the Multi-Ethnic Study of Atherosclerosis (MESA) recruited 6814 individuals, aged 45–84 years, who self-identified as Caucasian (white), African-American, Hispanic, or Chinese and were free of clinical CVD. Participants were recruited from 6 US communities: Baltimore, MD; Chicago, IL; Forsythe County, NC; Los Angeles, CA; North Manhattan and the Bronx, NY; and St. Paul, MN. Exclusion criteria included physician-diagnosed CVD or heart failure, pregnancy, active treatment for cancer, residency in a nursing home, cognitive impairment, weight >300 pounds, or having received a CT scan in the preceding year (26). The institutional review boards of all participating centers, the coordinating center, and University of California San Diego approved the study, and all participants provided written informed consent.
All participants underwent chest CT scans for assessment of CAC at their baseline examination and were invited to return for follow-up CT scans. Half had their follow-up CT scans performed at the second study visit and the remainder at the third visit (approximately 3 years after baseline). Timing of the follow-up CT was assigned at the time of the baseline visit by computer random number generator. To maximize statistical power to detect incident CAC, this study enrolled all 2774 individuals who had both the baseline and third-visit CT scan. From these, we excluded 224 individuals with insufficient serum for fetuin-A measurement and 93 participants with missing covariate data, resulting in a final analytic sample of 2457 participants for this analysis.
Venous blood samples were collected at the baseline MESA examination and serum was frozen at −70 °C. In 2009, specimens were thawed and we measured fetuin-A at the Clinical Chemistry Laboratory at the University of Maryland with a human ELISA kit (Epitope Diagnostics). The assay uses a 2-site sandwich technique with polyclonal antibodies that bind different epitopes of human fetuin-A. We measured serum samples twice in each participant and averaged the results. Specimens were run in random order without reference to CAC prevalence. As an internal control, we ran pooled serum samples on each ELISA plate and noted an abrupt drift in mean concentrations after approximately half of the measurements had been completed. Thus, fetuin-A measurements performed after that point were recalibrated to mean of pooled serum measurements before the occurrence of the drift. The lower limit of detection of the assay is 0.05 mg/L. We used duplicate measurements in all participants to assess the intraassay CV across the whole spectrum of observed fetuin-A concentrations (range 0.20–0.94 g/L); the mean intraassay CV was 3.0%. We used pooled serum specimens to assess the mean interassay CVs across the 85 ELISA plates, which showed a mean interassay CV of 5.3% and 4.8% at fetuin-A concentrations of 0.29 and 0.58 g/L, respectively.
CORONARY ARTERY CALCIFICATION
We measured CAC with either an Imatron C-150XL electron beam CT scanner (GE-Imatron) or a multidetector CT scanner as described (27). CT scans were done twice at each visit, and identical protocols were used at the baseline and third visit. An expert committee developed a scanning protocol to standardize scan acquisition across the 2 slightly different technologies, and data quality was equivalent between CT scan techniques (28). Images were read centrally at the Harbor-UCLA Research and Education Institute (Torrance, CA) (27).
Prevalent diabetes mellitus was defined as fasting blood glucose ≥126 mg/dL (7.0 mmol/L) or use of hypoglycemic medications or insulin. We measured resting blood pressure 3 times with participants in the seated position by use of a Dinamap model Pro 100 sphygmomanometer (Critickon, General Electric), using the mean of the last 2 measurements. The medications used by study participants were recorded by study personnel. Smoking status was categorized as current, former, or never. We measured height and weight with participants wearing light clothing and no shoes and calculated body mass index (in kg/m2). Concentrations of total cholesterol, HDL cholesterol, and triglycerides were measured with fasting (8-h) morning samples. LDL cholesterol was estimated by use of the Friedewald equation (29). We measured interleukin-6 with an ultrasensitive ELISA from R&D Systems (30) and cystatin C concentrations with a BNII nephelometer (Siemens Healthcare Diagnostics); intraassay CVs were ≤3%. We used these measurements to calculate estimated glomerular filtration rate (eGFR) with the formula eGFR= 76.7 * cystatin C−1.19 (31). We collected spot morning urine specimens, measured urine albumin and creatinine with nephelometry and the rate Jaffe reaction, respectively, and calculated urine albumin-to-creatinine ratios (mg /g) (32). We used fasting morning blood samples to measure calcium and phosphorus with a Siemens Dimensions Vista 500 system. The limits of detection were 5 mg/dL (1.25 mmol/L) for serum calcium and 1.0 mg/dL (0.323 mmol/L) for serum phosphorus. CVs were <1% for serum calcium at 9.0 mg/dL (2.25 mmol/L) and serum phosphorus at 4.0 mg/dL (1.29 mmol/L).
We categorized participants into quartiles by the distribution of fetuin-A and compared differences in baseline characteristics across quartiles. P values are for linear trend derived from linear regression models. Prevalent CAC was defined as a score >0 at baseline. We modeled the association of fetuin-A with prevalent CAC with relative risk regression, by use of a generalized linear model with log link, gaussian error, and robust standard errors (33, 34). Given cross-sectional design, results are interpreted as prevalence ratios. We developed sequential models: model 1 was adjusted for age, sex, and race/ethnicity; model 2 added body mass index, diabetes, systolic blood pressure, blood pressure medication use, smoking (never, former, current), total cholesterol, HDL cholesterol, natural log of serum triglycerides, lipid medication use, and natural log of interleukin-6; and model 3 added eGFR, urine albumin-to-creatinine ratio, and serum calcium and phosphorus concentrations. Multiplicative interaction terms evaluated in model 3 tested for effect modification by sex, race, and diabetes status. We evaluated statistical significance of the interaction term by the likelihood ratio test comparing nested models with or without inclusion of the interaction term.
Among participants with detectable CAC at baseline, we assessed the relationship of fetuin-A with severity of CAC (natural log of CAC score) by use of multiple linear regression. The β coefficient from linear regression was exponentiated, subtracted from 1, and multiplied by 100 to allow interpretation as the percentage difference in CAC severity per SD change in fetuin-A.
Incident CAC was defined as any CAC on follow-up CT among participants with CAC scores of 0 at the baseline visit. We used Poisson regression with a log-link, robust variance estimation, and an offset for time between CAC measurements. Among individuals with detectable CAC at baseline, we evaluated the relationship of fetuin-A with progression of CAC from the baseline CT to the follow-up CT with multiple linear regression. We evaluated the natural log of change in CAC score, and the β coefficient was exponentiated, subtracted from 1, and multiplied by 100 to allow interpretation as the percentage change in CAC between examinations.
We performed statistical analyses using S-Plus 8.0 (Insightful Corp.) and Stata version 10.1 (Stata Corp.). P values <0.05 were considered statistically significant for all analyses including interaction terms.
Among the 2457 study participants, the mean age was 62 (10) years and 52% were female. Forty percent were white, 27% were African-American, 20% were Hispanic, and 13% were Chinese. Twelve percent (n = 288) had diabetes, and the mean eGFR was 94 (22) mL · min−1 · (1.73 m2)−1. The mean fetuin-A concentration was 0.48 (0.10) g/L. Twelve hundred participants (49%) had prevalent CAC at the baseline examination.
Participants with higher fetuin-A concentrations were younger, more frequently male and white, and less frequently African-American (Table 1) (see Supplemental Table 1, which accompanies the online version of this article at http://www.clinchem.org/content/vol58/issue5). Individuals with higher fetuin-A concentrations also had greater body mass index; lower prevalence of diabetes but higher fasting insulin; lower systolic blood pressure and less use of blood pressure medications; higher total cholesterol, LDL cholesterol, and triglycerides; and lower interleukin-6, higher calcium, and lower phosphorus concentrations.
We observed a monotonic inverse association of fetuin-A with CAC prevalence in unadjusted analysis (Fig. 1). Table 2 shows the association of fetuin-A with CAC prevalence in a series of adjusted models. In unadjusted models, compared with the lowest quartile, participants in the highest quartile had 15% lower prevalence of CAC, and each SD (0.10 g/L) higher fetuin-A was associated with a 5% lower prevalence of CAC. This association was attenuated when adjusted for age, sex, and race; age accounted for most of the attenuation. With adjustment for traditional CVD risk factors (model 2) and kidney function, calcium, and phosphorus (model 3), a threshold effect was observed wherein the lowest fetuin-A quartile had the highest prevalence of CAC but the prevalence in quartiles 2 through 4 were all about 7% lower compared with the lowest quartile, although no quartile in isolation was significantly different from the lowest. When quartiles 2 through 4 were collapsed and compared to quartile 1 in the final model, individuals in quartiles 2 through 4 had 7% (95% CI 1%–13%; P = 0.043) lower prevalence of CAC compared with quartile 1.
At the baseline examination, 1200 participants had CAC scores >0. Among these individuals, higher fetuin-A concentration was associated with progressively less severe CAC (Fig. 1 and Table 3). Comparing extreme quartiles, participants in the highest quartile had 34% less severe CAC than the lowest quartile, and each SD of higher fetuin-A concentration was associated with 17% lesser CAC severity. This association was attenuated with adjustment for age, sex, and race. In models adjusted for traditional CVD risk factors (model 2) and kidney function, calcium, and phosphorus (model 3), a monotonic inverse association of fetuin-A with CAC severity emerged again, such that in the final model, participants in the highest quartile had 26% lower CAC severity than the lowest quartile.
Among the 1200 participants without CAC at the baseline examination, 272 individuals developed CAC on a follow-up CT scan performed a median 3.2 years (range 2.2–4.9 years) after baseline. Although the point estimates suggested associations in similar directions to the cross-sectional analyses described above, the association of fetuin-A with incident CAC was not statistically significant in any of the models (Table 4). Among individuals with detectable CAC at baseline, 92% had greater CAC scores at the time of the follow-up CT scan. Individuals with lower fetuin-A concentrations at baseline were less likely to experience CAC progression in unadjusted analyses. However, although point estimates remained in the same direction, this association was also attenuated with adjustment for age, sex, and race and was not statistically significant through the remaining sequence of adjusted models (Table 5). Last, because point estimates suggested a possible association of fetuin-A with CAC incidence and progression in Tables 4 and 5, we evaluated a combined outcome [defined as Ln(CAC+1) at the follow-up CT − Ln(CAC+1)at the baseline CT] to maximize statistical power. In the fully adjusted model, we did not observe a statistically significant association of fetuin-A with this outcome (P = 0.37).
We determined the association of fetuin-A [per SD (0.1g/L) greater] with % CAC severity among all individuals, and within strata defined by sex, race/ethnicity, and diabetes status (see online Supplemental Fig.). Although point estimates varied somewhat, none of the tests for interaction were statistically significant (all P interactions >0.20), nor were any interactions present for outcomes of CAC prevalence, incidence, or progression (all P interactions >0.05).
Fetuin-A concentrations were inversely associated with CAC severity in a large multiethnic community-living population without clinically apparent CVD, independent of traditional CVD risk factors and kidney function. Results were similar irrespective of sex, race/ethnicity, or diabetes status. Conversely, fetuin-A concentrations were not significantly associated with CAC incidence or progression over 3 years.
Prior studies have evaluated fetuin-A among ESRD patients and in select populations with prevalent clinical CVD. Findings have been conflicting, showing inverse, absent, and positive directions of associations (12, 13, 35–38), which may reflect differences in enrollment criteria and small sample sizes. To our knowledge, only 1 prior study has examined the association of fetuin-A concentrations with CAC in community-living individuals without advanced chronic kidney disease (CKD) or known clinical CVD. In a pilot study among older individuals in the Rancho Bernardo Study, we observed that fetuin-A concentrations were inversely associated with CAC severity (13). These findings were considered preliminary, as only 28% of participants with available fetuin-A measurements had CT scans for CAC and the fetuin-A and CAC measurements were not concurrent (median time between measurement was 4.6 years). Moreover, compared to the study participants evaluated here, the study sample was considerably older and exclusively white. These results are confirmed in a larger sample here, and extended to a younger, geographically and racially diverse cohort. The substantially larger sample size also allowed us to test heterogeneity by sex, race, and diabetes status: an important contribution, as prior studies have suggested that relationships may differ across these subgroups (12, 39).
Whereas lower fetuin-A concentrations were independently associated with CAC severity, fetuin-A was only associated with CAC prevalence when we compared the lowest quartile with quartiles 2 through 4, and we did not observe statistically significant associations of fetuin-A with CAC incidence or progression over 3 years. Our study may have been underpowered to detect a small or moderate association of fetuin-A with these outcomes. Although the difference was not statistically significant, individuals with higher fetuin-A had lower point estimates for each. Future studies with larger sample sizes are required to determine whether the lack of statistical significance for these outcomes was due to insufficient statistical power or absent biological effects.
A priority for future research will be to definitively establish whether fetuin-A is associated with incident CVD events in community-living individuals. Unfortunately, at present, existing data regarding this question are also conflicting. In the largest study to date, Weikert et al. (40) evaluated 2593 middle-aged Europeans and reported that higher fetuin-A concentrations were associated with incident myocardial infarction and stroke. The direction of association reported in that study is surprising in light of our findings, as we report that individuals with higher fetuin-A concentrations had less severe CAC. Because higher fetuin-A concentrations may also induce peripheral insulin resistance, (41) it is possible that the mechanisms linking high fetuin-A with CVD events in the study by Weikert were through pathways distinct from arterial calcium deposition. In contrast, 2 other smaller studies have reported that lower fetuin-A concentrations measured in individuals with myocardial infarction were associated with greater mortality risk (42, 43). Thus, future studies are required to determine the direction of association of fetuin-A with the incident development of CVD events in community-living individuals and to evaluate the extent to which factors such as arterial calcification and insulin resistance may mediate or modify the association.
Our study has important limitations. Fetuin-A was measured at 1 point in time. To our knowledge, the test–retest correlation of fetuin-A within individuals is unknown. Moreover, whether intraindividual changes in fetuin-A concentrations over time may provide additional information about risk of CAC prevalence, severity, incidence, and progression is an important question for future research. Participants were of 4 race/ethnicities and middle and older age, and prevalent clinical CVD was an exclusion criteria. Few participants had moderate or severe CKD. Our results may not generalize to other populations.
In conclusion, serum fetuin-A concentrations are inversely associated with CAC severity independent of traditional CVD risk factors and kidney function in community-living individuals free of clinically apparent CVD. Future studies are required to determine the association of fetuin-A with incident CVD events and, if the association is observed, to determine whether diabetes or the presence of arterial calcification may mediate or modify the association.
The authors thank the other investigators, the staff, and the participants of the MESA study for their valuable contributions. A full list of participating MESA investigators and institutions can be found at http://www.mesa-nhlbi.org. This material is the result of work supported with resources of the VA San Diego Healthcare System. We thank Clydene Nee for administrative review and assistance with this manuscript.
↵14 Nonstandard abbreviations:
- end-state renal disease;
- coronary artery calcification;
- cardiovascular disease;
- computed tomography;
- Multi-Ethnic Study of Atherosclerosis;
- estimated glomerular filtration rate;
- chronic kidney disease.
Author Contributions: All authors confirmed they have contributed to the intellectual content of this paper and have met the following 3 requirements: (a) significant contributions to the conception and design, acquisition of data, or analysis and interpretation of data; (b) drafting or revising the article for intellectual content; and (c) final approval of the published article.
Authors' Disclosures or Potential Conflicts of Interest: Upon manuscript submission, all authors completed the Disclosures of Potential Conflict of Interest form. Potential conflicts of interest:
Employment or Leadership: None declared.
Consultant or Advisory Role: None declared.
Stock Ownership: None declared.
Honoraria: None declared.
Research Funding: J.H. Ix (principal investigator), National Heart, Lung, and Blood Institute (NHLBI) grants R21HL091217 and contracts N01-HC-95159 through N01-HC-95169 with the NHLBI, to the VA San Diego Healthcare System; R. Katz, NHLBI grants R21HL091217 and contracts N01-HC-95159 through N01-HC-95169 with the NHLBI; D. Sisovick, NHLBI grants R21HL091217 and contracts N01-HC-95159 through N01-HC-95169 with the NHLBI; M. Shlipak, NHLBI grants R21HL091217 and contracts N01-HC-95159 through N01-HC-95169 with the NHLBI.
Expert Testimony: None declared.
Role of Sponsor: The funding organizations played no role in the design of study, choice of enrolled patients, review and interpretation of data, or preparation or approval of manuscript.
- Received for publication October 15, 2011.
- Accepted for publication February 2, 2012.
- © 2012 The American Association for Clinical Chemistry