Prognostic Usefulness of Free Fatty Acids in Patients With Stable Coronary Heart Disease

Prognostic Usefulness of Free Fatty Acids in Patients With Stable Coronary Heart Disease Lutz P. Breitling, MDa,*, Dietrich Rothenbacher, MDa,b, Norma C. Grandi, DiplBiola, Winfried März, MDc, and Hermann Brenner, MDa Circulating nonesterified or free fatty acids (FFAs) may contribute to the development of cardiovascular pathology and correlate with ischemia in acute cardiovascular conditions. The aim of this study was to assess whether serum levels of FFAs are associated with long-term prognosis in subjects with stable coronary heart disease. This observational prospective cohort study included 1,206 participants in 3-weeks inpatient rehabilitation programs after acute myocardial infarction, coronary syndromes, or coronary intervention at 2 rehabilitation clinics in Germany (1999 to 2000). Eight-year prognosis (time to a secondary fatal or nonfatal cardiovascular disease event including myocardial infarction and stroke [n ⴝ 153] and time to death from any cause [n ⴝ 124]) was examined according to FFA quartiles and in spline regression. FFAs were correlated with established serum markers of cardiovascular risk and strongly related to secondary cardiovascular events and all-cause mortality in age- and gender-adjusted analysis. When additionally controlling for multiple established risk factors and risk markers, the hazard ratio in the fourth versus first quartile was 1.34 (95% confidence interval 0.79 to 2.24) for secondary cardiovascular events and 1.09 (95% confidence interval 0.62 to 1.91) for all-cause mortality. Dose-response modeling suggested that very high FFAs might predict an increased risk for mortality (hazard ratio 1.98, 95% confidence interval 0.98 to 4.02, for 95th percentile vs first quartile). In conclusion, FFAs are closely correlated with cardiovascular risk markers, and in particular, very high FFA might identify patients with stable coronary heart disease with worse prognoses. © 2011 Elsevier Inc. All rights reserved. (Am J Cardiol 2011;108:508 –513) Levels of circulating free fatty acids (FFAs), which generally are known as important substrates for energy homeostasis, have more recently attracted attention as potential risk factors for cardiovascular disorders.1 Increased levels have been found in obesity, and prospective associations with insulin resistance and type 2 diabetes mellitus have been described.2,3 Apart from thus being markers and possibly causal determinants of metabolic syndrome–related disorders, FFAs might furthermore influence cardiovascular risk by interacting with endothelial function, inflammation, or other components of lipid metabolism.4 – 6 In nondiabetic subjects, FFAs are associated with familial cardiovascular risk.7 To further elucidate whether FFAs might indeed be useful for risk stratification in patients with stable coronary heart disease, we studied the association of serum FFA levels with prognosis in the Long-term Success of Cardiologic Rehabilitation Therapy (KAROLA) cohort, a welldefined cardiovascular high-risk population with 8 years of

a

Division of Clinical Epidemiology and Aging Research C070, German Cancer Research Center, Heidelberg, Germany; bInstitute of Epidemiology and Medical Biometry, Ulm University, Ulm, Germany; and cSynlab Center of Laboratory Diagnostics Heidelberg, Eppelheim, Germany. Manuscript received February 4, 2011; revised manuscript received and accepted March 22, 2011. This work was supported in part by the German Federal Ministry of Education and Research (Grant 01GD9820/0) and the Pitzer Foundation, Bad Nauheim, Germany. *Corresponding author: Tel: 49-6221-421343; fax: 49-6221-421302. E-mail address: [email protected] (L.P. Breitling). 0002-9149/11/$ – see front matter © 2011 Elsevier Inc. All rights reserved. doi:10.1016/j.amjcard.2011.03.076

follow-up for nonfatal secondary cardiovascular disease (CVD) events as well as mortality.8,9 Appreciating that whereas most circulating FFAs are bound to albumin, the unbound FFA fraction (FFAu) might be more closely related to pathology in acute cardiovascular conditions,10,11 we additionally explored the prognostic value of the FFA/albumin ratio, a good proxy of FFAu levels.11 Methods Our analyses were based on data from the KAROLA study, which has been described in detail elsewhere.8,9 In brief, patients aged 30 to 70 years and admitted from January 1999 to May 2000 to 1 of 2 participating rehabilitation clinics for in-hospital rehabilitation within 3 months (mean 6 weeks) after acute cardiovascular events (acute myocardial infarction, coronary syndromes, or coronary artery intervention) were eligible for participation in this prospective follow-up study. Inclusion was conditional on written informed consent. The study protocol was approved by the ethics boards of the physicians’ chambers of Hessen and Baden-Württemberg and of the University of Ulm and the University of Heidelberg. Baseline information was obtained from standardized self-administered questionnaires at the beginning of the rehabilitation program and from hospital records. Active follow-up was conducted 1, 3, 4.5, 6, and 8 years after baseline, obtaining information on incident secondary CVD events (myocardial infarction or stroke) from treating general practitioners and, in the case of deceased participants, www.ajconline.org

Coronary Artery Disease/FFA in Coronary Heart Disease

509

Table 1 Baseline characteristics and free fatty acids* in 1,139 patients with stable coronary heart disease

Table 2 Age- and gender-adjusted Spearman’s correlations of free fatty acids with laboratory parameters

Characteristic

Parameter

Age (years) 30–39 40–49 50–59 60–70 Women Men Acute coronary syndromes Coronary intervention Myocardial infarction Body mass index (kg/m2) ⱕ25 ⬎25–30 ⬎30 Diabetes mellitus Hypertension Discharge prescriptions Angiotensin-converting enzyme inhibitors Lipid-lowering drugs Smoking Never Formerly Currently

n

%

FFAs (mmol/L) Mean

95% CI

27 149 334 629 176 963 138 330 671

2.4% 13% 29% 55% 16% 85% 12% 29% 59%

0.38 0.40 0.41 0.45 0.49 0.42 0.40 0.41 0.44

0.30–0.49 0.36–0.43 0.39–0.44 0.43–0.47 0.45–0.53 0.41–0.43 0.36–0.44 0.39–0.44 0.43–0.46

318 644 176 199 627

28% 57% 16% 18% 55%

0.38 0.43 0.50 0.54 0.45

0.36–0.41 0.42–0.45 0.47–0.54 0.51–0.58 0.43–0.47

608

54%

0.46

0.44–0.48

876

77%

0.42

0.41–0.44

350 726 63

31% 64% 5.5%

0.45 0.42 0.43

0.43–0.48 0.40–0.44 0.37–0.50

N-terminal–pro-brain natriuretic peptide Fasting glucose High-density lipoprotein cholesterol Total cholesterol Triglycerides Cholesteryl ester transfer protein High-sensitivity C-reactive protein Interleukin-6 Adiponectin Fetuin A Adipocyte fatty acid–binding protein Retinol-binding protein-4 Lipoprotein-associated phospholipase A2 Secretory phospholipase A2

r

p

⫺0.033

0.27

0.195 ⫺0.017 0.016 0.150 ⫺0.144 0.074 0.093 ⫺0.097 0.071 0.289 ⫺0.080 0.056

⬍0.0001 0.57 0.59 ⬍0.0001 ⬍0.0001 0.012 0.0017 0.0011 0.017 ⬍0.0001 0.0072 0.058

0.075

0.011

* Reported are geometric means. Numbers do not always sum to 1,139, because of missing values for some covariates.

death certificates indicating the cause of death from public health authorities. Blood samples were taken at discharge from rehabilitation using serum Monovettes (Sarstedt AG & Company, Nümbrecht, Germany), centrifuged according to the manufacturer’s instructions, cooled at ⫺20°C for a maximum of 4 weeks, mailed to the study center on dry ice, and stored at ⫺80°C until analysis. FFAs were measured using the NEFA FS* kit (DiaSys Diagnostic Systems, Holzheim, Germany), which quantifies FFAs through an enzymatic end point method with manufacturer-provided intra- and interassay coefficients of variation of about 1.1%. Details of other laboratory measurements and respective assays have been described elsewhere.9,12–15 To maximize the comparability with the most relevant previous publication,16 FFA quartiles were used in most analyses. The associations of baseline FFAs with secondary CVD events (nonfatal myocardial infarction or stroke, or death from CVD [up to year 4.5 of follow-up: International Classification of Diseases, Ninth Revision, codes 390 to 459; years 6 and 8 of follow-up: International Classification of Diseases, Tenth Revision, codes I00 to I99, 1 case of code R57.0]) and all-cause mortality were first examined by Kaplan-Meier plotting. In subsequent Cox regression analysis estimating hazard ratios (HR), which can be considered a measure of relative risk, model 1 was adjusted for age and gender; model 2 was additionally adjusted for the most important established risk indicators to investigate whether FFAs, regardless of causal considerations, could contribute

Figure 1. Scatterplot of concentrations of FFAs and triglycerides in 1,138 patients with stable coronary heart disease.

to risk stratification beyond these traditional predictors of prognosis in patients with stable coronary heart disease (body mass index, history of diabetes, hypertension, and myocardial infarction, smoking status, renal function, discharge prescription of angiotensin-converting enzyme inhibitors, extent of heart failure as assessed by N-terminal– pro-brain natriuretic peptide); model 3 was additionally adjusted for markers of metabolic dysregulation to investigate if any association in model 2 would be attenuated by the inclusion of these additional potentially mediating variables (blood glucose, lipid-lowering drugs, high-density lipoprotein and total cholesterol, and triglycerides); and model 4 was adjusted like model 3 but excluded covariates with p values ⬎0.20 to reduce possible overfitting. Strongly skewed continuous variables (N-terminal–pro-brain natriuretic peptide, fasting glucose, triglycerides, high-sensitivity C-reactive protein, and interleukin-6) were log transformed. The proportional-hazards assumption was assessed by testing interaction terms of the main exposure with log(survival time)17 while adjusting for age and gender. The nonlinearity of associations of continuous covariates with outcomes was assessed in model 3 by including quadratic terms.

510

The American Journal of Cardiology (www.ajconline.org)

Figure 2. Kaplan-Meier curves of the time to a secondary CVD event (top) or death from any cause (bottom) by FFA quartile.

The potential effect heterogeneity of FFAs according to the presence of diabetes was tested by introducing an interaction term between these 2 variables in model 4. The dose-response relations of the main exposures with outcomes were examined using restricted cubic splines, a regression method that allows nonlinear associations between a risk factor and an outcome.18 Sensitivity analyses included additional adjustment for the potentially mediating variable cholesteryl ester transfer protein,4 or the inflammatory marker interleukin-6,5 or correcting smoking status on the basis of self-report through serum cotinine measurements.19,20 As a surrogate for FFAu, the FFA/albumin ratio was also examined in the main models. In particular, its associations with the outcomes were examined after categorization into quartiles and adjusted like model 4, and its dose-response relation was studied in spline regression models. More sophisticated procedures predicting FFAu from parametric approximations of complex FFA/albumin binding dynamics have been developed21 but were disregarded here in favor of the FFA/albumin ratio put forward by the same group.11 Results Baseline FFA measurements were available for 1,045 of 1,091 subjects (96%) with secondary CVD follow-up and

for 1,139 of 1,204 subjects (95%) with mortality follow-up (1 subject with an outlying FFA concentration of 3.44 mmol/L was excluded). The FFA concentrations were right skewed, with a geometric mean of 0.43 mmol/L (95% confidence interval [CI] 0.42 to 0.44) and quartile cutoffs of 0.32, 0.43, and 0.60 mmol/L. Table 1 lists the distributions of baseline characteristics and FFAs in the analysis population. Mean FFA level appeared higher in older subjects and in women and furthermore in the presence of components of the metabolic syndrome (obesity, diabetes, and hypertension). For various laboratory markers assessed in KAROLA, fairly pronounced correlations with FFAs emerged (Table 2). Figure 1 shows a scatterplot of FFAs versus triglycerides. During a median of 8.1 years of follow-up, 153 nonfatal or fatal secondary CVD events occurred (51 nonfatal myocardial infarctions, 41 nonfatal strokes, and 61 cardiovascular deaths) and an additional 63 deaths due to noncardiovascular causes. Kaplan-Meier curves of the time to secondary CVD events or death from any cause suggested rather limited survival differences between subjects with different FFA concentrations (Figure 2). In the Cox regression models adjusted only for age and gender, the risk for an adverse outcome appeared almost doubled in subjects in the fourth compared to the first quartile of FFAs (Table 3). These associations, however, were strongly attenuated when adjusting for additional risk markers, and in the case of the all-cause mortality disappeared, suggesting that there might be only limited independent correlation of FFA with prognosis. The Cox regression models appeared well specified. In particular, the interaction terms of log(observation time) and FFA quartiles revealed no significant violation of the proportional-hazards assumption, with p ⫽ 0.40 for secondary CVD events and p ⫽ 0.46 for all-cause mortality. The continuous covariates in model 3 showed no statistically significant nonlinearity: testing 7 covariates and 2 different end points, the smallest p value was 0.016, which was observed for the squared body mass index term in the all-cause mortality model. The dose-response relations obtained by spline modeling are shown in Figure 3. In line with the results listed in Table 3, the 2 curves were fairly flat. For all-cause mortality, extremely high values of FFAs seemed to be associated with an increase in risk. Given this pattern, we also explored the association after splitting the highest quartile at the 95th percentile (FFA ⱖ 0.90 mmol/L). The estimates for these extreme FFA values compared to the lowest quartile were unremarkable for secondary CVD events (HR 1.43, 95% CI 0.67 to 3.03) and only suggestive of an association for all-cause mortality (HR 1.98, 95% CI 0.98 to 4.02) (adjusted like model 4 in Table 3). The interaction terms of FFAs with prevalent diabetes were significant neither for secondary CVD events (p ⫽ 0.92) nor for all-cause mortality (p ⫽ 0.092). For secondary CVD events, the HRs were consistently higher if diabetes was present (2.76, 2.29, and 2.45, for quartiles 2, 3, and 4) than otherwise (1.31, 1.28, and 1.28), but the confidence intervals were excessively wide (not shown). The pattern for mortality was less homogenous (1.10, 0.45, and 0.63 with diabetes and 0.84, 1.16, and 1.29 without). A post hoc

Coronary Artery Disease/FFA in Coronary Heart Disease

511

Table 3 Cox regression models* of free fatty acids as predictors of secondary cardiovascular disease events or all-cause mortality Outcome

Secondary CVD FFA ⱕ0.32 mmol/L FFA ⬎0.32–0.43 mmol/L FFA ⬎0.43–0.60 mmol/L FFA ⬎0.60 mmol/L Death FFA ⱕ0.32 mmol/L FFA ⬎0.32–0.43 mmol/L FFA ⬎0.43–0.60 mmol/L FFA ⬎0.60 mmol/L

Model 1

Model 2

Model 3

Model 4

HR

95% CI

HR

95% CI

HR

95% CI

HR

95% CI

1 1.65 1.51 1.94

Reference 1.01–2.70 0.91–2.49 1.18–3.17

1 1.50 1.36 1.43

Reference 0.90–2.49 0.81–2.30 0.84–2.44

1 1.39 1.33 1.33

Reference 0.83–2.32 0.79–2.25 0.78–2.29

1 1.40 1.31 1.34

Reference 0.84–2.33 0.78–2.21 0.79–2.27

1 1.34 1.25 1.87

Reference 0.77–2.31 0.72–2.17 1.11–3.17

1 1.17 1.04 1.22

Reference 0.66–2.07 0.58–1.86 0.69–2.16

1 1.09 0.96 1.04

Reference 0.61–1.94 0.53–1.72 0.58–1.88

1 1.05 0.94 1.09

Reference 0.60–1.85 0.53–1.67 0.62–1.91

* Model 1 adjusted for age and gender; model 2 also adjusted for body mass index, history of diabetes, hypertension, myocardial infarction, smoking status, impaired renal function, angiotensin-converting enzyme inhibiting drugs, and N-terminal–pro-brain natriuretic peptide; model 3 also adjusted for fasting glucose, lipid-lowering drugs, high-density lipoprotein cholesterol, total cholesterol, and triglycerides; model 4 adjusted like model 3 but with covariates with p values ⬎0.20 removed (secondary CVD: body mass index, hypertension, impaired renal function, fasting glucose, lipid-lowering drugs, total cholesterol; death: body mass index, hypertension, myocardial infarction, impaired renal function, fasting glucose, high-density lipoprotein cholesterol, and total cholesterol).

exploration of interactions between FFAs and the prescription of lipid-lowering drugs yielded no support for pronounced heterogeneity of the associations according to this covariate (not shown). For the 2 outcomes, additional adjustment of model 4 for either cholesteryl ester transfer protein, log(interleukin-6), or biomarker-based correction of smoking status resulted in no relevant changes of the HRs (not shown). Albumin concentrations ranged from 33.8 to 58.6 g/L (median 47.6, interquartile range 45.5 to 50.0). The associations for the FFA/albumin ratio were very similar to those for FFA itself. The HRs obtained when adjusting like model 4 in Table 3 were 1.36 (95% CI 0.81 to 2.30), 1.34 (95% CI 0.78 to 2.28), and 1.36 (95% CI 0.80 to 2.30) for secondary CVD events and 1.02 (95% CI 0.56 to 1.85), 1.13 (95% CI 0.63 to 2.02), and 1.11 (95% CI 0.62 to 1.97) for all-cause mortality, listed from the second to the fourth quartile in reference to the first quartile. The estimates for subjects with values higher than the 95th percentile were 1.42 (95% CI 0.67 to 3.01) for secondary CVD events and 2.13 (95% CI 1.04 to 4.38) for all-cause mortality. Cubic spline estimates for these analyses are shown in Figure 4. Discussion In this long-term follow-up study of patients with stable coronary heart disease, FFAs were correlated with established laboratory markers of cardiovascular risk. Associations were observed with cardiovascular events and allcause mortality, although they were substantially reduced and no longer statistically significant after controlling for established risk markers. The results of adjusted regression models nonetheless showed a point estimate of ⬎30% increased risk for secondary cardiovascular events in subjects in the highest FFA quartile. The associations of FFA with major participant characteristics, for example, age, gender, body mass index, and prevalent diabetes, were in line with the pertinent research.1,3 Furthermore, in correlation analyses including

several novel cardiovascular risk markers, the direction of the correlations with FFAs rather consistently supported the notion that higher FFAs might mark a cardiovascular highrisk profile (negative correlations with cholesteryl ester transfer protein13 and adiponectin14 and positive correlations with fasting glucose, triglycerides, and interleukin-6). Given the close physiologic relation between triglycerides and FFAs,1 it appears noteworthy that the pertinent correlation coefficient in the present study was very similar to those in previous studies.16 These patterns overall strengthen the confidence in the validity of our measurements. Pronounced associations of FFA with sudden (cardiac) death have been described in subjects free of CVD and in those referred for coronary angiography.3,22 The corresponding HRs were consistent with approximately a doubling of the cause-specific mortality per standard deviation increase22 or in the fourth compared to the first quartile of FFAs.3 Interestingly, no association with fatal myocardial infarction was observed in 1 of these studies (HR per standard deviation 0.94, 95% CI 0.75 to 1.09).22 In a different report on the same study, the absence of an association with death from coronary heart disease was described along with a pronounced association with cancer mortality.23 The contrast tested in this case was the fifth quintile of FFA versus the bottom 80%; crude cumulative mortality from coronary heart disease was 4.4% in the fifth and 2.5% in the first quintile, the ratio of these proportions being not unlike our findings in the less adjusted models. Additional other studies have described crude associations of FFA with incident heart disease or mortality that did not withstand covariate adjustment.24,25 Further analyses of the angiography referral cohort, in contrast, provided more convincing support for independent associations of FFAs with cardiovascular and all-cause mortality.16 In subgroup analyses restricted to participants with stable coronary artery disease, the extensively adjusted HRs of the fourth versus first quartile were 1.88 (95% CI 1.19 to 2.96) for cardiovascular mortality and 1.60 (95% CI 1.10 to

512

The American Journal of Cardiology (www.ajconline.org)

Figure 3. Relative risk for experiencing a secondary CVD event (top) or dying from any cause (bottom) according to FFA level, in comparison to subjects with 0.32 mmol/L (first quartile). Shown are restricted cubic spline estimates of the HRs on the y axis (dotted curves: 95% CI limits) at different values of FFAs in millimoles per liter on the x axis. The models were adjusted like model 4 in Table 3, and the spline knots were placed at the 10th, 50th, and 90th percentiles; the graphs were cut at the 99th percentile.

2.34) for all-cause mortality. It appears noteworthy that the concentrations in that cohort were altogether higher than in the present study, and the highest quartile cutoff of 0.89 mmol/L resembled our 95th percentile. Whereas differences in assay characteristics cannot be ruled out, this discrepancy also could be the result of the different timing of the blood draws, which were done just before angiography,16 in contrast to several weeks after the acute events in our study. If acute increases, especially of the unbound fraction, in FFAs indeed provide an indication of the extent of ischemia,10,11 the pronounced association reported in the angiography referral cohort might be due to FFAs being a marker of acute disease severity. This could explain why adjusted associations with cardiovascular mortality seem to be less consistent in generally healthy subjects23 or if FFA is determined several weeks after the acute event (i.e., when FFA measurements are more likely to reflect physiologic longterm levels, as in our study). We are not aware of prognostic studies pertaining to

Figure 4. Restricted cubic spline estimates (HRs on y axis; dotted curves: 95% CI limits) of the association of FFA/albumin ratio (millimoles per gram) on the x axis as a surrogate for unbound FFAs with secondary CVD events (top) or all-cause mortality (bottom), adjusted like model 4 in Table 3. Knots were placed at the 10th, 50th, and 90th percentiles; 0.00666 mmol/g (the 25th percentile) served as the reference ratio. The graphs were cut at the 99th percentile.

FFAu or the FFA/albumin ratio as its proxy.11 Whereas our data do not suggest that replacing FFA with FFA/albumin would result in extraordinary changes in the associations of interest, it appears reasonable to further explore the predictive value of this variable in other studies, in which albumin should be commonly available. This similarly applies to our exploratory analyses of interactions with prevalent diabetes or of the impact of very high FFA levels. Our study population was recruited after survivor selection through prehospital and early mortality had occurred. This probably is the main reason for the overall small number of events observed in our study, which resulted in limited statistical power, particularly for the interaction analyses. Furthermore, FFA measurements were available only from the end of the rehabilitation program. Whereas this was preferable for the purpose of avoiding the acute changes due to cardiac ischemia during the acute event, the 3-week inpatient rehabilitation itself also may have led to temporary changes in FFA levels,26 quite possibly diluting associations of FFAs with long-term prognosis. More con-

Coronary Artery Disease/FFA in Coronary Heart Disease

tinuous assessments of FFAs after acute CVD would be a powerful asset of any future study trying to correlate this marker with prognosis. 1. Pilz S, Marz W. Free fatty acids as a cardiovascular risk factor. Clin Chem Lab Med 2008;46:429 – 434. 2. Boden G. Interaction between free fatty acids and glucose metabolism. Curr Opin Clin Nutr Metab Care 2002;5:545–549. 3. Pilz S, Scharnagl H, Tiran B, Wellnitz B, Seelhorst U, Boehm BO, Marz W. Elevated plasma free fatty acids predict sudden cardiac death: a 6.85-year follow-up of 3315 patients after coronary angiography. Eur Heart J 2007;28:2763–2769. 4. Arii K, Suehiro T, Yamamoto M, Ito H, Hashimoto K. Suppression of plasma cholesteryl ester transfer protein activity in acute hyperinsulinemia and effect of plasma nonesterified fatty acid. Metabolism 1997;46:1166 –1170. 5. Mathew M, Tay E, Cusi K. Elevated plasma free fatty acids increase cardiovascular risk by inducing plasma biomarkers of endothelial activation, myeloperoxidase and PAI-1 in healthy subjects. Cardiovasc Diabetol 2010;9:9. 6. Steinberg HO, Tarshoby M, Monestel R, Hook G, Cronin J, Johnson A, Bayazeed B, Baron AD. Elevated circulating free fatty acid levels impair endothelium-dependent vasodilation. J Clin Invest 1997;100: 1230 –1239. 7. Carlsson M, Wessman Y, Almgren P, Groop L. High levels of nonesterified fatty acids are associated with increased familial risk of cardiovascular disease. Arterioscler Thromb Vasc Biol 2000;20:1588 – 1594. 8. Breitling LP, Grandi NC, Hahmann H, Wusten B, Rothenbacher D, Brenner H. Gamma-glutamyltransferase and prognosis in patients with stable coronary heart disease followed over 8 years. Atherosclerosis 2010;210:649 – 655. 9. Rothenbacher D, Koenig W, Brenner H. Comparison of N-terminal pro-B-natriuretic peptide, C-reactive protein, and creatinine clearance for prognosis in patients with known coronary heart disease. Arch Intern Med 2006;166:2455–2460. 10. Apple FS, Wu AH, Mair J, Ravkilde J, Panteghini M, Tate J, Pagani F, Christenson RH, Mockel M, Danne O, Jaffe AS. Future biomarkers for detection of ischemia and risk stratification in acute coronary syndrome. Clin Chem 2005;51:810 – 824. 11. Kleinfeld AM, Prothro D, Brown DL, Davis RC, Richieri GV, DeMaria A. Increases in serum unbound free fatty acid levels following coronary angioplasty. Am J Cardiol 1996;78:1350 –1354. 12. Roos M, von Eynatten M, Heemann U, Rothenbacher D, Brenner H, Breitling LP. Serum fetuin-A, cardiovascular risk factors, and six-year follow-up outcome in patients with coronary heart disease. Am J Cardiol 2010;105:1666 –1672. 13. Duwensee K, Breitling LP, Tancevski I, Rothenbacher D, Demetz E, Patsch JR, Ritsch A, Eller P, Brenner H. Cholesteryl ester transfer protein in patients with coronary heart disease. Eur J Clin Invest 2010;40:616 – 622.

513

14. von Eynatten M, Hamann A, Twardella D, Nawroth PP, Brenner H, Rothenbacher D. Relationship of adiponectin with markers of systemic inflammation, atherogenic dyslipidemia, and heart failure in patients with coronary heart disease. Clin Chem 2006;52:853– 859. 15. Koenig W, Vossen CY, Mallat Z, Brenner H, Benessiano J, Rothenbacher D. Association between type II secretory phospholipase A2 plasma concentrations and activity and cardiovascular events in patients with coronary heart disease. Eur Heart J 2009;30:2742–2748. 16. Pilz S, Scharnagl H, Tiran B, Seelhorst U, Wellnitz B, Boehm BO, Schaefer JR, Marz W. Free fatty acids are independently associated with all-cause and cardiovascular mortality in subjects with coronary artery disease. J Clin Endocrinol Metab 2006;91:2542–2547. 17. Hosmer DW, Lemeshow S. Applied Survival Analysis: Regression Modeling of Time to Event Data. New York, New York: John Wiley, 1999. 18. Desquilbet L, Mariotti F. Dose-response analyses using restricted cubic spline functions in public health research. Stat Med 2010;29: 1037–1057. 19. Twardella D, Rothenbacher D, Hahmann H, Wusten B, Brenner H. The underestimated impact of smoking and smoking cessation on the risk of secondary cardiovascular disease events in patients with stable coronary heart disease: prospective cohort study. J Am Coll Cardiol 2006;47:887– 889. 20. Hellerstein MK, Benowitz NL, Neese RA, Schwartz JM, Hoh R, Jacob P III, Hsieh J, Faix D. Effects of cigarette smoking and its cessation on lipid metabolism and energy expenditure in heavy smokers. J Clin Invest 1994;93:265–272. 21. Richieri GV, Anel A, Kleinfeld AM. Interactions of long-chain fatty acids and albumin: determination of free fatty acid levels using the fluorescent probe ADIFAB. Biochemistry 1993;32:7574 –7580. 22. Jouven X, Charles MA, Desnos M, Ducimetiere P. Circulating nonesterified fatty acid level as a predictive risk factor for sudden death in the population. Circulation 2001;104:756 –761. 23. Charles MA, Fontbonne A, Thibult N, Claude JR, Warnet JM, Rosselin G, Ducimetiere P, Eschwege E. High plasma nonesterified fatty acids are predictive of cancer mortality but not of coronary heart disease mortality: results from the Paris Prospective Study. Am J Epidemiol 2001;153:292–298. 24. Forsblom CM, Sane T, Groop PH, Totterman KJ, Kallio M, Saloranta C, Laasonen L, Summanen P, Lepantalo M, Laatikainen L, Matikainen E, Teppo AM, Koskimies S, Groop L. Risk factors for mortality in Type II (non-insulin-dependent) diabetes: evidence of a role for neuropathy and a protective effect of HLA-DR4. Diabetologia 1998;41: 1253–1262. 25. Pirro M, Mauriege P, Tchernof A, Cantin B, Dagenais GR, Despres JP, Lamarche B. Plasma free fatty acid levels and the risk of ischemic heart disease in men: prospective results from the Quebec Cardiovascular Study. Atherosclerosis 2002;160:377–384. 26. Kupper-Nybelen J, Rothenbacher D, Hahmann H, Wusten B, Brenner H. Changes of risk factors in patients with coronary heart disease after in-patient rehabilitation [article in German]. Dtsch Med Wochenschr 2003;128:1525–1530.