Aims: We studied the clinical and immunological importance of fasting cholesterol, HDL, LDL and triglycerides in patients with chronic heart failure in relation to plasma concentrations of tumor necrosis factor-alpha (TNFalpha), soluble TNF receptor-1 and -2 (sTNF-R1 and -R2), and a ratio potentially indicating recent endotoxin bioactivity (soluble [s] CD14/total cholesterol).
Methods and results: Fifty-eight stable, non-oedematous patients with established heart failure and 19 controls were studied prospectively. Concentrations of sTNF-R1 and sCD14 were higher in patients than in controls (1238+/-96 vs. 632+/-72 pg/ml, P=0.005 and 3401+/-120 vs. 2775+/-139 pg/ml, P=0.007, respectively), whereas those of TNFalpha (9.3+/-1.1 vs. 6.7+/-0.6 pg/ml) and sTNF-R2 (2464+/-145 vs. 1920+/-303 pg/ml) were not. Cholesterol (5.6+/-0.1 vs. 5.5+/-0.2 mmol/l) and LDL (3.5+/-0.1 vs. 3.6+/-0.2 mmol/l) were not different (both P>0.75). Patients had lower HDL (1.10+/-0.04 vs. 1.4+/-0.06 mmol/l, P=0.0004) and higher triglycerides (2.1+/-0.1 vs. 1.1+/-0.1 mmol/l, P=0.0006). Aetiology and the presence of cardiac cachexia did not influence the lipid profile. Correlations in patients: cholesterol vs. TNFalpha (r=-0.40, P=0.003), vs. sTNF-R1 (r=-0.24, P=0.08), vs. sTNF-R2 (r=-0.29, P<0.04); sCD14 vs. TNFalpha (r=0.44, P=0.005), vs. sTNF-R1: (r=0.65, P<0.0001), vs. sTNF-R2 (r=0.59, P<0. 0001). The sCD14/cholesterol ratio related powerfully to TNFalpha (r=0.60), sTNF-R1 (r=0.74), and sTNF-R2 (r=0.65, all P<0.0001). This sCD14/cholesterol ratio emerged as the strongest predictor of TNFalpha, sTNF-R1 and -R2 (all P<0.01), independently of renal and hepatic function, and conventional measures of disease severity. A cholesterol level <5.2 mmol/l (n=18) significantly predicted a poor clinical outcome (P<0.04, RR 3.5, 95% CI 1.1-11.0) independently of peak VO(2) (P=0.07), NYHA class (P=0.08), aetiology (P=0.14), and age, body wasting, sodium, LVEF, heart rate, and blood pressure (all P>0.20, follow-up 12 months, event rate 26%).
Conclusion: Our data supports previous findings that lower, rather than higher cholesterol levels are associated with poor clinical outcome in patients with chronic heart failure. This relationship is unrelated to heart failure aetiology, and suggests that the classic risk profile is not longer relevant in established heart failure. The little-recognised ability of all lipoprotein fractions to bind endotoxin and to serve as natural buffer substances may explain this relationship between lower lipoprotein levels, higher cytokine concentrations and impaired prognosis.