C-Reactive Protein Predicts Response to Pentoxifylline in Patients with Idiopathic Dilated Cardiomyopathy

We have previously reported that the addition of pentoxifylline to standard heart failure therapy results in improvements in systolic function in patients with idiopathic dilated cardiomyopathy (IDC) 1–3. However, these improvements, which are associated with a reduction in inflammatory cytokines 4–7 and Fas/Apo-1 8 only occur in some patients. C-reactive protein (CRP) is an inflammatory marker which is associated with adverse prognosis in patients with IDC 9,10. It is therefore possible that elevations of CRP at baseline may predict the response of patients with IDC to the immuno-modulating agent pentoxifylline. Our primary objective was to evaluate whether the degree of elevation of CRP at baseline predicts left ventricular systolic function in patients with IDC treated with pentoxifylline. A secondary objective was to determine whether elevations of CRP are associated with changes in TNF-alpha and a plasma marker of apoptosis (Fas/Apo-1). In the present analysis we used the databases of two single-center, double-blind, randomized trials of pentoxifylline treatment in patients with idiopathic dilated cardiomyopathy in New York Heart Association Functional Class II–III heart failure. In the first study (n=49) the patients received treatment with diuretics, digoxin and angiotensin-converting-enzyme inhibitors for at least 4 months prior to randomization 8. In the second trial (n=39) carvedilol was added to baseline medication and given for 4 months prior to randomization 2. In both studies the treatment arm received pentoxifylline 400 mg TID in addition to previous therapy. The investigation conforms with the principles outlined in the Declaration of Helsinki. Inclusion criteria for both trials were: (1) age≥18 and ≤70 years, (2) stable NYHA functional class II or III congestive heart failure of unknown etiology, (3) left ventricular ejection fraction <40% by radionuclide ventriculography after 4 months of standard therapy, (4) sinus rhythm, and (5) eligible patients in whom high quality echocardiographic images could be obtained. Exclusion criteria were: (1) chronic obstructive pulmonary disease, (2) significant valvular heart disease, (3) history or evidence of ischaemic heart disease, (4) systolic blood pressure >160 mmHg and/or diastolic blood pressure >95 mmHg, (5) clinical conditions other than cardiomyopathy that could increase cytokine levels, (i.e. rheumatoid arthritis, sepsis), (6) pregnancy, (7) severe liver disease, defined as enzymes >2 times the upper limit of normal, and (8) any clinical condition that according to the investigators precluded inclusion in the study. Clinical examination, echocardiography and radionuclide ventriculography studies were performed prior to and 6 months after randomization. Details of the measurement of left ventricular function are described elsewhere 1,2. Improvement in radionuclide ventriculography ejection fraction was predefined as an increase in more than 10 absolute units. For plasma CRP measurements blood was obtained prior to randomization and measured as a routine investigation with a commercially available enzyme-linked immunosorbent assay at the hospital laboratory (Roche Diagnostics). The assay used was of low sensitivity and uses 0.1–10 mg/l as a referent range for normal values. For plasma cytokine and Fas/Apo-1 receptor measurements blood was obtained prior to and 6 months after randomization. Plasma was stored at −70 °C within 15 min of blood collection. Plasma concentrations of TNF-α were determined from undiluted samples using commercially available enzyme-linked immunoassays (Amersham, Maidstone). Fas/Apo-1 concentrations were measured from appropriately diluted samples using a commercially available non-isotopic quantitative immunoassay (Calbiochem). Measurements were made in triplicate and the mean of 3 values used for statistical analysis. Data management and statistical analysis were performed with SAS software, version 8.0 (SAS Institute INC.). Between group and within group differences in continuous measurements were tested with Wilcoxon two sample tests and Wilcoxon matched pairs test respectively as well as unpaired and paired t-test when variables showed normal distribution. Proportions were compared using Chi-Square test or Fisher exact test when required. Spearman correlation coefficients adjusted for intake of carvedilol were calculated and single and stepwise multiple linear regression analysis was used to analyze the relationship between change in ejection fraction and various variables including CRP, TNF-alpha and Fas/Apo-1. Significance was assumed at a two-tailed value of P<0.05. At randomization no differences in clinical characteristics were noted between the 45 patients in the treatment group and the 43 patients in the placebo group (Table 1a). Importantly, the doses of enalapril, digoxin, furosemide and carvedilol were similar between the groups (Table 1b). In addition, there were no differences in the characteristics of the patients receiving a beta-blocker (as part of their standard therapy) vs. those not receiving a beta-blocker (data not shown). Thirteen patients died (nine in the placebo group, P=0.058 between groups) and three patients were not available for follow-up. After 6 months the group of patients treated with pentoxifylline had a significant increase in ejection fraction and a significant decline in the plasma levels of TNF-alpha and Fas/Apo-1 (Table 2) whereas no significant changes occurred in the placebo group. Eighteen of the pentoxifylline treated patients improved their radionuclide ejection fraction by >10 absolute units, and were arbitrarily defined as improvers. These patients had significantly higher CRP levels at randomization (n=18, 13.4±4.4 mg/l) compared to the patients that did not improve their ejection fraction on therapy with pentoxifylline (n=24, 8.3±4.4 mg/l; P<0.0014). In the placebo group CRP levels did not differ between the improvers (n=6) and the non-improvers (10.3±4.2 vs. 9.5±5.2 respectively, P=0.52). When splitting CRP levels in the pentoxifylline group along the median CRP=10.65 mg/l, above the median CRP levels 21 patients showed a mean change in EF of 15.0±12.9 (P=0.004) and below the median CRP levels 21 patients showed a mean change in EF of 3.4±9.5 (P=0.39). Instead in the placebo group (median CRP=10.00 mg/l), above the median CRP levels 12 patients showed a mean change in EF of −0.4±13.7 and below the median CRP levels 18 patients showed a mean change in EF of 0.9±6.9. CRP at randomization was correlated with the change in ejection fraction after therapy with pentoxifylline (P<0.0006) (Table 3). In addition, CRP at randomization was correlated with the change in Fas/Apo-1 after 6 months of therapy with pentoxifylline (P<0.05, Table 3). However CRP was not correlated with changes in TNF-alpha (Table 3). No significant correlations were observed in the patients treated with placebo (Table 3). Stepwise multiple linear regression analysis in the pentoxifylline treated group (multiple r=0.55; P=0.0001) showed that CRP and Fas/Apo-1 at randomization were independent predictors of change in ejection fraction after 6 months of therapy with pentoxifylline (Table 4). Like other treatment strategies for heart failure, improvements in systolic function with pentoxifylline occur only in a sub-fraction of the patients treated 1,2. Hence it is important to identify those patients who stand to benefit the most from this therapy. Improvements in systolic function with pentoxifylline are accompanied by reductions in the inflammatory cytokine TNF-α and in the marker of apoptosis Fas/Apo-1 1,2,8. However as TNF-α is subject to circadian variation 11 and has a short half-life, it is unlikely to be reliable as a marker to predict improvements. In contrast the inflammatory marker CRP shows no circadian variation and is highly reproducible within individuals 12,13. CRP is an acute-phase protein which recognizes a range of pathogenic targets including membranes of apoptotic and reactive cells 14. In addition CRP is implicated in the synthesis of TNF-α 15,16. Hence it is possible that CRP may predict those patients who are likely to respond to the immuno-modulating agent pentoxifylline. Indeed we found that those patients who responded to pentoxifylline had higher initial concentrations of CRP than those who did not. Furthermore, CRP was correlated with reductions in Fas/Apo-1 as well as with improvements in systolic function following pentoxifylline therapy. As CRP measurements are relatively inexpensive and considered routine in comparison to TNF-α and Fas/Apo-1 measurements, CRP could provide us with a readily available tool to target pentoxifylline therapy in a subgroup of patients with IDC.