Author + information
- Received October 20, 1999
- Revision received February 1, 2000
- Accepted March 30, 2000
- Published online August 1, 2000.
- Rakesh Sharma, BSc, MRCP∗,
- Mathias Rauchhaus, MD∗,§,
- Piotr P Ponikowski, MD, PhD∗,
- Susan Varney, BSc∗,
- Philip A Poole-Wilson, MD, FACC∗,
- Douglas L Mann, MD, FACC†,
- Andrew J.S Coats, DM, FACC∗ and
- Stefan D Anker, MD, PhD∗,‡,* ()
- ↵*Reprint requests and correspondence: Dr. Stefan Anker, Cardiac Medicine, NHLI London, Dovehouse Street, London SW3 6LY, United Kingdom
The object of the study was to assess the relationship between erythrocyte sedimentation rate (ESR) and inflammatory cytokine production in chronic heart failure (CHF). Our findings lead us to re-evaluate the prognostic value of the ESR in assessing patients with CHF.
The search for simple prognostic markers in CHF that can be assessed anywhere at low cost is important. Increases in ESR are related to the acute phase response in states of inflammation and infection.
Initially, we studied ESR in relation to plasma levels of inflammatory cytokines in 58 CHF patients. The findings prompted us to analyze the mortality predictive power of ESR compared with established risk factors in these patients and (retrospectively) in a second group of 101 clinically stable CHF patients who had ESR measured.
In all 159 CHF patients (age 62 ± 2 years, New York Heart Association [NYHA] class 2.7 ± 0.1), ESR ranged from 1 to 96 mm/h (median 14 mm/h). The ESR was correlated with tumor necrosis factor (TNF)-alpha (r = 0.31, p < 0.05), soluble TNF receptor-1 (r = 0.48, p < 0.0005), soluble TNF receptor-2 (r = 0.39, p < 0.005) and interleukin 6 (r = 0.45, p < 0.005) levels. High ESR levels indicated a poor prognosis (p < 0.0001), and this was independent of age, NYHA class, ejection fraction and peak oxygen consumption (p < 0.005). Patients with ESR above median (≥15 mm/h) compared with patients with ESR <15 mm/h had an impaired survival (hazard ratio 2.62, 95% confidence interval 1.58–4.36, p < 0.0001).
Our study demonstrates that in CHF a high ESR is an unfavorable prognostic sign, independent of patients’ symptomatology and ventricular function. These results are in diametrical contrast to previous results. This may reflect a change in the underlying pathophysiology due to today’s treatment with angiotensin-converting enzyme inhibitors.
The search for markers of prognosis in chronic heart failure (CHF) is important. Of particular interest are markers that can be assessed anywhere at (nearly) no cost. The use of angiotensin-converting enzyme (ACE) inhibitors for the treatment of CHF patients started only about 10 years ago. This major treatment change might have affected the strength or even the direction of prognostic markers established in the pre-ACE inhibitor era. The possibility that the introduction of new therapies can reverse the direction of conventional prognostic markers is a novel concept that may be applicable to many other diseases. To our knowledge this phenomenon has never been shown before, and this may have implications for other prognostic markers that are currently used in the management of cancer, infectious diseases or other chronic illnesses.
The erythrocyte sedimentation rate (ESR) is a particularly simple measure of clinical status. It has been available to physicians for over eight decades (1), but its value in clinical cardiology is controversial (2). In 1991, Haber et al. (3) reported that a low ESR was associated with an unfavorable prognosis in 242 patients with heart failure. When these patients were re-evaluated after the introduction of new treatments, the clinical improvement and the changes in hemodynamic status were directly related to an increase in ESR. These findings appear to contradict our clinical experience suggesting that a high ESR is a reflection of an acute phase response to inflammation, or infection is present (4), and that it is usually associated with a poor clinical outcome. Since 1991, no single study has been published on ESR in CHF patients.
Initially, we intended to assess the relationship between ESR and inflammatory cytokines and markers of cytokine production in CHF. Our findings lead us to re-evaluate the prognostic value of ESR in assessing patients with CHF. Our results show the opposite of previous findings (3). We put forward a hypothesis to explain the apparent change in the relationship between the ESR and survival in CHF.
Study population and follow-up
We performed a prospective study assessing inflammatory cytokines and ESR in 58 stable patients with CHF (age 61 ± 1, New York Heart Association [NYHA] class 2.7 ± 0.1, left ventricular ejection fraction [LVEF] 27 ± 2%, mean duration of heart failure 5 ± 1 years). The study was approved by the local ethics committee, and all patients and controls gave written informed consent. The results from this study prompted us to analyze mortality in these 58 patients and (retrospectively) in a second group of 101 clinically stable CHF patients (age 63 ± 1, NYHA class 2.7 ± 0.1, LVEF 30 ± 2%) who had ESR measured between May 1992 and June 1998 as part of their routine clinical investigations. The clinical characteristics (age, NYHA class, LVEF and peak oxygen consumption) did not differ significantly between the two study groups (all p > 0.3). Results of survival analyses (using the Cox proportional hazards model) also did not demonstrate a significant difference between the two groups. Baseline clinical details are, therefore, reported for the whole study population in Table 1.
The diagnosis of CHF was based on clinical presentation and standard investigations. The 159 CHF patients were aged 28 to 87 years (mean 62 ± 2 years). The majority of patients (92%) were in NYHA class II to IV (64% were in class III and IV), and 85% were on an ACE inhibitor (see Table 1). The etiology of CHF was ischemic in 72% of patients and idiopathic dilated cardiomyopathy in 28%. In 130 patients LVEF was determined by radionuclide ventriculography or echocardiography. A total of 125 patients underwent treadmill exercise testing with respiratory gas analysis, and the mean peak oxygen consumption (VO2) was 16.8 ± 0.6 ml/kg/min. Patients were excluded from the study if they were clinically unstable, were edematous or had severe renal failure. No subject had clinical signs of infection, rheumatoid arthritis or cancer at the time of ESR assessment. Follow-up for survival status (as of 31/12/98) was available in March/April 1999 from the hospital information system and from the Office of National Statistics where all patients had been flagged for death as part of the Royal Brompton Hospital heart failure registry.
ESR and cytokine analyses
Erythrocyte sedimentation rate was analyzed using standard procedures. The cytokine assays were performed on EDTA plasma from venous blood samples taken from patients after an overnight fast and after at least 20 min of supine rest. Tumor necrosis factor alpha (TNF alpha) levels were measured using an ELISA assay (Medgenix, Fleurus, Belgium, sensitivity 3 pg/ml). As TNF alpha levels are known to be variable due to a short plasma half-life time, we also measured the more stable soluble TNF receptors (sTNFR)-1 and -2 using ELISA tests (R&D Systems, Minneapolis, Minnesota; sensitivity 25 and 2 pg/ml, respectively). Interleukin 6 (IL-6) levels were also measured using an ELISA assay (R&D Systems, Minneapolis, Minnesota; sensitivity 0.094 pg/ml).
All results are presented as mean value ± SEM. Unpaired Student t tests and both univariate and multivariate regression analysis were performed. Cox proportional hazard analysis was used to assess the association between variables and all-cause mortality. Hazard ratio (RR) and 95% confidence interval (CI) for risk factors as well as significance levels for chi-square (likelihood ratio test) are given, and Kaplan-Meier cumulative survival plots were constructed (StatView 5, Abacus Concepts, Berkeley, California).
In the 58 patients initially studied, a mean ESR of 19 ± 2 mm/h (range 1 to 96 mm/h) was found. Plasma levels of TNF alpha (p < 0.05), sTNFR-1 and sTNFR-2 (both p < 0.001, Table 1) were higher than in healthy control subjects of similar age reported earlier (5). Regression analyses showed that the ESR was positively correlated with TNF alpha (r = 0.31, p < 0.05), sTNFR-1 (r = 0.48, p < 0.0005, Fig. 1), sTNFR-2 (r = 0.39, p < 0.005) and IL-6 (r = 0.45, p < 0.005) levels. The second group of 101 CHF patients showed a mean ESR of 20 ± 2 (p > 0.5 vs. first study group).
Considering all 159 CHF patients, the ESR ranged from 1 to 96 mm per h (median 14). When patients were grouped for ESR above median (i.e., ≥15 mm/h, n = 79, mean: 32 mm/h) and ESR <15 mm/h (n = 80, mean: 7 mm/h), patients with low ESR were, on average, four years younger (p < 0.05), but NYHA class, LVEF and the proportion of patients treated with ACE inhibitors were similar (Table 1). There were 26 patients (16%) who had an ESR < 5 mm/h. The mean furosemide equivalent dose for the patients in the study was 101 ± 10 mg. There was a weak correlation between ESR and age (r = 0.16, p = 0.046), but ESR did not correlate with NYHA class, LVEF, duration of CHF and diuretic dose (all p > 0.11). During follow-up (duration 7 to 80 months in survivors, mean 37 ± 3 months), 66 (42%) of the 159 CHF patients died after 2 to 2,443 days (median 569 days). All deaths could be attributed to cardiovascular causes or were of a sudden nature.
In both studies ESR was associated with an impaired survival in CHF patients (prospectively studied patients RR 1.031, p = 0.0016; retrospective study RR 1.026, p = 0.0009). In both studies this was independent of age and NYHA class (data not shown). In the combined data set, survival analyses showed that a high ESR is associated with an impaired survival (p < 0.0001, Table 2). Every mm increase of ESR was related to a 2.9% higher risk (95% CI 1.8–3.9%) of death during follow-up. The group of patients with high ESR (≥15 mm/h) had an impaired survival compared with patients with ESR <15 mm/h (RR 2.62, 95% CI 1.58–4.36, p < 0.0001, Fig. 2). The negative prognostic value of high ESR was independent of age, NYHA class, LVEF (p < 0.0001) and peak VO2 (p < 0.005, Table 2). When restricting the survival analysis to patients with an LVEF <30% (n = 75, mean 17 ± 1%), a high ESR remains predictive of an impaired prognosis (RR 1.029, p < 0.0001).
Considering an ESR cut-off value of 25 mm/h (previously used in reference 3), patients with ESR >25 mm/h had a worse prognosis than patients with ESR ≤25 mm/h (RR 3.70, 95% CI 2.27–6.00, p < 0.0001). The mortality predictive power of high ESR was similar in the subgroup of 135 patients treated with an ACE inhibitor (ESR in mm/h: p < 0.0001, ESR ≤ 25 mm/h: p < 0.0001) and in the group of 101 patients with NYHA class III and IV (ESR in mm/h: p = 0.0002, ESR ≤ 25 mm/h: p < 0.0001). Also, in the subgroup of 58 patients in NYHA I/II, a high ESR related to a poor prognosis (p < 0.0005).
We categorized patients according to whether they were taking a high, medium or low dose of an ACE inhibitor (at the time of assessment), but we found no significant correlation between ACE inhibitor dose and ESR (data not shown).
This study shows that in CHF patients, the ESR is positively correlated with plasma levels of inflammatory cytokines. An increased ESR is a strong predictor of impaired survival in patients with CHF, independent of established risk factors such as NYHA class, LVEF and peak VO2. This study shows for the first time that, in patients with a well-known and frequent chronic disease, it is possible that a parameter that was previously thought to relate to a better prognosis can (within 10 years) become a predictor of impaired survival.
In this study, only 16% of patients had an ESR < 5 mm/h, which is similar to what was reported by Haber and colleagues (3) (10% of patients in their study had an ESR < 5 mm/h). Therefore, this tends to support the hypothesis that the ESR is high in CHF. However, the important difference between Haber’s study and the present one is in the prognostic value of the ESR in CHF. Our results are in contrast to the study of Haber et al. (3) where high ESR levels (>25 mm/h, median in their study) indicated better survival in 242 patients with CHF. These patients were nonedematous, in NYHA class III and IV, and, at baseline, these patients had been treated with diuretics and digitalis. Only during follow-up (i.e., after the initial ESR assessment) were patients treated with ACE inhibitors (n = 126), other vasodilators (n = 106) or amrinone (n = 3). Haber et al. (3) also demonstrated for 35 patients (15% of their total study group) that after one to three months of this new therapy, patients whose ESR increased showed improvement in both cardiac performance and functional status. These results were in agreement with Paul Wood’s (6) findings in 1936, when he reported on 22 patients with heart failure and found that the ESR decreased as heart failure worsened and rose again when the condition improved.
The results of our investigations apparently contradict those of the previous studies. We have no reason to doubt that both the older studies and ours correctly describe the role of the ESR in heart failure. If that is so, there must be a fundamental difference between the studies to explain the diametrically opposed results. The technique for measuring the ESR has not changed in the past few decades. In Haber’s study (3) as well as in ours, nonedematous CHF patients were studied. The major difference between the study populations is treatment with ACE inhibitors. In our study the majority of patients were taking ACE inhibitors (85%), whereas at initial investigation none of the patients studied by Haber et al. (3) and Wood (6) were on this therapy.
The role of ACE inhibitors
What may be the importance of ACE inhibitors causing a fundamental change in the relationship between the ESR and survival in CHF? Changes in ESR are part of body’s acute phase response to illness (4). It could be hypothesized that, before the advent of ACE inhibitors, patients with CHF were compromised in their response to the disease. Independently of symptomatology (i.e., NYHA class) and ventricular function (i.e., LVEF) patients with a high ESR (and without an ACE inhibitor) may have represented those patients who were able to respond adequately to metabolic and immunological abnormalities, and, thus, these patients had a more favorable outcome. Patients with low ESR were compromised in their responsiveness and died earlier. When new treatment was started, measures of heart failure status improved, the response to metabolic and immunological abnormalities improved, and ESR was then found to increase proportionally. Today most patients are on long-term ACE inhibitor therapy, which may result in an intact responsiveness and a different meaning of a high or low ESR. When the metabolic/immunological CHF status is normal (again, independently of NYHA class and LVEF), no stress is present, ESR is low, and the prognosis is favorable. When the CHF status is poor, metabolic/immunological abnormalities are present, ESR is high, and the prognosis is impaired.
Inflammatory cytokines and the ESR
In support of the above hypothesis, we found that ESR is significantly and positively correlated with markers of inflammatory immune activation. These correlations, and the prognostic results for ESR, support the hypothesis that inflammatory cytokines (particularly TNF receptors) in their own right are strongly related to impaired prognosis in CHF. There is some preliminary evidence for the latter (7).
The ESR is not only a marker of inflammation but also of infectious processes. We have hypothesized that bacterial lipopolysaccharide may play a role in the immune activation of CHF patients (8), and it has been shown that increased lipopolysaccharide-inducible TNF alpha production of peripheral monocytes (9) and cardiomyocytes (10) can be observed in a proportion of CHF patients. Angiotensin-converting enzyme inhibitors have been found to be able to suppress lipopolysaccharide-induced production of TNF alpha in vitro and in vivo (11). Metabolic and immunological abnormalities are strongest in patients with cardiac cachexia (12). If an ACE inhibitor makes a patient more tolerant to pathophysiological changes, this may also prevent the development of cachexia in CHF. There is evidence for the latter (13).
Our model of impaired metabolic and immunological responsiveness in CHF patients somewhat resembles the picture of immune incompetence seen in a number of patients with bacterial sepsis. In sepsis, depending on the precise time point of investigation or initiation of treatment, high, and, in other cases, low levels of inflammatory cytokines may predict poor outcome (14), and immune system stimulatory treatment (with interferon-gamma in patients with an exhausted or incompetent immune system ) or immune system inhibitory treatment (for instance with TNF antibodies in patients with high cytokine levels ) may improve outcome. Nevertheless, it is difficult to test our hypothesis prospectively. Today, if CHF patients are not treated with ACE inhibitors, they may receive angiotensin 2 receptor blockers. An increasing number of CHF patients also receive beta-adrenergic blocking agent treatment that has been proven to exert survival benefit (17), and these new treatments may have similar metabolic effects to those of ACE inhibitors. Withholding these treatments is not possible.
Angiotensin-converting enzyme inhibitor dose was not related to ESR. We appreciate that there is no ideal method to compare ACE inhibitor doses as various ACE inhibitors have differing pharmacological properties. We were not able to analyze the mean drug dose over time. A comparison between patients taking and not taking ACE inhibitors is not feasible due to the low number of patients in this study not on ACE inhibitors (15%). Fourteen of the 24 patients not on an ACE inhibitor at the time of investigation had previously been on an ACE inhibitor for at least three months.
The precise mechanism whereby ACE inhibitors appear to change the relationship between ESR and patients’ survival are not known. Our hypothesis may be one explanation. Alternatively, CHF patients may now live longer (on ACE inhibitors). The somewhat worse general survival and higher median ESR in the previous study (3) may indicate a moribund population with more severely impaired liver function. The introduction of ACE inhibitor therapy in the previous study may have transiently allowed improved liver function with an initial increase in inflammatory mediators, hence, the increased ESR as patients began to improve clinically.
In many chronic diseases, new drugs are developed as a result of advances in basic and clinical research. Some of them may dramatically alter the pathophysiology of the disease and may lead to changes in the prognostic value of “well established” survival indicators. Doctors, scientists and other health care professionals need to be aware of this phenomenon when assessing the value of conventional prognostic indicators. We believe that this phenomenon could also be relevant to cancer, infectious diseases and other chronic conditions where novel treatments are developed. To our knowledge this has never been shown before, and, therefore, this finding may be of major significance.
This study shows that in heart failure, as is the case in many other diseases, a high ESR is a bad prognostic sign. The prognostic value of the ESR in CHF patients is independent of patients’ symptomatology and ventricular function. These results are in contrast to previous work. We hypothesize that today’s treatment with ACE inhibitors may improve patients’ responsiveness to metabolic and immunological abnormalities. In a chronic disease an “established” prognosticator can potentially, on the background of new treatment, completely change its clinical information value to predict the opposite outcome. This could be relevant for other “long-established” survival markers in heart failure as well as other diseases.
☆ Dr. Sharma and the Department of Cardiac Medicine are supported by the British Heart Foundation. Dr. Rauchhaus is a fellow of the European Commission, Brussels and has been supported by the Deutsche Herzstiftung, Frankfurt, Germany. Dr. Coats is supported by the Viscount Royston Trust Fund. Dr. Anker has been supported by the Ernst and Bertha Grimmke Stiftung, Düsseldorf, Germany. Dr. Anker is holding a postgraduate fellowship of the Max-Delbrück-Centrum for Molecular Medicine, Berlin, Germany.
- angiotensin-converting enzyme
- chronic heart failure
- confidence interval
- erythrocyte sedimentation rate
- interleukin 6
- left ventricular ejection fraction
- New York Heart Association
- hazard ratio
- soluble tumor necrosis factor receptor
- tumor necrosis factor
- oxygen consumption
- Received October 20, 1999.
- Revision received February 1, 2000.
- Accepted March 30, 2000.
- American College of Cardiology
- Fahraeus R
- ↵Anonymous. Not the ESR again. Lancet 1991;337:949.
- Anker S.D,
- Clark A.L,
- Kemp M,
- et al.
- Wood P
- ↵Deswal A, Petersen NJ, White BG, Mann DL. Cytokines as predictors of mortality in patients with advanced heart failure (abstr). Circulation 1999;100 Suppl:I–246.
- Wagner D.R,
- McTiernan C,
- Sanders V.J,
- Feldman A.M
- ↵Anker SD, Negassa A, Coats AJS, Poole-Wilson PA, Yusuf S. Weight loss in chronic heart failure and the impact of treatment with ACE inhibitors—results from the SOLVD treatment trial (abstr). Circulation 1999;100 Suppl:I–781.