Higher C‐Reactive Protein Predicts Worse Prognosis in Acute Heart Failure Only in Noninfected Patients

Abstract

Background:

The prognostic role of C‐reactive protein (CRP) in acute heart failure (HF) is not fully understood, and the impact of an infectious process in its risk‐stratification power was not previously evaluated.

Hypothesis:

As CRP is an inflammatory marker, its prognostic value in acute HF is probably different in patients with and without concurrent infection.

Methods:

We recruited patients admitted to our hospital due to acute HF from October 2006 to October 2007. All patients were given treatment at the discretion of the attending physician. Serum CRP was measured at discharge in 225 patients. We followed patients for 3 months after discharge to assess occurrence of all‐cause death or readmission due to HF. Infection was defined according to diagnoses registered on the discharge record. Patients were classified according to CRP tertiles, in the entire sample and in groups according to infection occurrence.

Results:

An infectious condition occurred in 109 patients (first and second CRP tertiles: 8.8 and 27.4 mg/L, respectively). No infection was detected in 116 patients (5.0 and 12.3 mg/L, respectively). In the group with infection, CRP was not a good predictor of adverse outcome. In the noninfected group, the hazard ratio of those with CRP > 12.3 mg/L was 2.46 (95% confidence interval: 1.29–4.70) in comparison with those with lower CRP. Adjusted hazard ratio for ischemic heart disease and diabetes was 2.03 (95% confidence interval: 1.06–3.91).

Conclusions:

CRP had no prognostic value in acute HF patients with an infectious complication. Noninfected patients with higher CRP at discharge had worse prognosis. Copyright © 2010 Wiley Periodicals, Inc.

This study was supported by a grant from FCT: PIC/IC/82773/2007. The authors have no other funding, financial relationships, or conflicts of interest to disclose.

Introduction

In patients with chronic heart failure (HF), the prognostic significance of C‐reactive protein (CRP) is well known.1, 2, 3, 4, 5 Several studies have demonstrated that levels of CRP are elevated in (all‐cause) chronic HF and that increasing levels are associated with increasing morbidity and mortality in ischemic6 and non‐ischemic7, 8, 9, 10 etiologies.

The rationale for this association rests in the inflammatory hypothesis, according to which the toxic effects of endogenous cytokine cascades on the heart and other organs play an important role in the pathophysiology of HF, contributing to its self‐perpetuation.11 Many pro‐inflammatory markers, such as interleukin 6 and tumor necrosis factor α, are elevated in HF. CRP is currently the most suitable marker to assess inflammation and the most widely used in clinical practice.12,13

In the acute HF setting, the usefulness of CRP as a risk marker is challenged by the possible confounding effect of a concomitant infection, frequently underlying decompensation. The prognostic value of CRP in acute HF is still unknown. Previous studies suggested the association of an elevated CRP value with higher morbidity and mortality.14, 15, 16, 17 In these studies, no comparison was made between patients considered infection‐free and those with an infectious complication.

Our aim was to study the prognostic value of CRP after an episode of acute HF according to concurrent infection.

Methods

We recruited 225 patients admitted to our internal medicine department between October 2006 and October 2007 with the diagnosis of acute HF, whether de novo or decompensation of chronic HF. This recruitment corresponded to a registry of acute HF performed at our department with the main objective of identifying prognostic predictors in acute HF. All patients were given treatment at the discretion of the attending physician. Time of discharge was also a decision of the attending physician. In all patients a fasting venous blood sample was obtained between 7 am and 8 am at admission and on discharge day.

Patients were followed for 3 months to assess the occurrence of all‐cause death or admission due to worsening HF. Follow‐up was made by consulting hospital registries of each patient and by telephone contact. No patient was lost during follow‐up.

HF was defined according to the European Society of Cardiology criteria.18 Arterial hypertension was defined as the presence of previous diagnosis and record of antihypertensive pharmacological treatment. Diabetes was defined as either a history of diabetes or the current prescription of either an oral hypoglycemic agent or insulin. Ischemic heart disease was defined as either history of angina, history or electrocardiographic evidence of ischemia, or significant coronary heart disease (CHD) confirmed by coronary angiography. Chronic obstructive pulmonary disease (COPD) was defined as either self‐reported history of medical diagnosis of COPD or by spirometry according to the Global Initiative for Chronic Obstructive Lung Disease (GOLD) criteria.19 Anemia was considered when hemoglobin level was below 13 g/dL in men or 12 g/dL in women. Creatinine clearance was estimated using the Cockroft‐Gault equation.

An infectious condition was considered according to the discharge diagnosis list as judged by the attending physician. We included both infectious conditions that precipitated the current HF decompensation and infections diagnosed during hospitalization.

CRP was measured by use of an immunoturbidimetric assay (Olympus AU640 Olympus America, Inc., Center Valley, PA). The detection limit was 0.20 mg/L; the coefficient of variation was 4.9% for a mean CRP measurement of 0.2 mg/L, 1.7% for a mean CRP of 59 mg/L, and 1.6% for a mean CRP of 147 mg/L. All patients provided written informed consent to participate in the study, and the study protocol conforms to the ethical guidelines of the Declaration of Helsinki. The study was approved by the local ethics committee.

Statistical Analysis

Continuous variables are presented as mean (SD) or median (interquartile range) if non‐normally distributed; categorical variables are presented as counts and proportions.

Comparisons between groups of patients were made by use of an χ² test for categorical variables, independent samples t test for normally distributed continuous variables, and Mann‐Whitney U test when the distribution was skewed.

A stratified analysis according to occurrence of infection during the index hospitalization was performed. In each group, patients were classified according to tertiles of CRP. The occurrence of outcome was compared between patients with CRP increase and patients with CRP decrease during hospitalization.

We used Kaplan‐Meier cumulative survival curves to display hospitalization‐free survival according to tertiles of CRP in the infected and noninfected groups, as well as according to CRP tertiles in the entire sample. We used univariate Cox regression analysis to quantify the association of independent variables with the outcome. A multivariate Cox regression model was used to determine the independent prognostic value of CRP. Potential confounders were defined by their association with serum CRP and the outcome. The confounding effect of each of these variables was confirmed if the prognostic value of CRP was changed when adjusting for them. Only variables whose confounding effect was confirmed were kept in the final model.

All the analyses were conducted using SPSS 13.0 (SPSS Inc., Chicago, IL). A P value of < 0.05 was considered to be statistically significant.

Results

Among the 225 patients with CRP measurement at discharge, 116 had no clinically detected infection and 109 had an infectious condition at admission or during hospital stay. Infections were mainly respiratory or from the urinary tract, although skin and gastrointestinal tract infections were also detected.

Baseline characteristics of the study sample and comparison between infected and noninfected groups are shown in Table 1. All laboratory parameters presented, including b‐type natriuretic peptide (BNP) and troponin I, were measured at discharge. In addition to significant differences concerning CRP levels (higher levels in infected patients), infected patients were older and more often had known chronic HF and concomitant COPD (Table 1).

Table 1.

Comparison Between Infected and Noninfected Patients

	All Patients (n = 225)	Infected Patients (n = 109)	Noninfected Patients (n = 116)	P Value
Clinical characteristics, n (%)
Age (y), median (IQR)	75 (64–81)	78 (70–82)	71 (59–79)	<0.001
Male sex	144 (64.0)	64 (58.7)	80 (69.0)	0.14
Chronic HF	186 (82.7)	97 (89)	89 (76.7)	0.024
Acute lung edema	53 (23.6)	26 (23.9)	27 (23.3)	1.00
NYHA class at admission
II	6 (2.7)	3 (2.8)	3 (2.6)
III	94 (42.2)	39 (36.1)	55 (47.8)	0.12
IV	123 (55.2)	66 (61.1)	57 (49.6)
Ischemic heart disease	114 (50.7)	58 (53.2)	56 (48.3)	0.54
AF	107 (47.6)	55 (50.5)	52 (44.8)	0.48
Arterial HT	145 (64.7)	75 (68.8)	70 (60.9)	0.27
DM	95 (42.2)	51 (46.8)	44 (37.9)	0.20
COPD	59 (26.2)	40 (36.7)	19 (16.4)	0.001
Anemia	97 (43.1)	51 (46.8)	46 (39.7)	0.34
LVF
Normal	48 (24.4)	25 (26.6)	23 (22.3)
Mild dysfunction	17 (8.6)	8 (8.5)	9 (8.7)
Moderate dysfunction	40 (20.3)	17 (18.1)	23 (22.3)
Severe dysfunction	92 (46.7)	44 (46.8)	48 (46.6)	0.65
Medications in use, n (%)
Aspirin	91 (40.8)	52 (48.1)	39 (33.9)	0.04
β‐Blocker	94 (42.3)	49 (45.8)	45 (39.1)	0.38
ACEI or ARB	120 (53.8)	55 (50.9)	65 (56.6)	0.48
Spironolactone	28 (12.6)	15 (13.9)	13 (11.3)	0.70
Statin	108 (48.0)	59 (54.1)	49 (42.2)	0.11
Laboratory parameters at discharge
Hemoglobin, g/dL, mean (SD)	12.3 (2.4)	12.3 (2.3)	12.3 (2.5)	0.90
CCl, mL/min, median (IQR)	47.4 (35.7–65.0)	47.9 (35.0–62.0)	47.2 (37.3–70.1)	0.61
Tn I, mg/L, median (IQR)	0.056 (0.03–0.11)	0.06 (0.04–0.117)	0.051 (0.03–0.109)	0.20
BNP, pg/mL) median (IQR)	712 (285–1444)	706 (339–1468)	722 (243–1339)	0.27
Sodium, mEq/L, median (IQR)	138 (136–141)	138 (136–140)	138 (136–141)	0.32
CRP, mg/L, median (IQR)	10.8 (5.8–23.1)	14.3 (7.4–36.6)	8.0 (4.2–14.3)	<0.001

Abbreviations: ACEI, angiotensin‐converting enzyme inhibitor; AF, atrial fibrillation; ARB, angiotensin II receptor blocker; BNP, b‐type natriuretic peptide; CCl, creatinine clearance; COPD, chronic obstructive pulmonary disease; CRP, C‐reactive protein; DM, diabetes mellitus; HF, heart failure; HT, hypertension; IQR, interquartile range; LVF, left ventricular function; NYHA, New York Heart Association; Tn I, troponin I

Hospitalization or death within 3 months occurred in 77 out of the 225 patients, 40 (36.7%) in infected and 37 (31.9%) in noninfected patients (P = 0.54). In infected patients, the hazard ratio (HR) of death or readmission per mg/L of discharge CRP, analyzed as a continuous variable, was 1.00 (95% confidence interval [CI]: 1.00–1.01, P = 0.49) against 1.02 in the noninfected patients (95% CI: 1.00–1.03, P = 0.02). In the 116 noninfected patients, those with CRP > 12.3 mg/L (upper third) had an HR for an adverse outcome of 2.46 (95% CI: 1.29–4.70, P = 0.006) when compared with those patients with CRP < 12.3 mg/L. Table 2 depicts the results of univariate Cox regression analysis in the noninfected group.

Table 2.

Univariate Analysis of Potential Prognostic Determinants of Adverse Outcome

	HR (95% CI)	P Value
Clinical characteristics
Age (per year)	1.01 (0.99–1.04)	0.41
Male sex	1.24 (0.60–2.55)	0.57
Chronic HF	4.08 (1.25–13.29)	0.02
Acute lung edema	1.35 (0.65–2.79)	0.42
NYHA class at admission
II	1
III	0.65 (0.09–4.92)	0.68
IV	0.93 (0.12–6.94)	0.94
Ischemic heart disease	2.68 (1.34–5.33)	0.005
AF	0.53 (0.27–1.06)	0.07
Arterial HT	1.15 (0.58–2.28)	0.68
DM	1.62 (1.17–2.25)	0.004
COPD	0.72 (0.29–1.77)	0.47
Anemia	2.31 (1.20–4.43)	0.01
LVSD (vs preserved systolic function)	1.86 (0.72–4.82)	0.20
Medications in use
Aspirin	2.42 (1.27–4.62)	0.007
β‐Blocker	1.66 (0.87–3.16)	0.12
ACEI or ARB	2.17 (1.07–4.39)	0.032
Spironolactone	1.27 (0.50–3.27)	0.62
Statin	2.16 (1.13–4.15)	0.020
Laboratory parameters at discharge
CCl ≤30 mL/min	1.39 (0.61–3.18)	0.44
Tn I, per mg/L	0.94 (0.72–1.23)	0.66
BNP, per 100 pg/mL	1.06 (1.03–1.08)	<0.001
Sodium, per mEq/L	1.00 (0.93–1.08)	0.91
CRP >12.3 mg/L	2.46 (1.29–4.70)	0.006

Abbreviations: ACEI, angiotensin‐converting enzyme inhibitor; AF, atrial fibrillation; ARB, angiotensin II receptor blocker; BNP, b‐type natriuretic peptide; CCl, creatinine clearance; CI, confidence interval; COPD, chronic obstructive pulmonary disease; CRP, C‐reactive protein; DM, diabetes mellitus; HF, heart failure; HR, hazard ratio; HT, hypertension; LVSD, left ventricular systolic dysfunction; NYHA: New York Heart Association; Tn I, troponin I

When adjusting for variables found to have prognostic value in univariate analysis, one at a time, a CRP discharge value > 12.3 mg/L maintained a significant association with the outcome, even when adjustment was made for BNP (Table 3). The final multivariate model is displayed in Table 4.

Table 3.

Effect of CRP at Discharge on Risk of Death or Readmission, Crude and Adjusted for Each Covariate

	HR (95% CI) for CRP at Discharge >12.3 mg/L
Crude	2.46 (1.29–4.70)
Adjusted for:
Chronic HF	2.31 (1.21–4.41)
Ischemic HF	2.28 (1.19–4.36)
DM	2.14 (1.11–4.12)
Anemia	2.34 (1.22–4.46)
Aspirin use	2.36 (1.23–4.52)
Statin use	2.49 (1.31–4.76)
ACEI or ARB use	2.49 (1.30–4.76)
BNP, per 100 pg/mL	2.38 (1.24–4.56)

Abbreviations: ACEI, angiotensin‐converting enzyme inhibitor; ARB, angiotensin II receptor blocker; BNP, b‐type natriuretic peptide; CI, confidence interval; CRP, C‐reactive protein; DM, diabetes mellitus; HF, heart failure; HR, hazard ratio

Table 4.

Multivariate Cox Regression Model for the Association of Hospitalization‐Free Survival With CRP at Discharge, Adjusting for Other Prognostic Factors Found to Be Confounding Factors

	HR (95% CI)
CRP >12.3 mg/L	2.03 (1.06–3.91)
Ischemic heart disease	2.35 (1.18–4.71)
DM	1.47 (1.06–2.06)

Abbreviations: CI, confidence interval; CRP, C‐reactive protein; DM, diabetes mellitus; HR, hazard ratio

Figure 1 represents the Kaplan‐Meier cumulative survival curves according to CRP tertiles in infected (8.8 and 27.4 mg/L) and noninfected (5.0 and 12.3 mg/L) patients, as well as in the entire group of patients (6.7 and 17.3 mg/L). In the entire sample, those in the lowest CRP third (<6.7 mg/L) had a significantly better prognosis (P = 0.03). When patients were stratified according to infection, this association of CRP with outcome was not reproduced in both groups. In the absence of infection, hospitalization‐free survival was significantly reduced in patients with CRP > 12.3 m/L (upper third). No significant differences existed in hospitalization‐free survival according to CRP tertiles in infected patients.

Figure 1.

Kaplan‐Meier survival curves according to CRP tertiles at discharge in noninfected patients (A) and infected patients (B), as well as in the entire group of patients (C). Hospitalization‐free survival was significantly reduced in patients with CRP > 12.3 mg/L (upper third) in the noninfected group (P = 0.01); CRP had no impact on survival in infected patients (P = 0.78). Hospitalization‐free survival in the entire group was significantly better in patients with discharge CRP < 6.7 mg/L. Abbreviations: CRP, C‐reactive protein

In the noninfected group, CRP at discharge was compared with admission CRP. There was a decrease in 72 (61.7%) patients and an increase in 35 (36.9%). All‐cause death or readmission for HF worsening occurred in 40% of the patients whose CRP value increased (14 out of 35) compared with 26.4% (19 out of 72) of those whose CRP decreased (P = 0.23). Although not significant, there was a trend for patients in whom CRP value decreased from admission to discharge to have a better outcome than those in whom the CRP value increased.

Discussion

Our results expand previous knowledge on the prognostic interest of CRP in acute HF. We showed that the value of CRP is specially significant in acute HF patients without concomitant infection.

Alonso‐Martinez and co‐workers14 evaluated 76 patients admitted to the hospital for acute HF and reported that patients with higher CRP at admission had higher readmission rates and shorter periods before readmission. In a larger multicenter study15 including 620 patients, investigators used median CRP at admission as the cutoff and found an almost doubled risk of death during a 1‐year follow‐up period in patients with CRP > 10 mg/L. Two other groups16,17 reported the same association of increased inflammatory activation and worse prognosis. In all these studies except one,17 infected patients were not excluded and, even when adjusting for infection, it is still possible that infection has determined, at least in part, the CRP values. Also, as only admission CRP measurements were considered, the possibility of a yet‐to‐be‐manifested infectious disease could not be discarded.

In the study sample, 34.2% of the patients were hospitalized or died within a 3‐month period after an admission for acute HF. This rate illustrates the very‐high‐risk population we are studying, composed of elderly patients and those with markedly compromised left ventricular function. Previous reports observed similar results in acute HF patients. Our group reported a 6‐month risk of death or readmission of 42.9%,20 and more recently Ross et alreported an all‐cause readmission rate of 20% within 30 days.21

In the entire sample, a lower CRP (<6.7 mg/L) was associated with better hospitalization‐free survival, reinforcing the idea that stronger inflammatory activation carries an ominous prognosis. This observation however, was only reproduced in noninfected patients. In infected patients, CRP at discharge had no prognostic value; but in patients considered infection‐free, a CRP > 12.3 mg/L (upper third) was associated with double the risk of death or readmission due to HF within 3 months. The cutoff value shown to have prognostic significance—12.3 mg/L—is higher than the 3 mg/L proposed for the general population and patients with stable CHD.22 This higher cutoff, although arbitrary and grown out of this specific sample, is not surprising, if we consider the fact that this is a group of patients with American College of Cardiology/American Heart Association stage C heart failure and in an acute setting—that is, patients with an exacerbation of a condition with known inflammatory activation. The increase in risk was nonlinear, with extra risk being exclusive of patients in the upper third of CRP.

The prognostic value of CRP was independent of factors known to interfere with CRP: ischemic heart disease and diabetes, aspirin and statins (drugs with anti‐inflammatory properties able to reduce CRP levels),23,24 and chronic HF (pro‐inflammatory activation).

Although no known pathophysiological mechanism linking CRP and BNP has been described in HF, CRP retained its prognostic value even when adjusted for BNP concentration (HR: 2.28, 95% CI: 1.34–4.58), a well‐established and powerful prognostic predictor.

To the best of our knowledge, this is the first study that considers separately infected and noninfected patients in acute HF. This is important, because infection is a recognized and relevant determinant of HF decompensation and simultaneously a factor known to determine increases in inflammatory markers. By analyzing the patients separately, we overcame that potential bias. In addition to higher serum CRP, infected patients were older and more often had known established HF and COPD. They were otherwise similar to infection‐free patients, including in death or readmission rate within 3 months.

Study Limitations

Important limitations of our study are the small sample size and short follow‐up period. Another setback comes from the definition of infection. We relied on the clinical evaluation of the attending physician and considered the presence of infection anywhere in the discharge diagnosis list. Criteria were probably different between different physicians, but in general there was information in the discharge note to support this diagnosis (the most common situations were fever, new‐onset cough with sputum, and elevated leukocyte count leading to the use of antibiotics). None of the patients had obvious noninfectious inflammatory conditions such as gout or autoimmune diseases. For the general population and stable CHD patients, an arbitrary cutoff value of 10 mg/L is accepted as the value beyond which an inflammatory or infectious disorder has to be excluded before assuming the CRP measurement to be associated with atherosclerosis and low‐grade inflammation.22 No such value has been suggested for chronic or acute HF, although this should be higher if we consider the increased inflammatory activation in these settings.

Conclusion

CRP has no role in the prognostic evaluation of acute HF patients in the context of an infectious complication. Noninfected patients with CRP > 12.3 mg/L had approximately double the risk of death or readmission for acute HF, independent of other well‐established predictors of prognosis. Further studies are needed to clarify the best CRP cutoff value with prognostic impact in acute HF.