OUP user menu

★ editor's choice ★

Prevalence, incidence, and prognostic value of anaemia in patients after an acute myocardial infarction: data from the OPTIMAAL trial

Stefan D. Anker, Adriaan Voors, Darlington Okonko, Andrew L. Clark, Margaret K. James, Stephan von Haehling, John Kjekshus, Piotr Ponikowski, Kenneth Dickstein
DOI: http://dx.doi.org/10.1093/eurheartj/ehp116 1331-1339 First published online: 21 April 2009

Abstract

Aims The prevalence, incidence, and prognostic value of anaemia in patients with an acute myocardial infarction (AMI) complicated by heart failure is unclear.

Methods and results We analysed the relationship between haemoglobin (Hb) and outcome in 5010 patients with AMI complicated by heart failure in the OPTIMAAL study. In 3921 patients, follow-up Hb levels were available at 365 (±90) days. In a subgroup of 224 patients, iron-related haematinics were assessed at baseline and during follow-up. At baseline, mean Hb was 12.6 ± 1.3 g/dL in women and 13.7 ± 1.4 g/dL in men. Hb < 11.5 g/dL was found in 9.3% of patients (women: 18.2%, men: 5.8%). Lower haemoglobin at baseline was clearly associated with female gender and the presence of diabetes, higher age and Killip class, lower body mass index, systolic blood pressure, total cholesterol, and the absence of current smoking (all P < 0.05). Higher Hb [per one standard deviation (SD)] related to lower mortality [adjusted hazard ratios (HR) 0.88; 95% confidence interval (CI) 0.83–0.93], CHF hospitalizations [HR 0.85 (0.77–0.93)], and all-cause hospitalizations [HR 0.96 (0.92–0.99), all P < 0.05]. In patients without anaemia at baseline, the anaemia incidence after 1 year of follow-up was 10.1% in women and 10.0% in men. Of patients with anaemia at baseline, 65% did not have anaemia at 12 months and 46% did not have anaemia at any time during follow-up (median 3.0 years, inter-quartile range, Q1–Q3 = 2.7–3.3 years). At 12 months, an increase in Hb (per SD) was related to lower mortality [HR 0.73 (0.63–0.85; P < 0.0001)] independent of baseline Hb and other clinical characteristics.

Conclusion In patients with complicated AMIs, anaemia on admission and/or reductions in haemoglobin during follow-up are independent risk factors for mortality and hospitalization. Studies are warranted to determine whether correcting anaemia after a complicated AMI improves outcome.

Keywords
  • Myocardial infarction
  • Anaemia
  • Iron deficiency
  • Erythropoietin
  • Mortality
  • Hospitalization
  • OPTIMAAL study

Introduction

Anaemia is frequently found in patients with an acute myocardial infarction (AMI), although the prevalence varies widely between 6.4 and 43%, depending on the definition and the patient characteristics.18 Low haemoglobin (Hb) levels augment myocardial ischaemia and adverse ventricular remodelling in experimental animal studies, particularly in the presence of coronary arterial stenoses.910 In humans, lower haemoglobin levels have been consistently associated with adverse cardiovascular outcomes in the general population,11 in patients undergoing a percutaneous coronary intervention12 or coronary bypass surgery,13 and in particular in patients with chronic heart failure.1418

In addition, several studies have demonstrated that in patients with an AMI, lower haemoglobin levels at or during admission were associated with increased morbidity and mortality.18,19,20

In general, these studies were mostly retrospective, focusing mainly on mortality, and were conducted in heterogeneous cohorts. Only one study specifically aimed to determine the prognostic value of low haemoglobin levels in patients with an AMI who developed heart failure.5 However, patients in this study were enrolled between 1990 and 1992, and treatment for AMI has substantially changed since then. Also, the cause of anaemia was unknown. Moreover, no data are available on the evolution of Hb levels in the months and years following AMI, and how such temporal trends relate to their subsequent outcome.

The Optimal Trial in Myocardial Infarction with the Angiotensin II Antagonist Losartan (OPTIMAAL) randomized patients with AMI and de novo signs and symptoms of heart failure to receive captopril or losartan within 10 days of symptom onset.21 In this unique cohort, we set out to assess (i) the association of baseline Hb levels to hospitalization and survival, (ii) the temporal trends in Hb levels, (iii) the association of changes in Hb levels over time to survival; and (iv) whether treatment allocation affected the relationship between Hb and outcomes. Our primary hypothesis was that anaemia on admission, and/or during follow-up, would predict an enhanced risk of mortality in incident cases of heart failure after AMI.

Methods

Patient cohort

In OPTIMAAL, 5477 patients with AMI complicated by the development of signs or symptoms of heart failure were randomized to receive either captopril or losartan (protocol 179-09). A past history of myocardial infarction was found in 18% of the patients, but only 6% had a past history of heart failure. None of the subjects was receiving either ACE-inhibitor or angiotensin receptor antagonist at randomization. There was no statistically significant difference in outcome between the two groups.21

Haemoglobin values

At baseline, data for haemoglobin were available for 5010 patients (91.5% of all patients) and these form the study group in the present paper. We defined anaemia as a haemoglobin level of <12 g/dL for women and <13 g/dL for men. For summary tables, patients were categorized into four a priori defined groups according to baseline haemoglobin (g/dL): <11.5, 11.5 to <12.5, 12.5 to <16.0, ≥16.0. In 3921 of these patients, haemoglobin was also available at 12 months follow-up. In this subset of patients, we studied the effect of change in haemoglobin on the subsequent outcome. This analysis necessarily excludes all patients reaching the outcome during the first year of follow-up. Patients were divided into five groups using quintiles for change in haemoglobin at 1 year.

Iron-related haematinics

In a prospectively designed sub-study of the main OPTIMAAL trial comprising 234 consecutive patients from six centres that was designed to analyse plasma/serum levels of inflammatory/anti-inflammatory mediators, iron-related haematinics could be measured in 224 patients. Blood samples were taken at baseline, at 1 month, 1 year, and 2 years after randomization to evaluate standard iron-related haematinics, such as transferrin, soluble transferrin receptor, ferritin, iron, and total iron-binding capacity.

Statistical analyses

All data are presented as mean ± SD. Associations between baseline categorical covariates and baseline haemoglobin groups were assessed using χ2 tests. Kruskal–Wallis tests were used to test differences in continuous baseline covariates and baseline haemoglobin and change from baseline quintile groups. Similar tests were used to assess associations between baseline covariates and change in haemoglobin quintiles at 1 year.

Cox proportional-hazards analysis was performed to assess the association between baseline variables and endpoints which included mortality (all-cause, due to sudden cardiac death, and due to progressive heart failure), hospitalization (all-cause, for CHF, and cardiovascular reasons). Factors which are known to be of prognostic value in heart failure were included in a multivariable model. These factors included: age, sex, randomized treatment group, baseline BMI, estimated glomerular filtration rate (eGFR; simplified MDRD formula), baseline creatinine, baseline uric acid, Killip class, heart rate, systolic blood pressure, total cholesterol, current smoking, history of diabetes, in-hospital beta-blocker, statin, digitalis, nitrate, aspirin, thrombolytic, warfarin, and diuretic use. Hazard ratio (HR) and 95% confidence interval (CI) for risk factors as well as significance levels for χ2 (likelihood ratio test) were calculated from separate models which assessed the impact of haemoglobin as a continuous variable (standardized to one standard deviation) and the impact of the presence of anaemia (defined as haemoglobin <12.0 g/dL for females and <13.0 g/dL for males). The proportional hazards assumption was assessed by evaluating the interaction between these baseline variable and time. The proportional hazards assumption was satisfied for these effects based on the interaction term with time which was not significant (P > 0.05). A quadratic term was added to the model for each baseline variable to assess the linearity assumption. The linearity assumption was met for all continuous variables in the model, with all quadratic terms being non-significant at P = 0.01. Kaplan–Meier cumulative survival plots were constructed to illustrate the risk of death and death or CHF hospitalization observed within the baseline haemoglobin groups. Similarly, a Cox proportional analysis model was constructed to evaluate the impact of 1 year change in haemoglobin (g/dL) on mortality. In an additional model, the effects of change in haemoglobin (g/dL) were parameterized to allow for separate estimation of the effect of an increase vs. a decrease in haemoglobin(g/dL). Hazard ratios and 95% CIs were calculated to estimate the impact of these changes in haemoglobin (g/dL), standardized per one standard deviation. A Kaplan–Meier plot was generated to display the cumulative risk of death observed within each change in haemoglobin quintile group. A P-value <0.05 was considered significant.

In the subset of patents with iron-related haematinics data, multivariable linear regression models were employed to assess the relationship between haemoglobin levels and transferrin, sTfR, Epo, Ferritn, Iron, and TIBC at baseline. Correlations between haemoglobin and these factors were assessed at 1 month, and at 1 and 2 years.

Results

The mean age of the 5010 patients in the present study population was 67.4 ± 9.8 years; 28.4% were female. At baseline, in the entire population, haemoglobin level was 13.4 ± 1.44 g/dL (range 7.8–21.4) and was lower in women (12.6 ± 1.31 g/dL; range 8.7–17.5) than in men (13.7 ± 1.37 g/dL; range 7.8–21.4). A prior history of anaemia was reported in only 10 patients (0.2%). A prior history of gastro-intestinal problems (gastric/duodenic ulcers, oesophagitis, gastritis, duodenitis, hiatus hernia) was reported in 116 patients (2.3%). The only chronic illness that was reported in more than 0.5% of patients was rheumatoid arthritis in 43 patients (0.9%). Other baseline characteristics of patients divided into pre-defined subgroups are demonstrated in Table 1. The patients with lower haemoglobin were more likely to be female, were significantly older, had a lower BMI, eGFR, uric acid, heart rate, systolic blood pressure, total cholesterol, and were more likely to have Killip Class III or IV, and a history of diabetes than patients with higher haemoglobin levels (P < 0.001).

View this table:
Table 1

Baseline characteristics of the study cohort according to a priori defined cut-offs for baseline haemoglobin

Haemoglobin group (g/dL)All patients<11.511.5 to <12.512.5 to <16.0≥16P-value
n4667693626149
Female (%)28.455.649.421.38.1<0.0001
Age67.4 ± 9.871.4 ± 9.469.8 ± 9.566.5 ± 9.764.2 ± 9.6<0.0001
BMI (n = 4819)26.6 ± 3.925.6 ± 3.926.2 ± 4.126.8 ± 3.827.7 ± 4.0<0.0001
Estimated glomerular filtration rate (mL/min/1.73 m2) (n = 4986)66.8 ± 15.961.3 ± 17.864.9 ± 17.067.9 ± 15.265.4 ± 14.8<0.0001
Uric acid (µmol/L) (n = 4986)343 ± 99340 ± 124330 ± 107345 ± 93388 ± 101<0.0001
% in Captopril-treated group49.751.748.549.749.70.753
Killip class<0.0001
 Killip Class I (%)31.726.028.933.034.2
 Killip Class II (%)57.056.057.457.154.4
 Killip Class III or IV (%)11.318.013.89.911.4
Heart rate (b.p.m.) (n = 4992)75.0 ± 14.177.2 ± 14.975.6 ± 13.974.5 ± 14.077.3 ± 13.8<0.0001
Systolic blood pressure (mmHg) (n = 4986)122.8 ± 16.9121.9 ± 19.2122.2 ± 16.1122.9 ± 16.8126.1 ± 16.20.005
Total cholesterol (mmol/L) (n = 4986)5.5 ± 1.15.0 ± 1.25.3 ± 1.15.5 ± 1.15.8 ± 1.1<0.0001
Current smoking (%)33.727.932.934.536.90.027
History of diabetes (%)17.321.220.916.017.40.001
In-hospital beta-blocker (%)76.275.176.376.475.20.920
In-hospital statin (%)29.827.927.630.727.50.226
In-hospital digitalis (%)10.115.710.79.210.70.0006
In-hospital nitrate use (%)78.082.880.477.173.20.005
In-hospital aspirin use (%)94.593.694.994.791.30.268
In-hospital thrombolytic use (%)54.346.153.155.849.00.0005
In-hospital warfarin use (%)7.57.37.97.46.70.944
In-hospital diuretic use (%)62.368.766.360.372.5<0.0001
Endpoints
 Any hospitalization (%)65.264.867.064.867.1
 Cardiovascular hospitalization (%)52.848.154.952.857.7
 CHF hospitalization (%)10.413.114.29.56.0
 Non-cardiovascular hospitalization (%)32.537.635.531.426.2
 Death (%)17.329.023.114.616.8
 Sudden cardiac death (%)6.27.97.45.76.0
 Death due to progressive heart failure (%)3.66.46.22.81.3
  • Data are mean ± SD unless otherwise noted.

During a median follow-up of 3.0 years (interquartile range Q1–Q3 = 2.7–3.3), there were 866 deaths, 523 CHF hospitalizations, and 3268 hospitalizations for any reason. Serious adverse events related to gastro-intestinal problems during follow-up were reported in only five patients (two gastric ulcers, two gastritis, and onw haematemesis). Decreasing baseline haemoglobin was strongly related to death and heart failure hospitalizations in both unadjusted analyses and after adjustment for important clinical (Table 2). For each 1 standard deviation increase in haemoglobin, the risk of death fell 12% (95% CI 7–17%, P < 0.0001). The Kaplan–Meier plots for time to death and time to death or CHF hospitalization for the haemoglobin subgroups are provided in Figures 1 and 2.

Figure 1

Kaplan–Meier plots for time to death according to baseline haemoglobin level in OPTIMAAL. Patients at risk and event rates at 1 and 3 years (with 95%CI) are indicated.

Figure 2

Kaplan–Meier plots for time to death or chronic heart failure hospitalization according to baseline haemoglobin level in OPTIMAAL. Patients at risk and event rates at 1 and 3 years (with 95%CI) are indicated.

View this table:
Table 2

The prognostic impact of baseline haemoglobin as continuous variable [heart rate based on haemoglobin standardized per one standard deviation(SD)], as dichotomized variable for the presence of Hb < 12.0 g/dL (females) and <13 g/dL (males) in the OPTIMAAL study cohort (n = 5010)

Unadjusted resultsAdjusted resultsa
HR (95% CI)P-valueHR (95% CI)P-value
Analyses for haemoglobin as continuous variable
 All-cause death0.76 (0.71–0.82)<0.00010.88 (0.83–0.93)<0.0001
 Cardiovascular hospitalization0.98 (0.95–1.02)0.3910.98 (0.94–1.02)0.357
 CHF hospitalization0.77 (0.71–0.84)<0.00010.85 (0.77–0.93)0.001
 Any hospitalization0.95 (0.92–0.98)0.0020.96 (0.92–0.99)0.024
 All-cause death or CHF hospitalization0.77 (0.73–0.82)<0.00010.86 (0.81–0.92)<0.0001
 Sudden cardiac death0.85 (0.76–0.95)0.0040.86 (0.80–1.03)0.141
 Death due to progressive heart failure0.67 (0.58–0.77)<0.00010.80 (0.69–0.94)0.006
Analysis for presence of anaemia
 All-cause death1.59 (1.38–1.82)<0.00011.35 (1.16–1.56)<0.0001
 Cardiovascular hospitalization1.00 (0.92–1.09)0.9381.01 (0.92–1.11)0.806
 CHF hospitalization1.48 (1.23–1.77)<0.00011.31 (1.08–1.59)0.006
 Any hospitalization1.06 (0.94–1.15)0.1231.04 (0.96–1.13)0.330
 All-cause death or CHF hospitalization1.56 (1.38–1.75)<0.00011.33 (1.17–1.51)<0.0001
 Sudden cardiac death1.31 (1.04–1.66)0.0241.14 (0.89–1.48)0.303
 Death due to progressive heart failure1.88 (1.40–2.53)<0.00011.55 (1.13–2.13)0.006
  • aAdjusted for age, sex, randomized treatment group, baseline BMI, estimated glomerular filtration rate, baseline creatinine, and baseline uric acid, Killip class, heart rate, systolic blood pressure, total cholesterol, current smoking, history of diabetes, in-hospital beta-blocker, statin, digitalis nitrate, aspirin, thrombolytic, warfarin, and diuretic use.

Haemoglobin measures were available at 1 year in 3921 patients (average age at randomization 66.6 ± 9.5 years; 73% male). A total of 459 patients had a fall in haemoglobin of at least 1 g/dL and 1156 patients had an increase of at least 1 g/dL. The quintiles for haemoglobin change (12 month value minus baseline value) corresponded to the following 1 year changes in haemoglobin (g/dL): quintile 1 (−8.8 to −0.7 g/dL), quintile 2 (−0.6 to 0.0 g/dL), quintile 3 (0.1 to 0.6 g/dL), quintile 4 (0.7 to 1.4 g/dL), and quintile 5 (1.5 to 6.7 g/dL). Data for these patients divided by category of haemoglobin change quintiles are shown in Table 3. Because 51% of the patients were admitted to hospital before the 1 year haemoglobin was measured, only results for the primary endpoint (all cause mortality) are shown. The Kaplan–Meier plots for time to death for the haemoglobin change quintile groups are provided in Figure 3.

Figure 3

Kaplan–Meier plots for time to death according to change in haemoglobin from baseline to 12 months in OPTIMAAL. Patients at risk and event rates at 1 and 3 years (with 95%CI) are indicated.

View this table:
Table 3

Baseline characteristics of 3921 patients subgrouped according to change in haemoglobin from baseline to 365 (±90) days

Haemoglobin change quintile12345P-value
n749771784822795
% female26.427.127.026.026.50.988
Age68.4 ± 9.767.3 ± 9.666.8 ± 9.565.6 ± 9.365.1 ± 9.0<0.0001
BMI (n = 3785)26.8 ± 3.926.8 ± 3.926.8 ± 3.926.8 ± 3.926.6 ± 3.80.567
eGFR (mL/min/1.73 m2) (n = 3900)63.1 ± 15.666.8 ± 14.768.1 ± 15.269.5 ± 14.271.1 ± 16.3<0.0001
Uric acid (µmol/L) (n = 3900)373 ± 124340 ± 89329 ± 85328 ± 85317 ± 91<0.0001
% in Captopril treated group47.550.447.749.452.60.242
Killip class<0.0001
 Killip class I (%)26.637.537.834.631.6
 Killip class II (%)60.055.555.455.856.0
 Killip class III or IV (%)13.47.06.99.612.4
Heart rate (b.p.m.) (n = 3906)75.3 ± 14.073.6 ± 14.373.5 ± 13.973.5 ± 13.275.3 ± 13.20.001
Systolic blood pressure (mmHg) (n = 3902)125.1 ± 16.7124.9 ± 17.9122.4 ± 16.8121.0 ± 16.1120.6 ± 16.8<0.0001
Total cholesterol (mmol/L) (n = 3900)5.8 ± 1.25.6 ± 1.15.6 ± 1.05.3 ± 1.05.0 ± 1.0<0.0001
Current smoking (%)29.633.132.836.139.50.001
History of diabetes (%)18.314.115.614.018.00.044
In-hospital beta-blocker (%)69.678.777.479.682.6<0.0001
In-hospital statin (%)30.629.230.632.231.10.772
In-hospital digitalis (%)9.87.66.98.68.80.294
In-hospital nitrate use (%)73.275.875.880.884.9<0.0001
In-hospital aspirin use (%)93.395.595.894.995.10.244
In-hospital thrombolytic use (%)52.655.156.358.461.10.010
In-hospital warfarin use (%)7.27.06.88.48.60.543
In-hospital diuretic use (%)71.060.655.255.659.1<0.0001
Endpoints
 Death (%)10.18.75.14.76.2
 Sudden death (%)2.72.91.71.82.9
 Death due to progressive heart failure2.01.20.61.00.8
  • Data are mean ± SD unless otherwise noted. Patients were divided into five groups based on quintiles of change in haemoglobin at 1 year.

At 12 months, mean change of haemoglobin was 0.36 and 0.44 g/dL in patients on losartan and captopril, respectively (P = 0.065). During this time period, the mean change (±SD) in serum creatinine was 4.3 ± 17.8 µmol/L, corresponding to a mean change in eGFR of −3.1 ± 11.2. The cumulative event rates (95% CI) for development of anaemia (haemoglobin <12 g/dL for women and <13 g/dL for men) in patients without anaemia at baseline at 1, 2, and 3 years follow-up were 10.0 (8.8–11.2)%, 16.3 (14.8–17.8)%, and 21.7 (20,0–23.4)% in men and 10.1 (8.1–12.1)%, 17.8 (15.2–20.4)%, and 23.1 (20.1–26.1)% in women, respectively.

At baseline, 2877 of the 3921 (73.4%) patients were not anaemic (haemoglobin ≥12.0 g/dL for women and ≥13.0 g/dL for men). At 12 months, 258 of the 2877 (9.0%) patients without baseline anaemia had become anaemic. For those 1044 who were anaemic at baseline, 683 (65.4%) were no longer anaemic at the 12 month follow-up.

In multivariable analysis, a change in haemoglobin over 12 months (change defined as 12 month haemoglobin value minus baseline haemoglobin value standardized to one standard deviation) was strongly related to increasing survival [HR 0.73 (0.63–0.85); P < 0.0001] after adjustment for sex, age, randomized treatment group, body mass index, baseline creatinine, baseline eGFR, baseline urate, baseline haemoglobin, Killip class, heart rate, systolic blood pressure, total cholesterol, current smoking, history of diabetes, in-hospital beta-blocker, statin, digitalis, nitrate, aspirin, thrombolytic, warfarin, and diuretic use. When increases and decreases in haemoglobin were separately parameterized in the model, an elevated mortality risk was observed per decrease in haemoglobin (decrease per standard deviation) (HR 1.45, 1.20–1.74; P < 0.0001), and that risk was borderline significant after adjustment for covariates (HR 1.27, CI 1.00–1.60, P = 0.05). The benefit on survival associated with an increase (increase per standard deviation) was not statistically significant (HR 0.85, 0.68–1.06; P = 0.16) in unadjusted analysis, but did reach statistical significance after adjustment for covariates (HR 0.67, 0.51–0.81, P < 0.01).

Iron-related haematinics were assessed in a subgroup of 224 patients. Mean age of this subgroup was 68 ± 10 years, 30% was female, mean BMI was 26 ± 3.7, and mean creatinin was 93 ± 24. So, patients were very similar compared with the overall OPTIMAAL population. Baseline haemoglobin values and iron metabolism were available in 213 patients. Of these 213 patients, 83 (39%) were anaemic. Of the anaemic patients, at baseline, abnormal transferrin values were found in 28 (34%), abnormal soluble transferring receptor in 14 (17%), abnormal ferritin in 30 (36%), abnormal iron in 70 (84%), and abnormal total iron-binding capacity in 12 (14%). In univariable Cox survival analysis, both lower haemoglobin (HR 0.97/g/dL; 95% CI 0.95–1.00; P = 0.023) and higher soluble transferrin receptor (HR 2.14/mg/L; 95% CI 1.22–3.76; P = 0.008) were related to increased mortality. At baseline, after 1 month, 1 year, and 2 years, anaemia was present in 39, 13, 29, and 23% of the patients, respectively. Haemoglobin values and iron-related haematinics over time are presented in Table 4. Multivariable regression analysis demonstrated that at baseline, haemoglobin levels were significantly and independently correlated with (log)erythropoietin (r = −0.40, P < 0.0001), gender (r = 0.25, P = 0.0004), (log)transferrin (r = 0.20, P = 0.0032), (log)iron (r = 0.15, P = 0.027), and (log)ferritin (r = 0.14; P = 0.038). At 1 month, haemoglobin levels were significantly and independently correlated with gender (r = 0.37, P < 0.0001), (log)iron (r = 0.31, P < 0.0001), age (r = −0.27, P < 0.0001), and (log)ferritin (r = 0.18; P = 0.0043). Both at 1 and 2 years, beside age and gender, (log)iron (r = 0.30; P < 0.0001 and r = 0.32; P < 0.0001 at 1 and 2 years, respectively) was the main determinant of haemoglobin levels. Of note, haemoglobin was not independently associated with several inflammatory markers, including high sensitivity C-reactive protein.

View this table:
Table 4

Haemoglobin and haematinics at baseline, and after 1 month, 1 year, and 2 years after randomization in a subgroup of 224 patients

Baseline (n = 213)1 month (n = 193)1 year (n = 190)2 years (n = 189)P-value
Haemoglobin (g/dL)12.9 ± 1413.7 ± 1313.3 ± 1313.5 ± 14
Transferrin (mg/L)212 ± 58223 ± 45243 ± 45234 ± 53<0.0001
sTfR (mg/L)1.28 ± 0.401.33 ± 0.421.29 ± 0.361.29 ± 0.350.0018
Epo (mlU/mL)19.5 ± 20.613.3 ± 13.510.4 ± 4.910.5 ± 4.9<0.0001
Ferritin (ng/mL)255 ± 241219 ± 204151 ± 138145 ± 135<0.0001
Iron (μg/dL)47.4 ± 30.889.6 ± 33.3101 ± 30.297.6 ± 31.9<0.0001
TIBC (μg/dL)336 ± 58333 ± 48337 ± 52300 ± 39<0.0001
  • sTfR, soluble transferrin receptor; EPO, erythropoietin; TIBC, total iron binding capacity.

Significant changes in haematinics occurred over time (Table 4). In particular iron deficiency was restored after 1 year. In this small subgroup, changes in haematinics between baseline and 1 month, and between baseline and 1 year, were not significantly related to survival.

Discussion

The main and novel findings of the present study are that in post-myocardial infarction heart failure patients, both low baseline haemoglobin and a fall in haemoglobin during follow-up appear to be independently associated with an adverse outcome.

Anaemia is frequently found in patients during an AMI. Prevalence varies from 6 to 43%, and is related to the patient characteristics, and the definition that was used.18 In the present study, the prevalence of anaemia at baseline (<12 g/dL women, <13 g/dL men) was 28.0%. From measurements of iron metabolism in a subgroup of patients, it appeared that at baseline, increased erythropoietin, and during follow-up iron deficiency was associated with low haemoglobin levels.

Several studies have already related anaemia to an increased risk of a cardiovascular event. In a study with 14410 subjects without cardiovascular disease at baseline, the presence of anaemia was associated with a 1.4 times increased risk of a cardiovascular event.11 Similar findings were observed in another community study in an elderly population.22 In patients undergoing general surgery, anaemia was related to more peri-operative complications and an increased mortality rate.23 Also in patients with established cardiovascular disease, anaemia is an important predictor of outcome. Both in patients undergoing coronary bypass surgery13 and a percutaneous coronary intervention,12 low haemoglobin levels were associated with increased mortality. Finally, several groups have reported that anaemia is frequently found in chronic and acute heart failure patients, and also related to poor outcome.24,25

One of the main findings of the present study is that lower baseline haemoglobin at baseline during AMI is related to a higher mortality. This has already been described by others. In a retrospective study on 78 974 AMI patients, patients with lower haematocrit values on admission had higher 30 day mortality rates.1 After this, several others have showed that anaemia during an AMI was related to 30 day and 1 year mortality.27 In contrast, Al Falluji et al.8 could not demonstrate that anaemia during admission was independently related to 1 year mortality after an AMI. Only one study described a 2 years adverse outcome in anaemic patients after AMI.4 Moreover, only one smaller study described the adverse prognostic effects of anaemia in patients with left-ventricular dysfunction following AMI.5 So, the present findings that baseline haemoglobin was related to a higher mortality confirms findings by others, although to date this is the largest cohort describing long-term follow-up after AMI with heart failure.

A unique and novel finding of the present study is that the development of anaemia during follow-up after discharge for AMI is also independently related to increased mortality. Aronson et al.2 demonstrated that a drop in haemoglobin during index hospital admission for AMI was also related to increased mortality.

Another unique feature of the present study is that we have assessed iron-related haematinics in a subgroup of patients. Haemoglobin levels were lower at baseline compared with follow-up. Lower haemoglobin levels are frequently found in patients with acute coronary syndromes. The explanation for this has not been fully elucidated. Interestingly, a higher erythropoietin level at baseline was the main determinant associated with lower haemoglobin levels. This probably reflects larger infarcts that are accompanied both by higher erythropoietin and lower haemoglobin levels. Unfortunately, since patients in OPTIMAAL were included days after the onset of AMI, a good index of infarct size is missing. During follow-up, iron deficiency appeared to be the main determinant associated with lower haemoglobin levels. A likely explanation for the association between anaemia and mortality in the current study is the inflammatory reaction in response to tissue injury. However, inflammatory status was not related to haemoglobin levels neither at baseline, nor during follow-up. Haemodilution might also have cause anaemia in the current study, especially during follow-up, since these were patients with heart failure. In particular in heart failure, anaemia is in part caused by haemodilution.26 Finally, the use of blockers of the renin–angiotensin system might have caused anaemia. The change in haemoglobin was slightly larger in patients treated with captopril, which might be caused by inhibition of the breakdown of Ac-SDKP, a strong haematopoiesis inhibitor.27

Since anaemia is a very consistent and strong predictor of mortality in patients with AMI and heart failure, one might assume that correction of anaemia using erythropoiesis-stimulating proteins (such as the recombinant form of Epo–Rh-Epo) might be beneficial. Concerns have been raised to normalize haemoglobin levels with Rh-Epo in patients with chronic kidney disease.2829 However, first results in chronic heart failure patients are promising,3033 and studies on the effects of a single bolus of Rh-Epo in AMI patients are ongoing. Also, several studies have even indicated that erythropoietin might have a direct beneficial effect on cardiac function, independent of correction of anaemia.3435

In conclusion, in this large study in patients with heart failure after AMI, lower baseline haemoglobin was related to higher 2 year mortality. In addition, the development of anaemia during follow-up was also related to an increased mortality, independent of baseline haemoglobin.

Funding

The OPTIMAAL study was supported by an unconditional grant from Merck, Sharp and Dohme Research Laboratories, West Point, PA, USA. The haematinics substudy was separately supported by an unrestricted research grant from Merck, Sharp and Dohme Research Laboratories, West Point, PA, USA. A.A.V. is supported by the Netherlands Heart Foundation (Clinical Established Investigator grant 2006T37).

Conflict of interest: S.D.A. has received research grants and honoraria for consultancy and presentations from Amgen Inc. and Vifor International. He has also received honoraria for presentations from Roche AG. A.A.V. has received research grants from Amgen Inc. and Orthobiotech. M.K.J. is an employee of Merck and owns stock and stock options of Merck. P.P. has received consultancy fees from Amgen Inc. and Vifor International. K.D. has received research grants and/or honoraria from Amgen Inc. and Vifor International.

References

View Abstract