Anemia and Change in Hemoglobin Over Time Related to Mortality and Morbidity in Patients With Chronic Heart Failure
Results From Val-HeFT
Background— Anemia is known to be a prognostic marker for patients with heart failure. However, little is known about the prognostic value of changes in hemoglobin (Hgb) over time or about the causes of anemia.
Methods and Results— Retrospective analysis of Valsartan Heart Failure Trial data indicated that the quartile of patients with the biggest average decrease in Hgb over 12 months (from 14.2 to 12.6 g/dL) had significantly (P≤0.01) increased risk of subsequent hospitalization (hazard ratio [HR], 1.47), morbid events (HR, 1.41), and death (HR, 1.6) compared with the quartile that exhibited little change in Hgb over 12 months (from 13.7 to 13.8 g/dL). Increasing Hgb was significantly associated with lower mortality in patients with (HR, 0.78) and without (HR, 0.79) anemia at baseline. Anemia at baseline and the changes in Hgb were independently associated with serum albumin, blood pressure, glomerular filtration rate, B-type natriuretic peptide, and C-reactive protein. Lack of anemia at baseline and increases in Hgb over 12 months were not associated with smaller left ventricular diameters or higher ejection fractions.
Conclusions— Changes in Hgb over 12 months were inversely associated with subsequent risk of mortality and morbidity, independently of the effects of baseline anemia and other important predictors. Several factors were independently related to anemia at baseline and changes in Hgb, suggesting multiple causes of anemia in patients with heart failure. These findings raise important questions about the optimal level of Hgb in patients with moderate to severe heart failure and how to achieve them.
- angiotensin receptor blockers
- clinical trials
- natriuretic peptides
Received July 21, 2003; de novo received October 8, 2004; revision received May 5, 2005; accepted May 10, 2005.
There is increasing evidence that anemia is common in patients with heart failure (HF)1,2 and is correlated with increases in mortality and morbidity.2–4 However, it remains unclear whether changes in hemoglobin (Hgb) over time are also related to the risk of morbid events and mortality. Moreover, neither the factors that cause anemia in HF nor the mechanisms that worsen HF in anemic patients are well understood. Whereas chronic anemia can lead to high-output failure,5,6 it is unclear whether anemia is a cause or a consequence of the low-output HF.
A number of small clinical studies have shown that treatment of anemia with erythropoietin in patients with HF is clinically beneficial.7,8 However, therapeutic measures to increase Hgb may worsen hemodynamics,5,9 and raising hematocrit above 42% increased cardiovascular mortality in a large randomized trial.10 More recently, at least 4 clinical trials in Europe were stopped prematurely because of excessive vascular events in the erythropoietin-treated group.11,12 These data therefore raise concerns as to whether Hgb should be raised in patients with HF and questions about the ideal level to be achieved.
The Valsartan Heart Failure Trial (Val-HeFT)13 database was analyzed to address some of the unanswered questions about anemia and HF. The aims of the present analyses were to confirm the prognostic value of baseline Hgb in HF, to relate changes in Hgb over time to mortality and morbidity, and to identify factors related to changes in Hgb, including the effect of valsartan.
Study Design and Patients
Val-HeFT was a randomized, placebo-controlled, double-blind, multicenter trial in patients with symptomatic HF that evaluated the efficacy of the angiotensin receptor blocker valsartan.13 There were no exclusion criteria for Hgb. The upper cutoff for serum creatinine was 2.5 mg/dL. Valsartan caused a 13.3% decrease in the first morbid event and a 27.5% decrease in hospitalizations for HF.13
Hgb, Biochemical, and Echocardiographic Measurements
A complete blood cell count and measurements of serum albumin, uric acid, creatinine, and blood urea nitrogen were performed at a central laboratory with a commercially automated system at randomization and after 4, 6, 12, 18, and 24 months. Glomerular filtration rate (GFR) was calculated using the Modification of Diet in Renal Disease equation.14 Other variables used in this analysis included neurohormones (B-type natriuretic peptide [BNP], measured by Shionogi radioimmunoassay,15 norepinephrine [high-performance liquid chromatography with electrochemical detection],15 and aldosterone [by radioimmunoassay]), high-sensitivity C-reactive protein (CRP), left ventricular internal diastolic diameter, and left ventricular ejection fraction (LVEF). Similar findings were seen when hematocrit was analyzed instead of Hgb. To simplify, analysis of hematocrit is not be discussed further.
Val-HeFT had 2 primary end points: time to death and the first morbid event, which was defined as death, sudden death with resuscitation, hospitalization for HF, or administration of intravenous inotropic or vasodilator drugs for ≥4 hours without hospitalization.
Analyses of Data
Baseline Hgb values were grouped into anemia defined as Hgb <13.0 g/dL for men and <12.0 g/dL for women for analysis.16 Characteristics of groups with or without anemia at baseline were compared by χ2 tests, independent group t tests, and Wilcoxon rank-sum tests, depending on the level and distribution of measurements. Logistic regression was used to identify variables that were independently related to the presence of anemia at baseline.
Cox proportional-hazards regression models were used to relate anemia, change in Hgb, and other variables to time to death, hospitalization for HF, and first morbid event. Relationships were summarized as hazard ratios (HRs) with 95% CIs. Tests for linear relationships between scaled Schoenfeld residuals and time as ranks were conducted to assess the proportional-hazards assumption. The proportional-hazards assumption was tenable for all results shown. Change in Hgb from baseline to 12 months was analyzed by quartiles to examine the trend in relationships to event likelihood. Linearity was examined by plotting changes in Hgb as a continuous variable versus Martingale residuals estimated by Cox regression models.
Relationships between changes in Hgb and changes in other variables were examined by ANOVA with tests for linear trend across quartiles of changes in Hgb from baseline to 12 months. A Kruskal-Wallis analysis of ranks was also done to confirm the ANOVA. Multiple linear regression was used to identify independent correlates of change in Hgb.
A value of P≤0.05 was considered statistically significant without adjustment for multiple comparisons. SPSS (version 12) and STATA (version 8) software were used to perform all analyses.
Prevalence of Anemia
Overall, 23% of the patients enrolled in the Val-HeFT were anemic (23% of men, 24% of women). Baseline characteristics of anemic and nonanemic patients are compared in Table 1. Patients with anemia were slightly older and were less likely to be white or to have an ischemic origin of HF. More patients with anemia had diabetes. Clinical, echocardiographic, and neurohormonal indexes and use of medications indicated that anemia was associated with somewhat more severe HF. On average, Hgb levels did not differ between patients taking or not taking an ACE inhibitor (13.7±1.4 g/dL [n=4636] versus 13.6±1.6 g/dL [n=366]; P=0.74) or a β-blocker (13.7±1.5 g/dL [n=1749] versus 13.7±1.4 g/dL [n=3258]; P=0.75).
Baseline characteristics were entered into a multivariable logistic regression model to identify those that were associated with anemia independently of other factors (Table 2). Greater peripheral edema, BNP, and CRP were independently associated with increased likelihood of anemia, as were presence of diabetes and use of a diuretic and β-adrenergic receptor blocker. Anemia was less likely in patients with higher diastolic blood pressure, GFR, serum albumin, and weight. When other factors were controlled for, white people were also significantly less likely to have anemia. However, LV size and function were not associated with anemia.
As in previous studies, anemia was associated with increased risk of mortality (unadjusted HR, 1.39; 95% CI, 1.21 to 1.57), hospitalization for HF (unadjusted HR, 1.55; 95% CI, 1.34 to 1.81), and first morbid event (unadjusted HR, 1.36; 95% CI, 1.20 to 1.50). When other known prognostic variables were included in the analysis, anemia continued to be a significant independent predictor of mortality and first morbid event but was not quite statistically significant for hospitalization for HF (P=0.08; Table 3).
Changes in Hgb Over 12 months and Subsequent Events
Table 4 summarizes the results of analysis of the 12-month changes in Hgb by quartile. Because patients in the different change quartiles could have differences in baseline risk factors, estimates were adjusted for all the baseline covariates listed in Table 3, including anemia and random assignment to valsartan, which was associated with decreases in Hgb (Figure 1). Patients in quartile 1 (Q1) who, on average, had the greatest decrease in Hgb (from 14.2 to 12.6 g/dL) had a significantly greater subsequent risk of mortality, hospitalization for HF, and first morbid event than patients in Q3, who had the smallest average change in Hgb (from 13.7 to 13.8 g/dL; Table 4 and Figure 2). Patients in Q2 on average had a smaller decrease in Hgb (from 13.9 to 13.4 g/dL) that was not associated with a greater risk than patients in Q3. Patients in Q4 had a larger mean increase in Hgb (from 13.3 to 14.4 g/dL) than those in Q3 (from 13.7 to 13.8 g/dL). The unadjusted event rates shown in Table 4 for Q4 were slightly higher than in Q3. However, patients in Q4 had greater baseline risk. For example, 32% of the patients in Q4 were anemic at baseline compared with only 20% of those in Q3. Multivariable analysis that made adjustments for differences in baseline risk suggested that the risk tended to be lower in Q4 than Q3, albeit not statistically significant.
The interaction between the effects of changes in Hgb as a continuous variable and baseline anemia was not significant for any end point. Nevertheless, analysis of changes in Hgb was done separately for groups that did or did not have anemia at baseline. There were 668 patients with anemia at baseline who survived 12 months and had a complete set of baseline covariates. Of the patients with anemia at baseline, 59% had an increase in Hgb over 12 months averaging 1.0±1.0 g/dL. After adjustment for the baseline covariates listed in Table 3, an increase in Hgb of 1 g/dL was associated with a significantly lower risk of mortality (HR, 0.78, 95% CI, 0.65 to 0.93) but not hospitalization for HF (HR, 0.91; 95% CI, 0.80 to 1.03) or first morbid event (HR, 0.92; 95% CI, 0.78 to 1.07). In the much larger subgroup of 2424 patients who were not anemic at baseline, 43% had an increase in Hgb that averaged 0.6±0.5 g/dL. In the nonanemic group, an increase in Hgb of 1 g/dL was also associated with a significantly lower risk of mortality (HR, 0.79; 95% CI, 0.71 to 0.89), hospitalization for HF (HR, 0.86; 95% CI, 0.78 to 0.95), and first morbid event (HR, 0.84; 95% CI, 0.78 to 0.90).
Relationships Between Other Variables and Changes in Hgb Over 12 Months
Changes in several baseline correlates were related to change in Hgb over 12 months. Compared with patients who had a decrease in Hgb (Q1 and Q2 in Table 5), patients with an increase in Hgb (Q4) had larger increases in weight and serum albumin on average, a smaller decrease in diastolic blood pressure, and larger decreases in CRP and BNP. Interestingly, the increase in Hgb in Q4 was associated with slightly less improvement in left ventricular internal diastolic diameter and LVEF than Q1 and Q2. Changes in furosemide dose and peripheral edema, jugular venous distention, and orthopnea were not associated with a change in Hgb (data not shown). A multivariable regression analysis that included variables listed in Table 6 indicated that changes in serum albumin, diastolic blood pressure, GFR, BNP, and CRP were independently associated with changes in Hgb. The negative relationship between changes in Hgb and LVEF remained significant when these other factors were included in the model. Because valsartan also had an effect on a number other variables13 such as left ventricular internal diastolic diameter, LVEF, BNP, and Hgb, the random assignment of patients to valsartan was included as a control variable. When the signs and symptoms of fluid retention were added to this model, they were not significant and did not substantially alter the estimated relationships between changes in variables (data not shown).
This analysis of the Val-HeFT database confirms our previous findings that lower Hgb levels and anemia are associated with increased risk of mortality and morbidity.4 When World Health Organization criteria are used,16 nearly 23% of patients in this study were anemic at baseline. The prevalence of anemia varies from <5% to >60% in different studies, depending on Hgb cutoff and the population studied.17 Although the degree of anemia in the present study was mild, on average, anemia was independently associated with an ≈20% increase in the risk of morbidity and mortality (Table 3). Hence, even modest decreases in Hgb in patients with HF could have important healthcare implications.
The new findings of this study relate to the association of changes in Hgb over time with subsequent mortality and morbidity and the possible mechanisms responsible for changes in Hgb over time. The results show 2 important findings. First, decreases in Hgb over 12 months were associated with higher risk of mortality and morbidity when multivariable regression was used to control for baseline risk factors, including the presence of anemia. Second, baseline anemia and changes in Hgb were independently related to several other variables, including albumin, diastolic blood pressure, GFR, BNP, and CRP. Interestingly, although baseline anemia and BNP were related, anemia was associated with outcomes independently of BNP, suggesting that these variables may have their effects through different mechanisms. Moreover, patients with anemia at baseline did not have significantly more dilated left ventricles or lower LVEFs, and increases in Hgb over 12 months were not associated with greater improvement in left ventricular size and function.
Causes of Anemia in HF
Although the pathogenesis of anemia in patients with HF is unclear, several mechanisms have been implicated. Impaired renal perfusion with decreased erythropoiesis is probably an important factor.1 Tumor necrosis factor alpha-α is elevated in HF and may contribute to anemia through bone marrow depression.18 Use of ACE inhibitors may reduce Hgb by inhibiting erythropoietin synthesis.19 Iron deficiency resulting from malabsorption, nutritional deficiencies, impaired metabolism,20 and aspirin-induced gastrointestinal bleeding may contribute. Finally, hemodilution caused by an increase in plasma volume has been found to cause anemia in nearly half the patients with severe end-stage HF.21 Thus, multiple mechanisms could cause anemia in patients with HF. In this analysis, baseline anemia was independently associated with lower blood pressure and renal dysfunction as measured by GFR. Baseline anemia was also independently associated with peripheral edema, higher BNP, higher CRP, and lower body weight. Serum albumin was independently associated with both baseline anemia and changes in Hgb over 12 months. Both reduced serum albumin and Hgb could be due to hemodilution. However, albumin was related to anemia even when other variables affected by hemodilution were controlled for by multivariable regression. Thus, the relationship between albumin and Hgb might not depend entirely on hemodilution. Alternatively, low albumin might be an early manifestation of the development of a cachexia state. These results suggest that hemodilution, renal dysfunction, and cachexia might all contribute to the anemia seen in patients with HF.
Consequence of Anemia in HF
Chronic severe anemia can cause high-output HF.5,6 Anemic patients are severely vasodilated, which is due partly to both reduced blood viscosity and enhanced nitric oxide availability.9,22 Chronic vasodilation decreases blood pressure, leading to neurohormonal activation with consequent renal salt and water retention.6 The pathogenic role of severe anemia in patients with high-output failure is underscored by studies showing that correction of anemia promptly reverses fluid retention.5
Whether anemia worsens low-output HF or is just a marker of its severity remains unknown. Anemia can increase cardiac work and cause hypertrophy of the left ventricle, a risk factor for worse cardiovascular outcomes.23,24 We have recently shown that a 1-g/dL increase in Hgb was associated with a 4.1-g/m2 decrease in LV mass over 24 weeks.25 Interestingly, in this study, patients with anemia at baseline did not have significantly more dilated left ventricles or lower LVEF, and increases in Hgb over 12 months were associated with less improvement in LV size and function than seen in patients with decreasing Hgb. Thus, it is not clear what effect raising Hgb will have on cardiac remodeling and function.
Should Anemia in HF Be Treated?
Several recent studies have shown that treatment of anemia with erythropoietin was associated with improvements in LVEF, peak oxygen consumption, and NYHA functional class and a decrease in diuretic requirement.7,8 However, the mechanisms for these beneficial effects remain unknown and may involve more than just an increase in Hgb.26 Endogenous erythropoietin levels are elevated in patients with HF,27 raising the question of a relative erythropoietin resistance. Moreover, increases in Hgb can elevate systemic vascular resistance5,9 and increase blood pressure, as seen in this study, which is unlikely to be beneficial to patients with HF.
What is the ideal Hgb level to target in patients with HF? This study found that decreases in Hgb averaging 1.6 g/dL over 12 months in a group of patients with a mean baseline Hgb of 14.2 g/dL were associated with increased risk for subsequent events compared with a group with a mean baseline Hgb of 13.7 and minimal change in Hgb. Furthermore, changes in Hgb were inversely associated with risk in patients who did not meet the definition of anemia at baseline. Therefore, the ideal Hgb in patients with HF might fall in the normal range. These data, however, do not demonstrate that an average increase in Hgb from 13.3 to 14.4 g/dL was associated with a lower risk of mortality or morbidity.
ACE inhibitors depress erythropoietin synthesis and cause an ≈0.5 g/dL decrease in Hgb.19 In our study, baseline Hgb in patients receiving ACE inhibitors was no different from that in patients not receiving them. Although the exact reason for this discrepancy with the literature is unclear, patients not receiving ACE inhibitors in Val-HeFT had more severe HF.28 Therefore, these patients would be expected to have a lower Hgb, confounding the interpretation of our findings. Valsartan was associated with a modest decrease in mean Hgb by 4 months that remained unchanged over the next 2 years (Figure 1). This effect is similar to another angiotensin receptor blocker, losartan.29 A mechanism similar to ACE inhibition has been postulated for reduced Hgb. Despite the decrease in Hgb, valsartan demonstrated a significant 13.2% reduction in the combined end point of mortality and morbidity.17
In conclusion, in patients with moderate to severe HF, anemia and decreases in Hgb over 12 months were independently associated with higher mortality and morbidity. A decrease in Hgb was independently associated with increased risk of death even in patients who were not anemic at baseline. However, those who had an increase in Hgb did not have a significantly reduced risk, although no treatment was given to specifically increase Hgb. These findings suggest that an Hgb level well into the normal range might be ideal to reduce the risk of death and morbidity in patients with HF. In addition to hemodilution, early cachexia, as evidenced by reduced serum albumin and renal dysfunction, appears to contribute to the reduced levels of Hgb seen in patients with HF. Further studies are required to understand the basis of the remarkable association of anemia in patients with HF and mortality and morbidity, to prospectively assess the potential benefit of correcting anemia, and to evaluate the ideal threshold at which therapy should be initiated and the extent of correction considered safe and desirable. Such studies are in progress.
This work was supported by a grant from Novartis Pharmaceuticals AG, Basel, Switzerland.
Drs Glazer and Chiang and Nora Aknay are employees of Novartis Pharmaceuticals Corp.
Cromie N, Lee C, Struthers AD. Anaemia in chronic heart failure: what is its frequency in the UK and its underlying causes? Heart. 2002; 87: 377–378.
Ezekowitz JA, McAlister FA, Armstrong PW. Anemia is common in heart failure and is associated with poor outcomes: insights from a cohort of 12 065 patients with new-onset heart failure. Circulation. 2003; 107: 223–225.
Anand IS, Chandrashekhar Y, Ferrari R, Poole-Wilson PA, Harris PC. Pathogenesis of oedema in chronic severe anaemia: studies of body water and sodium, renal function, haemodynamic variables, and plasma hormones. Br Heart J. 1993; 70: 357–362.
Silverberg DS, Wexler D, Sheps D, Blum M, Keren G, Baruch R, Schwartz D, Yachnin T, Steinbruch S, Shapira I, Laniado S, Iaina A. The effect of correction of mild anemia in severe, resistant congestive heart failure using subcutaneous erythropoietin and intravenous iron: a randomized controlled study. J Am Coll Cardiol. 2001; 37: 1775–1780.
Mancini DM, Katz SD, Lang CC, LaManca J, Hudaihed A, Androne AS. Effect of erythropoietin on exercise capacity in patients with moderate to severe chronic heart failure. Circulation. 2003; 107: 294–299.
FDA ODACMot. Safety of erythropoietin receptor agonists in patients with cancer. Available at: http://www.fda.gov/ohrms/dockets/ac/04/agenda/4037A_Final.htm. Accessed May 6, 2004.
Anand IS, Fisher LD, Chiang YT, Latini R, Masson S, Maggioni AP, Glazer RD, Tognoni G, Cohn JN. Changes in brain natriuretic peptide and norepinephrine over time and mortality and morbidity in the Valsartan Heart Failure Trial (Val-HeFT). Circulation. 2003; 107: 1278–1283.
Dallman PR. Iron Nutrition in Health and Disease. Hampshire, UK: John Libbey & Co; 1996.
Albitar S, Genin R, Fen-Chong M, Serveaux MO, Bourgeon B. High dose enalapril impairs the response to erythropoietin treatment in haemodialysis patients. Nephrol Dial Transplant. 1998; 13: 1206–1210.
Anker SD, Chua TP, Ponikowski P, Harrington D, Swan JW, Kox WJ, Poole-Wilson PA, Coats AJ. Hormonal changes and catabolic/anabolic imbalance in chronic heart failure and their importance for cardiac cachexia. Circulation. 1997; 96: 526–534.
Androne AS, Katz SD, Lund L, LaManca J, Hudaihed A, Hryniewicz K, Mancini DM. Hemodilution is common in patients with advanced heart failure. Circulation. 2003; 107: 226–229.
Anand I, McMurray JJ, Whitmore J, Warren M, Pham A, McCamish MA, Burton PB. Anemia and its relationship to clinical outcome in heart failure. Circulation. 2004; 110: 149–154.
Ruschitzka FT, Wenger RH, Stallmach T, Quaschning T, de Wit C, Wagner K, Labugger R, Kelm M, Noll G, Rulicke T, Shaw S, Lindberg RL, Rodenwaldt B, Lutz H, Bauer C, Luscher TF, Gassmann M. Nitric oxide prevents cardiovascular disease and determines survival in polyglobulic mice overexpressing erythropoietin. Proc Natl Acad Sci U S A. 2000; 97: 11609–11613.