OUP user menu

The effect of micronutrient supplementation on quality-of-life and left ventricular function in elderly patients with chronic heart failure

Klaus K.A. Witte, Nikolay P. Nikitin, Anita C. Parker, Stephan von Haehling, Hans-Dieter Volk, Stefan D. Anker, Andrew L. Clark, John G.F. Cleland
DOI: http://dx.doi.org/10.1093/eurheartj/ehi442 2238-2244 First published online: 4 August 2005


Aims Chronic heart failure (CHF) is a common and leading cause of death in industrialized countries. The potential benefits of micronutrient supplementation in CHF are extensive. Therefore, we examined the influence of long-term multiple micronutrient supplementation on left ventricular (LV) function, levels of pro-inflammatory cytokines, and quality-of-life (QoL) in elderly patients with CHF.

Methods and results Thirty CHF patients [age 75.4 (0.7), mean (SEM), LV ejection fraction (LVEF) ≤35%] were randomized to receive capsules containing a combination of high-dose micronutrients (calcium, magnesium, zinc, copper, selenium, vitamin A, thiamine, riboflavin, vitamin B6, folate, vitamin B12, vitamin C, vitamin E, vitamin D, and Coenzyme Q10) or placebo for 9 months in a double-blind fashion. All subjects were on stable optimal medical therapy for at least 3 months before enrolment. At randomization and at study end, tumour necrosis factor-α and its soluble receptors TNFR-1 and TNFR-2 were measured and six-minute walk test and QoL were assessed. Cardiac magnetic resonance scanning was performed to evaluate cardiac dimensions and LVEF. Two patients died during follow-up. The remaining patients (14 randomized to placebo and 14 to micronutrients) were well matched for LV function, symptoms, and exercise capacity. At the end of the follow-up period, LV volumes were reduced in the intervention group with no change in the placebo group [−13.1 (17.1)% vs. +3.8 (10.0)%; P<0.05]. LVEF increased by 5.3±1.4% in the intervention group and was unchanged in the placebo group (P<0.05). Patients taking micronutrients also had a significant improvement in QoL score between enrolment and study end [+9.5 (1.6)%; P<0.05], whereas those taking placebo had a slight deterioration [−1.1 (0.8)%; P=0.12]. Six-minute walk test and inflammatory cytokine levels remained unchanged in both groups.

Conclusion Long-term multiple micronutrient supplementation can improve LV volumes and LVEF and QoL scores in elderly patients with heart failure due to LV systolic dysfunction.

  • Chronic heart failure
  • Micronutrients
  • Quality-of-life
  • Cytokines

See page 2215 for the editorial comment on this article (doi:10.1093/eurheartj/ehi490)


Chronic heart failure (CHF) remains a common and serious problem, especially among the rising proportion of the population aged >70 despite recent advances in therapy.16 The burden on health-care services, in terms of both resources and expenditure, is high.7

Patients with heart failure, especially if elderly, are prone to nutritional problems for a number of reasons. We have previously published an extensive review on this topic.8 Appetite and food intake may be reduced because of changes in taste and smell, dietary advice on salt and calorie intake, social isolation, and possible congestion of the hepatic and splanchnic circulation. There is conflicting evidence of malabsorption. Reduced mobility may lead to reduced exposure to sunlight and therefore vitamin D synthesis. Loop diuretics may increase the urinary excretion of micronutrients, including thiamine and calcium.

On the other hand, patients with heart failure may be more susceptible to the effects of micronutrient deficiency because of increased oxidative stress (requiring antioxidant protection), impaired skeletal muscle function (possibly exacerbated by vitamin D deficiency), and impaired myocardial contraction (some severe micronutrient deficiencies can cause heart failure and therefore it is reasonable to expect that less severe deficiency may exacerbate existing cardiac dysfunction). Carnitine, Coenzyme Q10, and creatine supplementation have resulted in improved exercise capacity in some studies of heart failure, although not in others.913 There is also conflicting evidence that supplementation with some micronutrients might retard the progression of coronary disease.1418

Although in many countries, micronutrient supplementation is prescribed by doctors only for specific developing or established deficiency, the public purchase and self-administer large quantities at considerable cost. Presumably, sales are high because of perceived benefit, which may reflect a placebo or a real effect. Recently, there have been moves to tighten control of micronutrient sales directly to the public. For all these reasons, it seems very important to establish whether micronutrient supplements are an expensive placebo or effective therapy. A major problem with existing supplementation studies is that they tend to focus on just one micronutrient that, owing to complex metabolic pathways, may be ineffective or even exacerbate deficiencies in other pathways. Also, existing studies have not focused on elderly patients with heart failure, those at greatest risk of deficiency and its adverse consequences.19,20

The aim of the present study, therefore, was to assess the influence that long-term multiple micronutrient supplementations has on left ventricular (LV) function, levels of pro-inflammatory cytokines, and quality-of-life (QoL) in elderly patients with CHF.



We enrolled 32 patients aged >70 with stable CHF due to ischaemic heart disease into the study. We only approached consecutive patients under follow-up who we knew were suitable for the study. Of those approached, two declined to participate and two dropped out because of claustrophobia in the MRI scanner. Ischaemic heart disease was confirmed in all subjects by the presence of a history and electrocardiographic evidence of previous myocardial infarction or important coronary artery disease in at least two vessels on selective coronary angiography. Each patient gave informed, written consent prior to the involvement in the study, which was approved by the local ethics committee. Each individual was at the time of enrolment under follow-up in a community-based heart failure unit. CHF was defined as the presence of symptoms of fatigue or breathlessness on exertion and a LV ejection fraction (LVEF) on echocardiography of ≤35% with no other cause of breathlessness. Stability was defined as a diagnosis of at least 3 months in duration with no exacerbation or change in medication. Patients were on otherwise optimal therapy including diuretics, angiotensin-converting enzyme inhibitors (ACE-I), and β-blockers if tolerated. We excluded patients with neurological or inflammatory conditions or other significant chronic morbidity affecting QoL (e.g. severe rheumatoid arthritis) or requiring long-term systemic steroid or non-steroidal anti-inflammatory drugs therapy (except low-dose aspirin). We also excluded patients in persistent atrial fibrillation to optimize the reproducibility of our estimation of LV function.

At a baseline visit, each subject completed a QoL questionnaire, undertook a six-minute walk test following a standard protocol21 and had blood samples taken for full blood count, renal, thyroid and liver function, N-terminal BNP, and cytokine measurements [tumour necrosis factor-α, (TNF-α) and soluble TNF receptor (sTNF-R)]. Within 5 days, each patient then underwent cardiac magnetic resonance (CMR) to assess cardiac function. Patients were reviewed every month.


Subjects were randomized by a two-by-two method to receive either micronutrient supplementation or placebo capsules in a double-blind fashion. This was coordinated by a remote pharmacy with which the investigators had no contact during the study. The patients were asked to take four visually identical, opaque capsules per day of either placebo or micronutrients. Table 1 shows the daily doses of the micronutrients received by those on active therapy. The placebo capsules contained cellulose. The micronutrients investigated were chosen in the light of results of previous human, animal, and basic science studies (reviewed in Witte et al.8). The doses of each were guided by previous studies and internationally recommended daily intakes (RDI) (Table 1). The pre-specified therapy duration was 1 year.

View this table:
Table 1

Doses of micronutrients, RDI, and upper safe daily intakes

NutrientDaily dose (four capsules)RDIUpper safe limit for total daily intake
Calcium250 mg800 mg2500 mg
Magnesium150 mga300 mg700 mg
Zinc15 mg15 mg30 mg
Copper1.2 mg1.2 mg9 mg
Selenium50 µg65 µg450 µg
Vitamin A800 µg800 µg3300 µg
Thiamine200 mga1.4 mgNo limit
Riboflavin2 mg1.5 mgNo limit
Vitamin B6200 mga2 mg300 mg
Folate5 mga200 µgNo limit
Vitamin B12200 µg1 µgNo limit
Vitamin C500 mga60 mg2000 mg
Vitamin E400 mga10 mg900 mg
Vitamin D10 µga5 µg25 µg
Co-enzyme Q10150 mga15 mgNo limit

aDoses taken from previous work8 or RDI.

Assessment of cytokines and cytokine receptors

Blood samples were taken at baseline and at study end. Commercially available ELISA kits were used for assay of immune markers in venous samples. We chose TNF-α and IL-6 prospectively because of their clear relationship to adverse prognosis in CHF, and the availability of reliable detection kits. After centrifugation, samples were frozen at −70° C until analysis. Concentrations of TNF-α, TNFR-1, and TNFR-2 were determined by test kits (R&D Systems Minneapolis, USA), the lower limit of detection being 0.18, 25, and 2 pg/mL, respectively. Plasma concentrations of interleukin-6 (IL-6) were measured by Immulite (Random Access Immunoassay Analyzer, DPC Biermann, Bad Nauheim, Germany), the lower limit of detection being 1.0 pg/mL.

LV function

LV volumes, mass, and EF were measured using ECG-gated breath hold gradient cine CMR (GE, Norway, and Lister Healthcare, UK) within 5 days before enrolment and at study end. The analysis of anonymous and undated paired images was performed at the end of the study, using MASS MRI software (Medis, Leiden).


All patients self completed the EuroQoL heart failure QoL score tool at baseline, every month and at study end. This questionnaire, which incorporates evaluation of NYHA class and assesses a broad range of health-related and QoL issues relevant to heart failure, has been used in several large CHF surveys.22 The patient is asked to respond to a series of 40 questions about symptoms of fatigue, breathlessness, ankle swelling, appetite, sleep, depression, mobility, and social activities. Each question is answered by grading the effect of each symptom. The answers are graded from ‘not at all’ (scoring 1) to ‘very much’ (scoring 0). Questions can be used individually to assess different aspects of the patient's life, while two general questions ask about the patient's perception of their overall health and overall QoL. In this way, a maximal score of 40 suggests an optimal QoL, whereas a score of 0 suggests a very poor QoL. The scores were then converted to a percentage of maximum at each time point. In a cohort of 250 individuals of a similar age chosen at random from the patient lists of local general practitioners with no history or symptoms of cardiovascular disease, the average QoL score was 36 (90%) [95% confidence intervals (CI) 35–37 (87–93%)] (Cleland et al., unpublished results). The same scoring system has been used in the Euroheart Failure survey in which patients with definite CHF scored 59% (95% CI 57–60%),22 and in a randomized controlled trial of revascularization in CHF.23 We analysed these at baseline and compared them with the results at study end. By examining the symptom assessment every month, we were also able to look at longitudinal changes in QoL scores in those randomized to vitamins and those to placebo.

The study was terminated in September 2003. At the final visit, all tests were repeated and the CMR scans were performed within 5 days.

Statistical analysis

The primary outcome measure was pre-specified as an improvement in LV function (LVEF) as measured by MRI. CMR is highly reproducible allowing for small sample sizes.24 A predicted average improvement of 10% in the active treatment arm, gave a sample size to confidently (90% power) detect a significant difference between populations of 30 patients (15 in each treatment arm).

The study code was not broken until all the data had been collected and analysed. Two-sided testing was used throughout the analysis. We recognized the risk of multiple testing, and the possible inflation of Type I error,25 but as there is no consensus on how to deal with this,26 we have not adjusted the data and the results should be interpreted accordingly.

We used Student's t-test to look at differences between the two groups of patients for continuous variables also for the changes in such variables between the two time points (Statview, SAS Institute). NYHA status was examined using the χ2 test. The results of the QoL questionnaire were treated as continuous data after confirming normality of distribution using the Kolmogorow–Smirnov test. Results are presented as means (SD). A P-value <0.05 was considered significant, and absolute P-values have been included for important but non-significant variables.


Of 32 patients enrolled, two were unable to tolerate the CMR scan. Our aim was to continue therapy for all patients for 1 year, but because of slow recruitment and the limited shelf-life of our supplements, follow-up was censored in September 2003. Patients were followed for an average of 295 days (inter-quartile range 105 days). Both groups were followed-up for the same length of time. The characteristics of the patients completing the study and their medication at enrolment are shown in Table 2. The patients in the two arms were similar in age, height, weight, EF, loop diuretic dose, NYHA status, TNF-α and TNF-R levels, and functional capacity. There was no change in β-blocker dose during the follow-up period. Two of the 30 patients (one in each group) enrolled into the study died in hospital during the follow-up period. One died of pneumonia (placebo group) and one as a consequence of extensive deep vein thrombosis and subsequent chest infection. QoL data of these patients from the time of enrolment to the last visit before death were included. There were no other adverse events. The capsules were well tolerated. There was no change in baseline renal, liver or thyroid function. The average dose of loop diuretic (frusemide equivalent) in the patients taking active capsules tend to decline during follow-up [63.5 (34) vs. 56 (31) mg], whereas that in the subjects taking placebo did not change [65 (23) vs. 67 (28) mg; P=0.09 for the difference].

View this table:
Table 2

Baseline characteristics

Total study population (n=28)Active therapy (n=14)Placebo (n=14)P-value (between active and placebo)
Age (years)75.4 (4.2)74.2 (2.8)75.5 (3.5)0.75
Height (cm)166 (10.7)170 (9.4)165 (10.0)0.07
Weight (kg)76.8 (13.8)80.4 (9.2)74.3 (12.4)0.06
Average NYHA class2.6 (0.5)2.6 (0.4)2.5 (0.4)0.78
BMI (kg/m2)26.4 (3.2)27.8 (2.4)26.4 (3.5)0.32
Heart rate (b.p.m.)63 (17)62 (28)67 (25)0.35
Mean BP120/76123/78119/740.21
pVo2 (mL/kg per minute)19.8 (4.9)19.1 (3.6)18.8 (5.0)0.33
6MWT (m)287 (119)279 (128)267 (110)0.54
LVEF (%)26.1 (6.7)25.6 (6.9)26.6 (6.8)0.71
LVEDV (mL)217 (92)222 (86)212 (95)0.23
SV (mL)50.1 (17.9)52.8 (18.1)47.8 (11.8)0.25
FEV1 (L/min)1.9 (0.6)1.7 (0.6)2.0 (0.7)0.26
FVC (L)2.8 (1.0)2.7 (0.9)2.8 (1.0)0.62
Mean frusemide equivalent (mg)a64 (35)63 (34)65 (23)0.74
ACE-I/AIIA (n)281414
Beta-blocker (n)231211
Statin (n)271314
Spironolactone (n)743
Serum sodium (mmol/L)138 (3.5)138 (3.3)138 (3.7)0.92
Serum creatine (µmol/L)136 (68)140 (78)133 (59)0.42
Haemoglobin (g/dL)13.3 (1.9)13.4 (1.7)13.1 (2.0)0.38
TSH (pmol/L)1.8 (8.7)1.9 (1.6)1.7 (1.1)0.34
Serum magnesium (mmol/L)0.9 (0.1)0.9 (0.1)0.9 (0.1)0.72
Serum vitamin B12 (ng/L)407 (179)392 (192)420 (171)0.12
Serum ferritin (µg/L)97 (95)103 (108)93 (87)0.09
Serum folate (µg/L)374 (150)372 (147)375 (155)0.28
Serum calcium (mmol/L)2.3 (0.1)2.3 (0.1)2.3 (0.1)0.56
Serum TNF-α (pg/mL)5.4 (2.6)5.6 (3.2)5.1 (2.0)0.11
Serum TNF-R1 (pg/mL)1192 (391)1228 (432)1161 (363)0.08
Serum TNF-R II (pg/mL)1974 (648)1951 (618)1993 (694)0.10
Baseline QoL66.3 (8.3)64.4 (13.4)67.2 (9.8)0.34

Values are mean (SD). NYHA, New York Heart Association classification (χ2 P-value); BMI, body mass index; pVo2, peak oxygen consumption; 6MWT, six-minute walk test distance; LVEDV, left ventricular end-diastolic volume; SV, stroke volume; FEV1, forced expiratory volume in 1 s; FVC, forced vital capacity; ns, non-significant; AIIA, angiotensin II inhibitor; TSH, thyroid-stimulating hormone; sTNF-RI, soluble tumour necrosis factor-receptor 1; sTNF-RII, soluble tumour necrosis factor-receptor 2.

aThe mean daily dose at enrolment of frusemide or frusemide equivalent (1 mg bumetanide=40 mg frusemide) is given (SD).

LV function

Table 3 shows the changes in CMR-measured variables in the two groups. LV end-diastolic and end-systolic volumes declined and EF improved in patient's randomized to micronutrient supplements compared with placebo (all P<0.05) (Table 3 and Figure 1).

Figure 1 Change in left ventricular end-diastolic volume with placebo (unfilled circles and dashed line) and micronutrient supplementation (filled circles and solid line), (error bars are SD).

View this table:
Table 3

CMR variables, TNF- α, selected micronutrients and QoL scores (% of maximum) at baseline and study end with micronutrient supplementation and placebo

Baseline active therapy group (n=14)Post-active therapy (n=14)Baseline placebo group (n=14)Post-placebo (n=14)P-value for change between active and placebo
LVESV (mL)169 (83)138 (86)164 (87)172 (81)0.02
LVEDV (mL)222 (86)193 (90)212 (95)224 (91)0.03
LV mass diastole (g)160 (38)158 (43)143 (34)154 (41)0.08
LV mass systole (g)174 (39)167 (41)158 (33)166 (45)0.08
SV (mL)52.8 (18.1)55.3 (20.1)47.8 (11.8)52.4 (13.5)0.17
LVEF (%)25.6 (6.9)30.9 (7.1)26.6 (6.8)26.2 (7.2)0.03
Heart rate (b.p.m.)62 (28)59 (29)67 (25)66 (16)0.12
Serum magnesium (mmol/L)0.9 (0.1)1.1 (0.1)0.9 (0.1)0.5 (0.1)0.04
Serum vitamin B12 (ng/L)392 (192)409 (194)420 (171)409 (162)0.02
Serum ferritin (µg/L)103 (108)576 (261)93 (87)77 (68)0.009
Serum folate (µg/L)372 (147)400 (20)375 (155)285 (124)0.008
TNF-α (pg/mL)5.6 (3.2)5.7 (4.5)5.1 (2.0)5.1 (2.1)0.86
sTNF-RI (pg/mL)1228 (432)1207 (443)1161 (363)1450 (374)0.16
sTNF-RII (pg/mL)1951 (618)2250 (643)1993 (694)2504 (716)0.31
QoL64.4 (13.4)73.9 (1.6)*67.2 (9.8)66.1 (11.1)0.02

Values are means (SD). LVESV, left ventricular end-systolic volume, other abbreviations as for Table 2.


The two groups were well matched for TNF-α, IL-6, and TNF-RI and II serum levels at baseline. There were no statistically significant changes in these variables with micronutrient supplementation (Table 3).

Exercise capacity

Baseline six-minute walk distance and change during follow-up were similar in the two groups.

Symptoms and QoL

There was no change in NYHA class in either group. The baseline scores, expressed as a percentage of maximum, for QoL are shown in Table 3 and show the two groups to be well matched. Patients taking micronutrients had an increase in their QoL score [64.4 (13.4) vs. 73.9 (9.4); P=0.02], and patients on placebo a decrease [67.2 (9.8) vs. 66.1 (7.9); P=NS]. There is a significant difference in the change of QoL scores between the two groups. The scores started to diverge after 6 months (Figure 2). Further analysis of the data suggested that the difference in overall score was driven predominantly by the improvements in scores for exertional breathlessness, quality of sleep, daytime concentration, and overall QoL ratings.

Figure 2 Quality of life score (percentage of maximum) during study period for patients taking placebo (unfilled circles and dashed line) or micronutrient supplementation (filled circles and solid line), (error bars are SD).


This is the first study to investigate the effects of multiple micronutrient supplementations on LV function and QoL in patients with heart failure. The study suggests that such treatment improves LV function when assessed using CMR imaging, an accurate and highly reproducible method for assessing ventricular volumes that allows interventions to be tested in a small number of patients.24 Improvement in ventricular function was associated with some evidence of improved patient well-being.

Previous studies of micronutrient supplements in heart failure have only randomized patients to a single agent. This reflects bias against the use of multiple supplements on the grounds that if improvement is observed, the mechanism will be unclear. However, using single supplements to correct only one deficiency when multiple exist may be useless and could be harmful by causing the accumulation of harmful metabolic products. On the other hand, micronutrients may have synergistic effects. For example, vitamin C is involved in the regeneration of cell-membrane bound vitamin E. Testing single substances in a series of randomized controlled trials would not only take a long time but could be misleading. Starting with a broad range supplement circumvents these problems and was also the strategy of a recent multivitamin supplementation study in patients with HIV infection.27 The value of identifying elements to be omitted without loss of benefit can then be assessed. This study reflects the first step on that route. It is possible that many components of our micronutrient cocktail were unnecessary or that further components may be added for greater effect.

Severe thiamine deficiency is a well-documented cause of heart failure. It has also been suggested that high doses of thiamine (200 mg daily) can improve cardiac function in patients with CHF receiving loop diuretics, which cause an increase in urinary excretion.28,29

One potential mechanism of benefit from micronutrient supplementation is the reduction of oxidative stress. Markers of oxidative stress are elevated in heart failure patients and correlate with functional class, reduced exercise tolerance, lower antioxidant levels, and indices of worse prognosis.30,31 Patients with heart failure have lower levels of co-enzyme Q10, a powerful endogenous antioxidant, in their myocardium compared with controls.32 Low plasma co-enzyme Q10 levels are associated with an increased mortality in heart failure.33 Supplementation with co-enzyme Q10 at a variety of doses in uncontrolled studies has shown improvements in EF, exercise tolerance, and NYHA status. However, data from randomized studies are less consistent, although studies using higher doses (≥2 mg/kg/day) were more likely to show benefit.913 Recent meta-analyses have clarified the situation somewhat suggesting benefits on symptoms and cardiac output and a trend to improved mortality in patients with CHF.34,35 Nevertheless, the reduction of the synthesis of co-enzyme Q10 by statins seems to be of sufficient concern that regulatory authorities required plasma co-enzyme Q10 measurements in a substantial proportion of patients in a large mortality trial of rosuvastatin in heart failure.

Vitamin C, vitamin E, magnesium, selenium, and zinc also appear to have anti-oxidant effects and may affect cardiac function.3639 Vitamin C acutely improves the ejection duration in patients with CHF independent of heart rate.40 Vitamin C also reduces apoptosis in cardiomyocytes in rats with experimental heart failure.41 Vitamin D and calcium deficiency can impair myocardial contractility, but the effects of supplementation in humans are unknown.42 The only study examining the effects of vitamin E in CHF patients was neutral.43

Peripheral vasodilation with a reduction in peripheral vascular resistance and afterload might also contribute to improved cardiac function. Selenium can cause vascular smooth muscle relaxation,44 and vitamins C and E can improve endothelial-dependent vasodilation.45,46 The B vitamins reduce homocysteine levels. Prolonged exposure to high homocysteine levels leads to a fall in nitric oxide production and impaired peripheral vasodilation.47

All our patients had coronary disease. Selected micronutrient supplementation in patients at risk of or with established coronary artery disease has generally not been shown to improve outcome.1618 However, few of these patients had heart failure and therefore may have had a lower risk of micronutrient deficiency or susceptibility and, as noted earlier, the results of studies using single agent supplements may be misleading.

There was no difference in NYHA status after micronutrient supplementation, but the study population was small. Our patients also completed QoL questionnaires at the end of every 1-month-follow-up period. The QoL of the patients improved significantly with the micronutrient supplementation.

The QoL score improved in our patients predominantly due to improved scores for fatigue and breathlessness during daily activity, improved quality of sleep, and improved daytime concentration. However, we did not observe an improvement in NYHA score or distance walked on the six-minute walk test. The effects of supplements on symptoms could reflect a chance finding, as multiple measurements were made and the QoL tool that we used has not yet been fully validated. However, the primary outcome measure of this study, LVEF, did improve and improved cardiac function might be expected to be associated with improved symptoms. It is possible that the failure to demonstrate an improvement in NYHA class or walk-test distance reflects inadequate study size rather than lack of effect, given the variability of these tests and probably only a modest benefit.


When added to conventional therapy with β-blockers and ACE-Is multiple micronutrient supplementations may improve LV function and patient well-being. Further trials are warranted with larger numbers of patients and a longer follow-up period.


Although carried out in a randomized, prospective, and double-blind fashion, the present study involved small number of patients and the data must be viewed in the light of this. We also did not measure blood levels of all the micronutrients to assess compliance, although the changes in the ferritin, vitamin B12, and folate levels in the patients randomized to the micronutrient combination suggest that they took them.


View Abstract