Professional Documents
Culture Documents
net/publication/267044426
CITATIONS READS
20 670
12 authors, including:
Some of the authors of this publication are also working on these related projects:
Endogenous erythropoetin and new onset heart failure in the general population: PREVEND View project
All content following this page was uploaded by Peter van der Meer on 12 November 2014.
ORIGINAL ARTICLE
Renal function was assessed using the estimated glomerular haemoglobin level, eGFR, systolic blood pressure and serum
filtration rate (eGFR, mL/min/1.73 m2), calculated from the sim- NT-proBNP, hs-CRP and erythropoietin levels.
plified Modification of Diet in Renal Disease equation. Vitamin A two-sided p value <0.05 was considered statistically signifi-
B12, folic acid and N-terminal prohormone brain natriuretic cant. Data were analysed using STATA V.12.0 (StataCorp LP,
peptide (NT-proBNP) levels were determined using an immuno- College Station, Texas, USA).
assay based on electrochemiluminescence (Elecsys, Roche
Diagnostics, Mannheim, Germany). Serum concentrations of RESULTS
high-sensitive C-reactive protein (hs-CRP) and other blood Baseline characteristics
markers were assessed using standard methods. Baseline characteristics for all 610 patients were stratified
according to serum vitamin B12 and folic acid levels in quartiles
Definitions and study end points (tables 1 and 2). Patients were recruited from Poland, the
Nutritional deficiencies were defined as follows: serum vitamin Netherlands and Spain. Baseline characteristics for all 610
B12<200 pg/mL8 and serum folic acid <4.0 ng/mL.11 Anaemia patients stratified by study cohort are shown in online supple-
was defined as a haemoglobin level <12 g/dL in women and mentary table S2. Median serum vitamin B12 and folic acid
<13 g/dL in men, in agreement with WHO standards.12 Iron levels (+IQR) were 435 (312–600) pg/mL and 9.4 (6.8–12.5)
deficiency was defined as a serum ferritin level <100 mg/L or ng/mL, respectively. Vitamin B12 levels were lowest in the Dutch
serum ferritin 100–299 with a TSAT <20%, reflecting both subcohort, whereas folic acid levels were lowest in the Polish
absolute and functional iron deficiency. Quality of life was subcohort. Vitamin B12 and folate deficiency were present in
assessed by means of the Minnesota Living with Heart Failure 5.4% and 4.1% of the patients, respectively. More than half of
Questionnaire (MLHFQ).13 Data regarding demographics and the patients were iron deficient (58%). The prevalence of
clinical signs (age at diagnosis, sex, HF aetiology, renal function, vitamin B12 and folate deficiency was equal in patients with and
EF, blood pressure and comorbidities), haematinics (haemoglo- without iron deficiency (5.9% versus 4.7%, p=0.516, and 3.1%
bin, MCV, serum iron, ferritin, transferrin, vitamin B12 and folic versus 5.5%, p=0.142, respectively). Folate deficiency was more
acid) and therapeutics were available for all patients. Follow-up prevalent in anaemic patients (6.7% versus 2.8% in non-
for surviving patients was censored after 9 years, when <5% of anaemic patients, p=0.020), whereas vitamin B12 deficiency was
all patients were at risk. The primary end point of this study not (6.2% versus 5.0%, p=0.538). None of the patients had
was all-cause mortality. both vitamin B12 and folate deficiency.
Statistical analysis
Clinical correlates
Data are presented as mean±SD when normally distributed, as
Univariable and multivariable linear regression models are
median and IQR when non-normally distributed or as percent-
shown in tables 3 and 4. Lower serum folic acid levels were
age when categorical. Intergroup differences were tested using
associated with the presence of atrial fibrillation and Polish and
parametric and non-parametric trend tests, when appropriate.
Spanish study cohorts ( p=0.001). Higher concentrations of
Baseline characteristics stratified by study cohort were compared
serum vitamin B12 were correlated with higher NT-proBNP and
using one-way analysis of variance test or Kruskal–Wallis test,
ferritin levels, the presence of atrial fibrillation and the Spanish
when appropriate. To assess the distribution of haematinics for
study cohort (all p<0.05). Lower serum vitamin B12 levels were
different MCV levels, serum levels of vitamin B12, folate, fer-
associated with systolic blood pressure ( p=0.008). There was
ritin and TSAT were stratified by MCV levels in tertiles.
no significant correlation between haematinics and treatment in
Changes in serum levels of haematinics were tested using a non-
all multivariable models.
parametric trend test (i.e. extension of Wilcoxon rank-sum test).
Multivariable linear regression models showed that better
After baseline analyses, the follow skewed variables were loga-
quality of life (as assessed by MLHFQ) was significantly corre-
rithmically transformed for all further analyses to achieve a
lated to serum folic acid levels ( p<0.029).
normal distribution: serum NT-proBNP, erythropoietin, iron,
ferritin, vitamin B12, folic acid, TSAT, hs-CRP and eGFR.
To determine clinical correlates of vitamin B12 and folic acid, Haematinic deficiencies and MCV
univariable linear regression models were constructed using age, Serum levels of haematinics stratified by MCV levels in tertiles
sex, body mass index (BMI), LVEF, systolic blood pressure, are shown in figure 1. TSAT was significantly lower in patients
haemoglobin, MCV, eGFR, TSAT, serum NT-proBNP, hs-CRP, with lower MCV at baseline ( p for trend <0.001). A similar
ferritin and erythropoietin, comorbidities, NYHA functional pattern was observed when only anaemic patients were selected
class, HF aetiology, health-related quality of life and treatment (see online supplementary figure S1). Interestingly, there was no
with angiotensin-converting enzyme inhibitors, angiotensin significant trend in prevalence of vitamin B12 and folate defi-
receptor blockers, aldosterone antagonists, β-blockers, statins, ciency for different MCV levels.
loop diuretics, antiplatelets and anticoagulants as covariates.
Variables with a significant univariable correlation with either Haematinics and survival
serum vitamin B12 and/or folic acid levels (i.e. p<0.10) were During a median follow-up of 2.10 years (IQR 1.31–3.60 years),
entered in the multivariable linear regression models. 254 patients (42%) died. Differences in event-free survival strati-
Kaplan–Meier curves were constructed to determine the fied by vitamin B12 and folic acid in quartiles are displayed in
effect of vitamin B12 and folic acid levels on cumulative survival. figures 2 and 3. Increased mortality rates were seen in patients
Differences in survival rates were tested using the log-rank with vitamin B12 levels >600 pg/mL versus those with levels
Mantel–Cox test. To determine the prognostic consequences of <600 pg/mL (47.7% versus 41.8%; p=0.026). In multivariable
vitamin B12 and folic acid, univariable and multivariable Cox Cox proportional hazard regression models, vitamin B12 lost its
proportional hazard regression models were constructed. In the predictive value for all-cause mortality (table 5). Mortality was
multivariable model, adjustment was made for all univariably similar in patients with higher and lower folic acid levels
significant variables, including age, sex, BMI, LVEF, (p=0.745), and the absence of an association between folic acid
2 van der Wal HH, et al. Heart 2014;0:1–9. doi:10.1136/heartjnl-2014-306022
Downloaded from http://heart.bmj.com/ on November 12, 2014 - Published by group.bmj.com
levels and mortality remained present in multivariable Cox pro- Prevalence and definition of haematinic deficiencies
portional hazard regression models. Unfortunately, there are no clear definitions of vitamin B12 and
folate deficiency, leading to different thresholds for the pres-
DISCUSSION ence of these deficiencies, and consequently to differences in
In an international pooled cohort comprising 610 patients with prevalence. Traditionally, serum vitamin B12 levels have been
chronic HF, we demonstrate that vitamin B12 and folate defi- considered as the gold standard for the detection of clinical
ciency are relatively rare (4% and 5%, respectively). Second, vitamin B12 deficiency. According to Stabler et al,8 serum
there is no significant correlation between MCV and vitamin B12 vitamin B12 levels <200 pg/mL are suggestive for clinical defi-
and folic acid levels. Finally, vitamin B12 and folic acid levels ciency (sensitivity and specificity 65%–100% and 50%–60%,
were not independently associated with mortality. respectively). With serum levels >300 pg/mL, vitamin B12
deficiency is unlikely (i.e. probability <5%).14 Exceptionally Measurement of methylmalonic acid and total homocysteine
low serum vitamin B12 levels (i.e. <100 pg/mL) are often asso- was introduced several decades ago to diagnose vitamin B12
ciated with overt deficiency, but such levels are hardly observed deficiency, in addition to serum vitamin B12 levels. However,
(n=3 in the present study). Additionally, false negative and both serum methylmalonic acid and homocysteine may be ele-
false positive value of laboratory assays for serum vitamin B12 vated in renal failure, making methylmalonic acid and homo-
levels can be as high as 50% due to suboptimal assays and cysteine unreliable screening markers for vitamin B12 deficiency
highly variable test results.11 15–17 New assays are currently in in patients with chronic HF as up to 60% of patients with
development (e.g. holotranscobalamin), but these assays need chronic HF are suffering from moderate to severe renal dysfunc-
to be validated clinically before they can be used in clinical tion (i.e. eGFR <60 mL/min/1.73 m2).19–21 In the current study,
healthcare.18 renal dysfunction is present in 39% of all patients. Therefore,
4 van der Wal HH, et al. Heart 2014;0:1–9. doi:10.1136/heartjnl-2014-306022
Downloaded from http://heart.bmj.com/ on November 12, 2014 - Published by group.bmj.com
Table 3 Clinical variables associated with vitamin B12 in chronic heart failure
Univariable regression Multivariable regression
Variable coefficient (95% CI) p Value t Value coefficient (95% CI) p Value t Value
serum methylmalonic acid and homocysteine would definitely and therefore, the prevalence of folate deficiency may be under-
overestimate the prevalence of vitamin B12 deficiency in our estimated in patients with chronic HF when using serum folic
study. In short, there are currently no reliable screening markers acid levels as a marker for folate deficiency.
for vitamin B12 deficiency in patients with chronic HF apart
from serum vitamin B12 levels. Haematinics and quality of life
Measurement of serum folic acid levels is mainly a reflection Our study shows that lower folic acid levels, but not vitamin B12
of short-term folic acid status. In the absence of recent fasting, levels, are related to an impaired health-related quality of life as
serum folic acid levels of <4 ng/mL are diagnostic for folate assessed by a disease-specific questionnaire. A recent study by
deficiency.11 Alternatively, erythrocyte folic acid levels can be Comin-Colet et al23 shows that iron deficiency is also correlated
used to diagnose folate deficiency as this parameter reflects to reduced quality of life. In this study comprising 552 patients
average folic acid availability over time and is less subject to with chronic HF, anaemia did not have any significant influence
daily fluctuations. However, test results are occasionally difficult on quality of life. The importance of quality of life in patients
to interpret and are, therefore, not always conclusive.11 Farrell with chronic HF has been recently stressed by Kraai et al,24
et al22 stated that serum folic acid measurements may be super- stating that most patients with HF prioritise quality of life above
ior to erythrocyte folate measurements. As a consequence, longevity.
initial screening for folate deficiency can be performed using
serum folic acid levels, which is a relatively inexpensive meas- Haematinics and survival
urement. More important, serum folic acid measurements are The predictive value of iron deficiency on mortality in patients
widely available and easy to perform. When serum folic acid with chronic HF has been recently stressed by Klip et al,1
levels are borderline, erythrocyte folic acid levels can be mea- making iron status a strong and independent predictor of
sured to confirm the diagnosis. outcome. Our results show that folic acid does not have such
The current study shows that serum folic acid levels are corre- predictive value on mortality. Univariable proportional hazard
lated with renal dysfunction (i.e. serum folic acid levels are analysis shows that patients with vitamin B12 levels >600 pg/mL
higher when renal dysfunction is present). As stated earlier, have higher mortality risks compared with patients with lower
renal dysfunction is a relatively common comorbidity in HF, levels. Vitamin B12 lost its predictive value on all-cause mortality
van der Wal HH, et al. Heart 2014;0:1–9. doi:10.1136/heartjnl-2014-306022 5
Downloaded from http://heart.bmj.com/ on November 12, 2014 - Published by group.bmj.com
Table 4 Clinical variables associated with folic acid in chronic heart failure (HF)
Univariable regression coefficient Multivariable regression coefficient
Variable (95% CI) p Value t Value (95% CI) p Value t Value
Age, per 5 years 0.022 (−0.002 to 0.046) 0.076 1.78 0.022 (−0.006 to 0.050) 0.135 1.50
Sex, women vs men −0.023 (−0.152 to 0.106) 0.724 −0.35
BMI, per 1 kg/m2 −0.012 (−0.023 to −0.001) 0.037 −2.08 0.010 (−0.001 to 0.021) 0.086 1.72
LVEF, per 1% −0.009 (−0.013 to −0.005) <0.001 −4.04 −0.004 (−0.009 to 0.001) 0.085 −1.72
SBP, per 1 mm Hg 0.001 (−0.002 to 0.004) 0.576 0.56
MCV, per 1 fL 0.006 (−0.004 to 0.016) 0.251 1.15
Hb, per 1 g/dL 0.040 (0.008 to 0.072) 0.014 2.46 0.002 (−0.034 to 0.037) 0.920 0.10
eGFR, per doubling −0.279 (−0.347 to −0.211) <0.001 −8.06 −0.045 (−0.135 to 0.046) 0.331 −0.97
NT-proBNP, per doubling 0.029 (−0.002 to 0.059) 0.064 1.86 0.003 (−0.033 to 0.038) 0.885 0.14
hs-CRP*, per doubling 0.003 (−0.036 to 0.042) 0.881 0.15
EPO, per doubling −0.033 (−0.088 to 0.022) 0.243 −1.17
Ferritin, per doubling 0.009 (−0.035 to 0.052) 0.699 0.39
TSAT, per doubling −0.111 (−0.162 to −0.059) <0.001 −4.19 −0.012 (−0.067 to 0.043) 0.662 −0.44
Vitamin B12, per doubling 0.033 (−0.044 to 0.109) 0.403 0.84
Atrial fibrillation, yes vs no −0.164 (−0.291 to −0.037) 0.011 −2.54 −0.191 (−0.308 to −0.074) 0.001 −3.21
Diabetes, yes vs no −0.153 (−0.272 to −0.035) 0.011 −2.54 −0.032 (−0.148 to 0.084) 0.591 −0.54
Ischaemic aetiology, yes vs no 0.053 (−0.064 to 0.170) 0.376 0.89
NYHA functional class III/IV vs I/II 0.378 (0.261 to 0.495) <0.001 6.34 0.009 (−0.138 to 0.155) 0.907 0.12
MLHFQ, per point −0.004 (−0.006 to −0.001) 0.007 −2.73 −0.003 (−0.005 to −0.0003) 0.029 −2.18
Study cohort (vs Holland)
Poland −0.602 (−0.734 to −0.470) <0.001 −8.98 −0.520 (−0.702 to −0.338) <0.001 −5.61
Spain −0.878 (−0.993 to −0.763) <0.001 −14.93 −0.794 (−0.983 to −0.606) <0.001 −8.28
Treatment
ACE inhibitor and/or ARB, yes vs −0.149 (−0.047 to 0.346) 0.137 1.49
no
Aldosterone antagonist, yes vs no −0.013 (−0.129 to 0.104) 0.833 −0.21
β-Blocker, yes vs no −0.103 (−0.278 to 0.072) 0.249 −1.15
Statins, yes vs no −0.103 (−0.219 to 0.014) 0.084 −1.73 0.032 (−0.077 to 0.141) 0.563 0.58
Loop diuretic, yes vs no 0.124 (−0.042 to 0.291) 0.144 1.46
Antiplatelet and/or anticoagulant, −0.047 (−0.209 to 0.115) 0.567 −0.57
yes vs no
*Serum hs-CRP levels were determined in 385 patients.
BMI, body mass index; eGFR, estimated glomerular filtration rate; EPO, erythropoietin; hs-CRP, high-sensitive C-reactive protein; MCV, mean corpuscular volume; MLHFQ, Minnesota
Living with Heart Failure Questionnaire; NYHA, New York Health Association; SBP, systolic blood pressure; TSAT, transferrin saturation; ARB, angiotensin receptor blockers; NT-proBNP,
N-terminal prohormone brain natriuretic peptide; Hb, haemoglobin.
in multivariable proportional hazard analyses, especially in folate deficiency since these deficiencies are frequently charac-
patients with vitamin B12 levels >600 pg/mL. The univariable terised by an increase in MCV. In the present study, no associ-
effect on mortality is caused by the paradoxical association ation between MCV and neither serum vitamin B12 nor folic
between serum vitamin B12 levels and disease severity (serum acid levels were observed, although we found a significant
NT-proBNP levels). This association was observed earlier by association between MCV levels and the presence of iron defi-
Herrmann et al25 and may be caused by the cardiohepatic syn- ciency. While MCV is a reliable screening marker for haemati-
drome. Chronic HF may cause blood congestion in the liver, nic deficiencies in otherwise healthy people, the diagnostic
leading to hepatic cell damage and, consequently, to mild liver value of this cellular index seems limited in patients with
dysfunction.26 Liver function deterioration may lead to the chronic HF.
release of hepatic and metabolically inactive vitamin B12 analo- In general, lower MCV levels increase the probability of iron
gues, making serum vitamin B12 measurements less accurate. deficiency but do not effectively rule out vitamin B12 and/or
However, this cardiohepatic syndrome remains only a hypoth- folate deficiency as the influence of serum vitamin B12 and folic
esis and is an incomplete explanation of the association between acid levels on MCV is limited when iron deficiency is present.
serum vitamin B12 levels and disease severity. Thus, MCV levels in patients with chronic HF seem to be
mainly driven by iron status. Since iron deficiency is a common
Clinical implications comorbidity in patients with chronic HF,1 MCV levels will be
Traditionally, MCV is one of the conventional erythrocyte para- inevitably lower as the result of microcytic anaemia caused by
meters and is used in the differential diagnosis of anaemia in iron deficiency. The simultaneous presence of more than one
mainstream clinical practice, prior to measuring haematinics. haematinic deficiency might cause MCV levels to be in the
The importance of MCV in the differential diagnosis of normal range (i.e. normocytic anaemia). Even normal MCV
anaemia has been recently emphasised by Brugnara and levels in the absence of iron deficiency do not rule out haemati-
Mohandas.27 Levels of MCV have always been considered as nic deficiencies as macrocytosis is frequently absent in vitamin
one of the hallmarks in the diagnosis of vitamin B12 and B12 and folate deficiency.28 Therefore, MCV should be used
Figure 1 Distribution of serum vitamin B12, folic acid, ferritin and transferrin saturation levels, stratified by mean corpuscular volume (MCV) levels
(in tertiles).
with caution when screening for vitamin B12 or folate deficiency lacking gold standard tests for vitamin levels and no clear defini-
in patients with chronic HF. tions of vitamin B12 and folate deficiency. Low serum vitamin
With respect to prognosis, vitamin B12 or folate supplementation B12 and folic acid levels are not always conclusive for clinical
does not constitute a therapeutic target in patients with chronic deficiency. However, current screening methods for haematinic
HF. Whether folate supplementation might be warranted in view deficiencies are validated thoroughly and are mainstream prac-
of a possible improvement of health-related quality of life is cur- tice in daily healthcare. The cut-offs for haematinic deficiencies
rently unknown. For iron deficiency, several studies have shown used in this study are based on an optimal balance between sen-
beneficial effects of intravenous iron supplementation with respect sitivity and specificity. Serum vitamin B12 and folic acid levels
to improved clinical status, exercise capacity and quality of life.4–6 remain the most important screening markers for these vitamin
deficiencies in patients with chronic HF as long as more reliable
Study limitations markers are lacking.
The main limitation of this study is that the diagnosis of vitamin Second, haematinics were only available at baseline.
B12 and folate deficiency is relatively challenging as a result of Therefore, haematinic changes over time could not be
Figure 2 Kaplan–Meier survival analysis, stratified by vitamin B12 Figure 3 Kaplan–Meier survival analysis, stratified by folic acid levels
levels (in quartiles). (in quartiles).
van der Wal HH, et al. Heart 2014;0:1–9. doi:10.1136/heartjnl-2014-306022 7
Downloaded from http://heart.bmj.com/ on November 12, 2014 - Published by group.bmj.com
Key messages
REFERENCES
1 Klip IT, Comin-Colet J, Voors AA, et al. Iron deficiency in chronic heart failure: an
What is already known on this subject? international pooled analysis. Am Heart J 2013;165:575–82.e3.
Iron deficiency is common in chronic heart failure (HF) and 2 Jankowska EA, Rozentryt P, Witkowska A, et al. Iron deficiency predicts impaired
affects clinical parameters and outcome. However, studies on exercise capacity in patients with systolic chronic heart failure. J Card Fail
2011;17:899–906.
haematinic deficiencies other than iron deficiency in chronic HF 3 van Veldhuisen DJ, Anker SD, Ponikowski P, et al. Anemia and iron deficiency in
are scarce, and clinical correlates or the effects on prognosis are heart failure: mechanisms and therapeutic approaches. Nat Rev Cardiol
currently unclear. 2011;8:485–93.
4 Anker SD, Comin Colet J, Filippatos G, et al. Ferric carboxymaltose in patients with
heart failure and iron deficiency. N Engl J Med 2009;361:2436–48.
What might this study add? 5 Comin-Colet J, Lainscak M, Dickstein K, et al. The effect of intravenous ferric
Vitamin B12 and folate deficiency are relatively rare in patients carboxymaltose on health-related quality of life in patients with chronic heart failure
with chronic HF. Mean corpuscular volume is neither related to and iron deficiency: a subanalysis of the FAIR-HF study. Eur Heart J 2013;34:30–8.
vitamin B12 nor to folic acid levels. Unlike iron deficiency, 6 Filippatos G, Farmakis D, Colet JC, et al. Intravenous ferric carboxymaltose in
vitamin B12 and folic acid levels are not related to prognosis. iron-deficient chronic heart failure patients with and without anaemia: a subanalysis
of the FAIR-HF trial. Eur J Heart Fail 2013;15:1267–76.
Lower folic acid levels are associated with an impaired 7 Pruthi RK, Tefferi A. Pernicious anemia revisited. Mayo Clin Proc 1994;69:144–50.
health-related quality of life as assessed by a disease-specific 8 Stabler SP. Clinical practice. Vitamin B12 deficiency. N Engl J Med
questionnaire. 2013;368:149–60.
9 Witte KK, Desilva R, Chattopadhyay S, et al. Are hematinic deficiencies the cause of
anemia in chronic heart failure? Am Heart J 2004;147:924–30.
How might this impact on clinical practice? 10 Beilby J, Olynyk J, Ching S, et al. Transferrin index: an alternative method for
Since no significant association was observed between mean calculating the iron saturation of transferrin. Clin Chem 1992;38:2078–81.
corpuscular volume and neither vitamin B12 nor folic acid levels, 11 Antony AC. Megaloblastic anemias. In: Hoffman R, Benz EJ, Shattil SJ, eds. Hematology:
this cellular index should be used with caution in the basic principles and practice. 4th edn. New York: Churchill Livingstone, 2005:519.
differential diagnosis of anaemia in patients with chronic HF. 12 [No authors listed]. Nutritional anaemias. Report of a WHO scientific group. World
Health Organ Tech Rep Ser 1968;405:5–37.
References This article cites 29 articles, 7 of which you can access for free at:
http://heart.bmj.com/content/early/2014/10/16/heartjnl-2014-306022
#BIBL
Email alerting Receive free email alerts when new articles cite this article. Sign up in the
service box at the top right corner of the online article.
Notes