Folic Acid Supplementation and Cardiac and Stroke Mortality among Hemodialysis Patients

Abstract

Objective

To assess whether use of folic acid vitamin supplements reduces cardiac and stroke mortality in hemodialysis patients. Further, we examined whether consumption of folic acid from vitamin supplements greater than 1000 μg compared to standard 1000 μg, and 1000 μg compared to either lower dose or no consumption were associated with reduced cardiac and stroke mortality risk.

Design

Secondary analysis of data from the Hemodialysis (HEMO) Study, a randomized clinical trial examining dialysis treatment regimens over three years follow-up. Participants: One thousand eight hundred and forty-six hemodialysis patients previously participating in the HEMO study.

Interventions

None

Main Outcome Measure

Cardiac and stroke mortality.

Results

From time-dependent Cox proportional hazard regression models, folic acid consumption from vitamin supplements, above or below the standard 1000 μg dose was not associated with decrease or increase in cardiac mortality (P = 0.53 above vs. standard dose and P = 0.46, below vs. standard dose). There was also no association between folic acid consumption and mortality from stroke (P = 0.27, above vs. standard dose and P = 0.64, below vs. standard dose).

Conclusion

Consumption of higher than the standard 1000 μg prescribed dose of folic acid was not beneficial in reducing cardiac or stroke mortality in hemodialysis patients. Similarly, consumption of lower than standard dose was not associated with an increase in either cardiac or stroke mortality.

Introduction

Mortality rates due to cardiovascular disease (CVD) are approximately 15 times higher among patients with chronic kidney disease (CKD) than in the general population ^1-3 CVD accounts for almost half of all deaths in hemodialysis patients ^1-3. Reducing cardiac risk factors for hemodialysis patients is therefore of major importance for them. In CKD, risk factors for CVD morbidity and mortality in CKD include unmodifiable risk factors such as older age, white race, male gender, as well as modifiable risk factors such as higher levels of low density lipoprotein and lower levels of high density lipoproteins, dyslipidemias including elevated serum lipoprotein(a) and apolipoprotein(a) isoforms and lipoprotein remnants, hypertension, diabetes mellitus, inflammation, infection, and anemia and other hemodynamic and metabolic abnormalities ³. Elevated circulating levels of homocysteine (a sulfur containing amino acid that is a vascular toxin) is also an established risk factor for CVD in CKD patients ³.

More than 90% of patients with CKD have mild to moderately elevated levels of homocysteine (in the range of 20-80 umol/L) ^4-10. The precise mechanisms for these elevated levels in CKD are not clear but include deficiencies of folic acid, vitamins B6 and B12 due to low intake, loss of these vitamins during hemodialysis, disorders in their metabolism, and the presence of uremic toxins ^8-10. Administration of folic acid, vitamins B6 and B12 reduces serum homocysteine levels in patients with normal renal function, but the effects on mortality due to CVD are variable ^11-24. High doses of folic acid (5-60 mg/day), vitamin B6 (up to 100 mg/day), and vitamin B12 (up to 1 mg/day) can reduce homocysteine levels in CKD patients, including those on hemodialysis. While homocysteine levels can be reduced by 30-50%, there is little benefit of folic acid doses greater than 15 mg/d and less than 10% of them achieve levels in the normal range ^25-35.

Since the 4 existing randomized trials of the effects of folic acid on CVD mortality in patients with CKD have been inconclusive and few hemodialysis patients were included in these CKD studies, we conducted a secondary analysis using data from the Hemodialysis (HEMO) Study to assess the relationship between the use of folic acid containing vitamin supplements and cardiac and stroke mortality in patients on hemodialysis ^36-39. The HEMO study was a large randomized clinical trial sponsored by the National Institutes of Health (NIH), conducted from 1995 through 2001 ⁴⁰. First, we assessed the association between use of folic acid vitamin supplements and cardiac and stroke mortality. Second, we investigated whether higher or lower doses of folic acid compared to the standard 1000 μg dose recommended clinically were associated with lower cardiac and stroke mortality ^41-42.

Methods

HEMO Study Design and Population

The design, methods, and primary outcomes of the HEMO Study are described elsewhere ⁴⁰. This prospective, multi-center, randomized clinical trial, which was sponsored by the National Institutes of Health’s National Institute of Diabetes, Digestive and Kidney Disease, compared effects of two dialysis doses and two dialysis membrane fluxes on morbidity and mortality in hemodialysis patients. Between 1995 and 2001, 1846 CKD patients of all races between the ages of 18-80 who underwent thrice weekly in-center maintenance hemodialysis therapy at 15 clinical center sites across the country participated in the trial. The mean years of follow-up were 2.84 years and the annual mortality rate was 16.6% ⁴⁰.

One thousand eight hundred and forty-six hemodialysis patients were randomized in the HEMO Study. This secondary analysis included 1790 patients. Forty-eight patients were excluded because they died of CVD in the first 6 months of the follow-up period in the study and an additional 8 were excluded due to missing vitamin supplement data forms during baseline and follow-up.

Use of Vitamin and Mineral Supplements in the HEMO Study

After randomization, all study patients were provided with a free high potency, high folic acid (1000 μg) renal formulated B-complex vitamin with vitamin C, “Nephro-Vite Rx” (R&D Laboratories). Patients were encouraged, but were not required, to use the “Nephro-Vite RX” supplement, and some chose not to do so. In addition, patients were permitted to use over-the-counter or other prescribed vitamin and mineral supplements. Reported intakes of “Nephro-Vite RX”, other physician prescribed vitamin and mineral supplements, and others the patients chose to take on their own were ascertained once at baseline and annually thereafter during follow-up by the HEMO study dietitian during patient interviews. The name, dosage and unit (ie. mg, μg, ml), amount, frequency (day, week, month), and form (tablet, liquid, intravenous) of such vitamin and mineral supplements that were prescribed and consumed by the patient were recorded onto a HEMO Study Diet Prescription and Supplement Documentation Form once at baseline and annually thereafter during follow-up. Nutrient values for each supplement were calculated based on the dosage indicated by the physician prescription, as determined by review of patient’s medical chart, and by the patient’s reported consumption.

Intake of Folic Acid from Vitamin Supplements

In this secondary analysis, intakes of folic acid from all vitamin supplements used by each patient were determined using the data derived from the supplement documentation form at baseline and annually thereafter during follow-up. Imputed values calculated for missing folic acid dosages were estimated using a proportions based median imputation method developed for the use in the HEMO Study that determined the proportions and respective medians of the available dosage data and then randomly assigned imputed values to replace the missing dosages.

Cardiac and Stroke Mortality

The methods used for the collection, classification and validation of causes of death in the HEMO study are published in detail elsewhere ^{40, 43}. Briefly, each death during follow-up was classified by the investigator at the clinical center using a standardized classification system devised by the HEMO study and reported to the Data Coordinating Center. The data that form the basis for the death reports included hospital records, ICD-9 codes, physician accounts, death certificates, and death notification forms to the United States Renal Data System. Causes of death were classified into 24 categories, four of which were cardiac causes including ischemic heart disease, congestive heart failure, arrhythmia and conduction problems, and other heart diseases and conditions. Cerebral vascular (stroke) was also classified as a death outcome. Death classifications were then audited by two members of the study’s Outcome Review Committee. Cause of death was adjudicated with the full Outcome Review Committee if reviewers were in disagreement with the initial classification.

Description of Covariates

The potential confounders in the analyses included baseline age, gender, race, intervention effects (dose of dialysis and membrane flux), total serum cholesterol (measured at each individual laboratory of the clinical centers), and serum albumin (pre-dialysis serum concentrations of albumin were measured in a central laboratory [Spectra East, Rockleigh, NJ] using the nephelomtery method). Additionally, baseline scores for the Index of Co-Existing Disease (ICED) and Index of Disease Severity (IDS) were also considered as potential confounders. The ICED Score is an estimate of the patient’s comorbidity status which is calculated using a composite comorbidity coding system that classifies the presence and severity of diseases and the impact of the diseases on physical function ⁴⁴. Scores of 0, 1, 2, or 3 represent normal, mild, moderate, or severe disease comorbidity, respectively. Comorbid conditions in the ICED Score include ischemic heart disease, congestive heart failure, arrythmias and conduction problems, other heart disease and conditions, hypertension, cerebral vascular disease, peripheral vascular disease, diabetes mellitus, respiratory disease, musculoskeletal and connective tissue disease, nonvascular nervous system disease, gastrointestinal disease, hepatobiliary disease, urinary tract disease, malignancy, ophthalmologic conditions, hematologic conditions, and anticoagulation therapy. Scores were assigned based on review of comorbidity from the patient’s medical record by a trained data abstractor. IDS scores were the single components of the ICED scores and followed the same scoring system as the ICED ⁴⁴. The IDS components included congestive heart failure, arrhythmias and conduction problems, cerebral and peripheral vascular disease, ischemic and other heart disease, respiratory disease, nonvascular nervous system disease, gastrointestinal and hepatobiliary disease, and malignancy.

Statistical Analysis

Descriptive statistics on the demographic and clinical characteristics of the population were carried out and the amount of folic acid consumed from vitamin supplements was determined. A P-value < 0.05 was considered statistically significant. Time-dependent Cox proportional hazard models were used to examine the intake of folic acid from vitamin supplements on cardiac and stroke mortality. The total folic acid consumed from vitamin supplements was presented as a continuous variable (for every 100 μg folic acid). Consumption of folic acid from vitamin supplements was also examined categorically; <1000 μg (0-999 μg), 1000 μg (the most commonly prescribed amount which was used as the reference group) and > 1000 μg. For the category of < 1000 μg folic acid, those with no use of folic acid supplements (0 μg) were combined with those who did use folic acid supplements but at dosages less than 1000 μg, since the groups with no intake and less than 1000 μg were small during the study follow-up years (e.g. n=109 [8%] for no intake and n=68 [5%] for those with < 1000 μg during follow-up year 1) and would have been insufficient for analytical purposes if they were treated as individual groups. Confounders listed previously, were entered individually in the time-dependent Cox proportional hazard models. These analyses utilized imputed values for missing folic acid dosages. All calculations were performed using the statistical software package SAS version 9.1 (SAS Institute, INC., Cary, NC).

Results

Table 1 presents the baseline clinical and demographic characteristics of the patients without a cardiac or stroke mortality outcome, with cardiac mortality, and with stroke mortality. Patients who died from cardiac causes and stroke tended to be older (P < 0.0001) and to have coexisting conditions of greater severity than those who did not (P < 0.0001). Cardiac related comorbidities were common; those with cardiac and stroke deaths had more ischemic and other heart disease, congestive heart failure, cerebral and peripheral vascular disease, arrhythmias and conduction problems, and conditions of greater severity than those without these causes of death (P < 0.0001).

Table 1.

Baseline characteristics of hemodialysis patients in the HEMO Study

Characteristics	Cohort without CVD or stroke mortality event (n=1436)	Cohort with CVD mortality event (n=292)	Cohort with stroke mortality event (n=62)	P†
Age (mean±SD years)	56±14	63±11	61±14	<0.0001
Female sex (%)	57	56	63	0.60
Black race (%)	64	56	69	0.02
Serum albumin (mean±SD gm/dL)*	3.9±0.3	3.8±0.3	3.8±0.3	0.07
Total cholesterol (mean±SD mg/dL)**	172±40	178±44	176±42	0.09
Index of Coexisting Disease (ICED) scores				<0.0001
none (%)	0	0	0
mild (%)	39	23	28
moderate (%)	31	35	24
severe (%)	30	42	48
Index of Disease Severity (IDS) score
Arrhythmia and conduction problems				<0.0001
none (%)	72	57	53
mild (%)	18	26	32
moderate (%)	7	13	10
severe (%)	3	4	5
Congestive heart failure				<0.0001
none (%)	64	45	52
mild (%)	26	37	30
moderate (%)	8	16	16
severe (%)	2	2	2
Cerebral and peripheral vascular diseases				<0.0001
none (%)	67	50	52
mild (%)	18	23	24
moderate (%)	11	15	14
severe (%)	4	12	10
Cerebral vascular disease				<0.0001
none (%)	83	71	73
mild (%)	4	8	5
moderate (%)	12	20	19
severe (%)	1	1	3
Ischemic and other heart disease				<0.0001
none (%)	31	13	19
mild (%)	62	72	68
moderate (%)	6	14	8
severe (%)	1	1	5
Respiratory disease				0.06
none (%)	86	81	76
mild (%)	7	11	14
moderate (%)	6	7	10
severe (%)	1	1	0
Malignancy (%)	5	8	7	0.04
Nonvascular nervous system disease				0.20
none (%)	64	58	60
mild (%)	18	21	22
moderate (%)	18	20	18
severe (%)	0	1	0
Gastrointestinal and hepatobiliary disease				0.18
none (%)	55	49	42
mild (%)	20	22	24
moderate (%)	21	24	26
severe (%)	4	5	8
High Kt/V group (%)	50	50	44	0.60
High flux group (%)	51	46	45	0.20

^†ANOVA was used to test differences among means for the continuous variables; Fisher’s exact test was used to compare frequency data among groups for ICED and IDS scores for cerebral vascular disease, ischemic and other heart disease, respiratory disease, and nonvascular nervous system disease; Chi-square test was used to compare all other frequency data among groups; P < 0.05 was considered statistically significant

^*n=1009 for group with no CVD or stroke mortality, n=175 for group with CVD mortality, n=43 for group with stroke mortality

^**n=1287 for group with no CVD or stroke mortality, n=277 for group with CVD mortality, n=60 for group with stroke mortality

Figure 1 presents the percent of patients consuming <1000 μg, 1000 μg and > 1000 μg of folic acid from supplements at baseline and at follow-up years 1-3. The minimum standard clinical recommendation at the time for supplementation among these hemodialysis patients was 1000 μg folic acid ^40-41. Most patients consumed 1000 μg, especially during follow-up, when patients were provided with the “Nephro-Vite RX” supplement containing 1000 μg of folic acid. Baseline mean ± SD folic acid intake from supplements was 814 ± 681 μg compared to 1007 ± 652 μg at follow-up year 1, 1093 ± 1232 μg at follow-up year 2, and 1110 ± 1208 μg during follow-up year 3. Median folic acid intake from supplements was 1000 μg at baseline and all 3 years of follow-up. Among patients who consumed greater than 1000 μg folic acid, the median dose was 2000 μg at baseline and it remained so during follow-up years 1-3. For those who took < 1000 μg folic acid, the median dose was 0 μg for baseline and at follow-up year 1, at follow-up year 2 it was 71 μg and 0 μg at follow-up year 3.

Figure 1.

Percent of patients taking <1000 μg, 1000 μg (standard dose) and > 1000 μg of folic acid from vitamin supplements at baseline and follow-up

*n=36 with missing folic acid dosage data

Table 2 presents the results for the time-dependent Cox proportional hazards regression models for folic acid intake from vitamin supplements on cardiac mortality. Neither the continuous nor categorical measures of folic acid consumed had statistically significant effects on cardiac mortality. The findings for cardiac mortality remained non-significant after controlling for confounders. Figure 2 shows the cardiac mortality survival in time-dependent Cox proportional hazards model for consumption of the 3 folic acid supplement groups during the 3 year follow-up period. Those consuming < 1000 μg folic acid had the highest cardiac mortality followed by those consuming 1000 μg, and the > 1000 μg group with the lowest cardiac mortality, although these differences in survival rates were small and not statistically significant.

Table 2.

Time-dependent Cox proportional hazards regression analyses for folic acid consumption and cardiac and stroke mortality^*

	Consumed folic acid (continuous)	Consumption of folic acid (categorical)
		<1000 μg vs. 1000 μg folic acid	>1000 μg vs. 1000 μg folic acid
Cardiac mortality
n**	206	47	14
Estimate	-0.011	0.13	0.17
Hazard Ratio (95% CL)	0.99 (0.97, 1.01)	1.14 (0.81, 1.62)	1.18 (0.69, 2.02)
P-value	0.25	0.46	0.53
Stroke mortality
n**	45	9	5
Estimate	0.02	-0.18	0.53
Hazard Ratio (95% CL)	1.02 (1.00, 1.03)	0.83 (0.38, 1.80)	1.70 (0.66, 4.39)
P-value	0.06	0.64	0.27

^*Results presented are for overall models; models remained non-significant after confounders (baseline age, gender, race, dose of dialysis, membrane flux, total serum cholesterol, serum albumin, ICED and IDS) were considered (data not shown)

^**n=event n

n=1790 patients at baseline

n=4628 patient years

Figure 2.

Cardiac survival by time-dependent consumed folic acid vitamins supplement group

For stroke mortality, the association between the categorical measure of folic acid intake and stroke mortality were not statistically significant (see Table 2). The results remained even after adjusting for all confounders. However, higher folic acid intakes from vitamins were associated with a slightly borderline significant increased stroke mortality risk (HR=1.02, P = 0.06) when the continuous measure of folic acid intake was used, as shown in Table 2. That is, for every 100 μg of total folic acid consumed at each annual collection date (baseline and follow-up years 1-3), patients had a 2 % increased risk of stroke mortality (P = 0.06). This association remained even after controlling for virtually all of the confounders (see Figure 3). The association became significant after controlling for ischemic and other heart disease (HR=1.02, P = 0.04). Conversely, after controlling for serum albumin, the association disappeared (HR=1.01, P = 0.28). Figure 4 shows the survival curves for stroke by time-dependent Cox proportional hazards models for consumption of the lower, standard and higher dose supplement groups during the 3 year follow-up period. Although not statistically significant, those consuming > 1000 μg folic acid had the highest stroke mortality followed by those consuming 1000 μg, and the < 1000 μg group had the lowest stroke mortality.

Figure 3.

Time-dependent Cox proportional hazards regression analyses for total consumed folic acid from vitamins and stroke mortality controlling for confounders in HEMO study patients

Figure 4.

Stroke survival by time-dependent consumed folic acid vitamins supplement group

Discussion

The benefits of high potency renal formulated B-complex vitamins in protecting the nutritional status of hemodialysis patients from losses of B vitamins during dialysis are well known and justify their use ^40-41. However, the utility of these vitamins in preventing cardiac or stroke mortality in hemodialysis patients is uncertain. We failed, as have others, to demonstrate that intake of folic acid containing vitamin supplements at the levels used in the HEMO Study reduced either cardiac or stroke mortality ^{12, 14-19, 36, 38}. We were also unable to demonstrate that doses greater than standard dose of 1000 μg of folic acid compared to standard dose had beneficial effects, or that lower doses of the vitamin or no intake had adverse effects on cardiac or stroke mortality compared to a standard dose intake.

The possible adverse association between folic acid and stroke mortality was only of borderline significance, however, it became non-significant after including serum albumin as a potential confounder. Malnutrition, indicated by hypoalbuminemia is highly prevalent in hemodialysis patients as are inflammation and atherosclerosis ^45-46. All of these conditions are associated with increased mortality risk in CKD patients ^{45, 47-50}.

While this study was a secondary analysis, it included the largest sample size of hemodialysis patients and the longest follow-up time to date in examining the effects of folic acid containing vitamin supplements use on cardiac and stroke mortality. Another unique strength of the analyses was its ability to define and identify the causes of mortality precisely among HEMO study patients and to use the ICED and IDS scores to quantify and explore the severity of existing comorbidities.

Our study had certain limitations. For the cardiac mortality analyses, with our sample size, the small percent (6%-9% during baseline and follow-up years 1-3) of patients consuming the higher dose of folic acid (> 1000 μg) and a cardiac mortality rate of 16%, we had only 54% power to detect a hazard ratio of 1.5 with an alpha of 0.05 and 80% power with a hazard ratio approaching 1.75. In our stroke analyses, with a relatively low stroke mortality rate of 3.4%, our sample size was even more limited in power for detecting an association between folic acid consumption and stroke mortality. Furthermore, combining the cardiac and stroke mortality outcomes would not have substantially improved the power of the analyses. The discrete nature of the folic acid vitamin supplement doses may also have been a limitation. Because the majority of the HEMO study patients took the standard dosage of 1000 μg, the dosage data had a smaller variance with dosages centered on 1000 μg and this may have impeded our ability to detect a possible beneficial effect of the higher folic acid doses on cardiac and stroke mortality. Another limitation was that folic acid intakes from other sources such as food and oral enteral nutritional supplements were not included and therefore the effect of total folic acid intakes on cardiac and stroke mortality could not be examined. If dietary folate varied markedly from person to person, this may have introduced a problem. The absence of a true control may have also hindered the observation of the effects of folic acid on cardiac and stroke mortality outcomes. Finally, plasma homocysteine and serum folate levels were not available on all HEMO study patients and so the relationship between folic acid, homocysteine, serum folate and subsequent cardiac and stroke mortality could not be ascertained.

Several large clinical trials are currently underway that address the use of folic acid as a homocysteine lowering therapy to reduce CVD related risks and mortality among patients with normal renal function and renal transplants ^51-55. Although our results were negative, further examination of the association between folic acid containing vitamin supplements and CVD mortality is warranted in hemodialysis patients, because of the high risk of these outcomes in this population.