Professional Documents
Culture Documents
net/publication/45185418
CITATIONS READS
17 1,141
1 author:
Richard Singer
The Canberra Hospital
33 PUBLICATIONS 262 CITATIONS
SEE PROFILE
All content following this page was uploaded by Richard Singer on 13 July 2017.
The Canberra Hospital Renal Service, Australian National University Canberra, ACT, Australia
Correspondence and offprint requests to: Richard Francis Singer; E-mail: richard.singer@act.gov.au
Abstract
Background. Vitamin C (ascorbate) deficiency and symp- using the Kidney Dialysis Quality of Life-Short Form
toms consistent with deficiency (fatigue, myalgia, dyspnoea, (KDQOL-SF ™ ) symptom subscale, and the study was
gingivitis, cardiovascular instability and depression) are 80% powered to detect a change of 10 in the KDQOL-SF™.
common in patients with renal failure. This study aimed to Results. Ninety-nine subjects were randomized, and ascor-
determine if supplementation with ascorbate in patients with bate deficiency was present in 40% at baseline. Mean symp-
severe renal failure improved symptoms or cardiovascular tom scores at follow-up were similar in the two groups
stability, or was associated with adverse effects. (P-value = 0.19). There was a trend to slightly worse nau-
Methods. The study was a 3-month, double-blind, rando- sea scores in the ascorbate group after controlling for the
mized trial of ascorbic acid 250 mg or matching placebo level of baseline nausea (P = 0.09), and there was no impact
given thrice weekly. Subjects were clinically stable and either on cardiovascular stability. Compliance appeared adequate
received conventional dialysis or had an estimated glomeru- at 91%, and deficiency was corrected in most (85%) of the
lar filtration rate of <20 mL/min. Symptoms were measured subjects in the active treatment group.
© The Author 2010. Published by Oxford University Press on behalf of ERA-EDTA. All rights reserved.
For Permissions, please e-mail: journals.permissions@oxfordjournals.org
Ascorbate in kidney failure study 615
Conclusions. This study indicates that ascorbate supple- Hospital databases were approached for their consent to participate. Those
mentation does not improve symptoms or cardiovascular meeting the selection criteria had baseline data collected and entered
stability in those with severe renal impairment, but is asso- randomization. Enrolment was planned to cease when 150 patients had
been enrolled, or 15 months after recruitment commenced. The study
ciated with a trend towards worse nausea. was listed with the Australia and New Zealand Clinical Trials Registry
(registry number ACTRN12608000016336) and approved by the Austra-
Keywords: ascorbic acid ; hypotension; kidney failure; quality of life; lian Capital Territory Health Human Research Ethics Committee. It was
renal dialysis conducted in accordance with the Declaration of Helsinki. The study was
part funded by the Canberra Hospital Private Practice Fund, with the
remaining funding derived internally from the Canberra Hospital Renal
Unit. All subjects provided written informed consent. The author de-
signed the study, analysed the data and wrote the paper.
Introduction
Inclusion criteria
Low vitamin C (ascorbate) levels are rare in the commu- Subjects were either receiving thrice weekly maintenance haemodialysis
therapy or receiving maintenance peritoneal dialysis therapy (defined as
nity [1] but common in both haemodialysis (HD) [2–4] having a PD catheter in situ that was being used or planned to be used
and peritoneal dialysis (PD) patients [5,6]. The detrimental within 2 months of enrolment), or had severe renal failure not on dialysis
effects of ascorbate deficiency (scurvy) have been recog- which was defined as an estimated glomerular filtration rate (eGFR) of
nized for over a hundred years, and the correlation between <20 mL/min corrected for body surface area (BSA) as measured by the
four-variable Modification of Diet in Renal Disease (MDRD) equation
blood levels and scorbutic symptoms has been delineated [15] with race assumed to be White.
in more recent human studies [7–9]. This has lead to a def-
inition of ascorbate deficiency as a level <11.4–17 μmol/L
(2–3 mg/L). A level of <23 μmol/L (4 mg/L) is considered Exclusion criteria
‘insufficient’ because deficiency would be expected after Subjects unable to give informed consent, concurrently enrolled in
several days of reduced intake [10]. another clinical intervention trial, aged less than 18 years, who were
clinically unstable, having a life expectancy <3 months, using ascorbate
In dialysis patients, ascorbate deficiency cannot be de- supplements within previous 2 weeks, not willing to abstain from non-
tected by clinical means as dietary intake correlates poorly study ascorbate supplements for the duration of the study, who lacked
with serum levels [5,6]. Further complicating clinical as- fluency in English or who had previous diagnosis of primary hyperox-
sessment in the chronic kidney disease population (CKD) aluria were excluded.
is that scorbutic symptoms such as fatigue, bleeding, ar-
thralgias, myalgias, dyspnoea, peripheral oedema, gingi- Randomization and intervention
vitis, cardiovascular instability and depression are also
Randomization was by computer-generated random number with subjects
commonly seen in uraemia. It is therefore possible that stratified according to diabetic status and by the need for maintenance
some of the symptoms commonly attributed to uraemia dialysis treatment.
are related to co-existent ascorbate deficiency. Only one Subjects were given a bottle containing 36 capsules with instructions
small study has looked at the effect of low-dose supple- to take one study capsule three times a week (post-dialysis for HD sub-
jects) for 3 months. Study drugs were compounded and packaged by an
mentation (300 mg ascorbate per week) on symptoms in
external pharmacy. Active (250 mg ascorbic acid) and placebo (lactose)
dialysis patients [11]. This small, quasi-randomized un- capsules were identical in appearance and taste. Both the subjects and
blinded study found no benefit on quality of life with investigators were blinded as to allocation until after the final subject
supplementation. There is currently no Australian guide- had completed the study, and all follow-up data had been collected. Oral
line for ascorbate supplementation in renal failure pa- dosing was used because intravenous dosing offered no advantage and
was considered impractical for non-haemodialysis subjects.
tients, but the potential risks of excessive ascorbate are
warned against in the Australian guidelines [12]. Most
previous studies indicate that ascorbate supplementation Outcomes and data collection
in dialysis patients equivalent to 120–130 mg/day does The primary outcome of the study was the effect of ascorbate supplemen-
not cause adverse effects [6,13,14], although serum oxalates tation on the Kidney Dialysis Quality of Life-Short Form version 1.3
were not measured in these studies. One study did suggest (KDQOL-SF™) symptom subscale [16]. The KDQOL-SF™ is a well-
an increase in the overall rate of adverse effects with ascor- accepted test for measuring quality of life in dialysis patients, and has
acceptable consistency, with repeated measure correlation of 0.73 for
bate [11], but the relationship to the medication in that study the symptom subscale [17]. Higher scores indicate better quality of life.
was doubtful. The KDQOL-SF™ symptom and cognitive subscales were administered
either face to face or by telephone by a single research assistant. Baseline
and follow-up data were collected prospectively with confirmation by
Materials and methods examination of the clinical record. Racial affiliation was self-assessed
by subjects. Vascular disease was defined as a history of a revasculariza-
The primary objective of this study was to determine if supplementation tion procedure or a history of acute myocardial infarction. Adequacy of
with 250 mg oral ascorbic acid thrice weekly in patients with severe renal blinding was assessed by asking subjects whether they felt they had been
failure reduced symptoms associated with ascorbate deficiency. The other given active or placebo at the completion of the study. Compliance was
objective of the study was to investigate (i) the relationship of ascorbate assessed by means of a capsule count at the completion of a 3-month fol-
deficiency and symptoms in CKD, (ii) the effect of supplementation on low-up and by analysing changes in plasma ascorbate levels. Compliance
cardiovascular stability in haemodialysis, (iii) the frequency of infections was defined as subjects taking 90% or more of the prescribed capsules.
and (iv) the effect on the rate of loss of native kidney function. Supplementation success (which encompasses both compliance and dose
The study design was a prospective, single-centre, double-blind, ran- effectiveness) was defined as a rise in ascorbate levels of >11 μmol/L at a
domized, placebo-controlled trial. All patients who appeared to meet the 3-month follow-up. A rise of this magnitude is sufficient to eliminate de-
selection criteria on the basis of information recorded in the Canberra ficiency, regardless of baseline level. Data collection methods and defini-
616 R.F. Singer
Table 1. Secondary outcome definitions and data collection methods
Cardiovascular instability A fall in blood pressure during dialysis Pre-printed sheet placed in dialysis notes,
necessitating either a fluid bolus, with confirmation from the clinical record
slowing of ultrafiltration or Trendelenburg
positioning
Bacteraemia Growth of bacteria from blood cultures Pathology Department database
Adverse effects Any event that the patient or investigator Open-ended question at a 3-month follow-up
thought might have been due to study with specific separate questioning for nausea
medication and worsening of diarrhoea
Residual renal function Residual glomerular filtration rate in Renal Unit databases
PD subjects assessed using standard
methods [23]
Three-variable MDRD equation for
non-dialysing subjects (race was
ignored) [15]
tions for the secondary outcomes in the trial are summarized in Table 1, subjects (87%) classified themselves as of ‘European’ or
with native kidney function reported as an eGFR. Ascorbate levels were ‘Australian’ race with one subject being Australian Aborigi-
measured by high-performance liquid chromatography [18] with the nal. Subjects classifying themselves as ‘Australian’ were all
plasma being centrifuged and deproteinized within 2 h.
of predominantly European descent. The subjects in each
group were well matched (Table 2), and ascorbate defi-
Statistical analysis ciency was present in 40% at baseline. Ascorbate prescrip-
Analysis was by intention-to-treat (randomized group) with all P-values tion was largely successful in eliminating deficiency at a
two-sided. The null hypothesis was rejected at a significance level of 3-month follow-up (Table 3).
<0.05. It was pre-specified that analyses were to be repeated for the sub- Mean symptom scores were not statistically significantly
group with ascorbate deficiency. A t-test was used for the comparisons of
normally distributed means. Medians were compared using the Wilcoxon different between the treatment groups at a 3-month follow-
rank-sum test. Cardiac instability was analysed using the method of up (P-value = 0.18). Because the symptom scores were not
Wei–Lin–Weissfeld [19] for correlated Cox regression; however, given perfectly normally distributed (Figure 2), the significance
that censoring was insignificant through the study, cardiac instability test was repeated using the non-parametric Wilcoxon rank-
was also analysed by the somewhat more familiar method of negative
binomial regression. The method used for calculating P-values for other sum test to check for any sensitivity of the P-value to the
data comparisons is indicated in the text by means of a superscript use of a parametric test. Very similar significance results
symbol as follows: ΞWilcoxon signed-rank test, ¶chi-square, §Fisher were found using this more conservative approach. Limit-
exact test, Θordered logistic regression and &binomial probability. Mul- ing the analysis to subjects who were ascorbate deficient at
tiple logistic regression models were compared using a process of back-
baseline suggested a non-statistically significant trend to-
ward selection with model comparison by the likelihood ratio test.
wards worse scores in ascorbate-supplemented subjects
who were deficient at baseline (absolute difference in mean
Power analysis assumptions symptom scores = 10.8, P-value = 0.07). However, this
The study was powered at 80% to detect a clinically significant change of trend was not apparent when baseline symptom scores were
10 in the KDQOL-SF™ symptom subscale with a two-tailed alpha of controlled for in a linear regression model limited to ascor-
0.05. The calculations were based on data suggesting that the distribution bate-deficient subjects at baseline (ascorbate therapy beta-
of KDQOL-SF™ symptom subscale scores in haemodialysis patients coefficient −4.47, P-value = 0.20). Limiting the analysis of
would be normally distributed, without a significant ceiling effect, with 3-month symptom scores to include only symptoms most
a mean of 71 and standard deviation of 16.6 [16,20]. Using a t-test for
analysing the difference in 3-month KDQOL-SF™ scores, a sample size relevant to adult ascorbate deficiency—muscle soreness,
of 88 was estimated to provide 80% power to detect a difference in itchy skin, dyspnoea and fatigue (questions 14a, d, f
KDQOL-SF™ of 10 with two-sided alpha of 0.05. and i in the questionnaire) [7,10,21]—did not affect the
results. There was no significant relationship between the
presence of baseline ascorbate deficiency and subscale
Results scores, nor to the severity of any individual component
of the subscales. Analysis of subjects in the active group
Numbers are expressed as medians with interquartile range as ‘before and after’ pairs did not reveal any significant
(IQR) in brackets or means with their standard errors (SEM) effect of ascorbate prescription on either cognitive or symp-
unless otherwise indicated. Figure 1 shows the flow of par- tom scores.
ticipants through each stage in the study. Baseline ascorbate Median 3-month nausea scores were slightly, but not sta-
levels were not available in three subjects. This was due to tistically significantly, worse in the ascorbate group [median
mishandling of the samples in two subjects and one subject 100 (IQR 50–100)] compared with the placebo group [me-
refusing venisection. A further one subject withdrew from dian 100 (IQR 75–100)] (P-value = 0.056). To check if this
the study after randomization and collection of an ascorbate trend was due to an effect of differing baseline scores be-
level, but before completing other baseline data. Data for all tween the groups, the nausea scores were reanalysed in an
subjects were included until their exit from the study. Most ordered logistic regression model which controlled for the
Ascorbate in kidney failure study 617
level of baseline nausea. In this model, the trend was weaker 0.87 (P-value = 0.74) in the proportional hazards model.
(P-value = 0.09). These results did not change if only ascorbate-deficient sub-
More than half of the 65 subjects on haemodialysis dur- jects at baseline were included, nor when the regression
ing the study had at least one episode of cardiovascular in- model controlled for other potential risk factors for cardiac
stability, but there was no difference between the treatment instability such as diabetes, dialysis vintage or a history of
groups (P-value = 0.36¶). There were a total of 129 cardio- cardiovascular disease.
vascular instability events during the study (range per sub- Information on native renal function was available for 30
ject 0–13). Analysis using correlated proportional hazards (88%) non-haemodialysis subjects, and the post hoc esti-
modelling and negative binomial regression yielded similar mated minimal detectable difference in native renal function
non-statistically significant results, with a hazard ratio of between groups (for beta of 0.2 and alpha of 0.05) was
618 R.F. Singer
Table 2. Baseline characteristics
#
Confirmed past myocardial infarction or arterial revascularisation.
Θ
Reported P-value is for the multivariate model controlling for baseline nausea.
4 mL/min/1.73 m2 BSA. In fact, the difference observed was Univariate predictors for the presence of baseline ascor-
smaller than this with a mean 3-month eGFR of 11.93 mL/ bate deficiency were haemodialysis status [odds ratio 3.88
min/1.73 m2 in the placebo group and 14.5 mL/min/1.73 m2 (P = 0.009)] and years on dialysis [odds ratio 1.13 (P =
BSA in the ascorbate group (P-value = 0.20). Controlling for 0.036)]. Inclusion of both variables in a multivariate logis-
baseline native renal function in a linear regression model tic regression model indicated that the best model included
did not significantly alter this finding. During the study, only a single independent variable, haemodialysis versus
there were no episodes of acute coronary syndrome, and non-haemodialysis status.
two subjects commenced dialysis, both to the PD modality. Compliance assessment was suboptimal as only 64 sub-
There were too few bacteraemias (two in ascorbate group jects returned their study medication bottles at the comple-
and one in placebo group) and dialysis access thromboses tion of the study. A total of four subjects (two in each
(one in ascorbate group) to analyse differences between group) had four or more capsules remaining at study
the groups. end. Questioning subjects who failed to return their medi-
cation bottle revealed a further two non-compliant sub-
jects, indicating that overall compliance was at best 91%.
Placebo Group Ascorbate Group Using ascorbate levels to determine the success of treat-
100
Gro Østli Eilertsen1, Silje Fismen2, Tor-Arne Hanssen2 and Johannes C. Nossent1,3
1
Department of Rheumatology, Institute of Clinical Medicine, Medical School, University of Tromsø, Norway, 2Department of
Pathology, University Hospital of Northern Norway, Tromsø, Norway and 3Department of Rheumatology, University Hospital of
Northern Norway, Tromsø, Norway
Correspondence and offprint requests to: Gro Østli Eilertsen; E-mail: gro.ostli.eilertsen@unn.no
© The Author 2010. Published by Oxford University Press on behalf of ERA-EDTA. All rights reserved.
For Permissions, please e-mail: journals.permissions@oxfordjournals.org