Evidences from randomized clinical trials and meta-analysis have claimed an association between the use of soluble dietary fiber from psyllium and a cholesterol-lowering effect. However, there is still uncertainty as to the dose–response relationship and its long-term lipid-lowering efficacy. This meta-analysis was primarily conducted to address the dose–response relationship between psyllium and serum cholesterol level and time-dependent effect of psyllium in mild-to-moderate hypercholesterolemic subjects.
Twenty-one studies, which enrolled a total of 1030 and 687 subjects receiving psyllium or placebo, respectively, were included in the meta-analysis. The studies were randomized placebo-controlled trials, double blinded or open label, on subjects with mild-to-moderate hypercholesterolemia. The dose of psyllium was between 3.0 and 20.4 g per day and intervention period was more than 2 weeks. Any type of diet background was permitted. Diet lead-in period was between 0 and 8 weeks.
Compared with placebo, consumption of psyllium lowered serum total cholesterol by 0.375 mmol/l (95% CI: 0.257–0.494 mmol/l), and LDL cholesterol by 0.278 mmol/l (95% CI: 0.213–0.312 mmol/l). With random-effect meta-regression, a significant dose–response relationship were found between doses (3–20.4 g/day) and total cholesterol or LDL cholesterol changes. Regression model of total cholesterol was −0.0222+0.2061 × log (dose+1), and that of LDL cholesterol was 0.0485+0.1390 × log (dose+1). There was a time effect of psyllium on total cholesterol (equation: 6.3640–0.0316 × treatment period) and on LDL cholesterol (equation: 4.3134–0.0162 × treatment period), suggesting that psyllium reduced serum total cholesterol more quickly than LDL cholesterol.
Psyllium could produce dose- and time-dependent serum cholesterol-lowering effect in mild and moderate hypercholesterolemic patients and would be useful as an adjunct to dietary therapy for the treatment of hypercholesterolemia.
In recent years, there has been growing interest in the use of dietary fibers in health maintenance and disease prevention (Petchetti et al., 2007). It has been speculated that a deficiency of soluble fibers in the Western diet might be contributing to the epidemics of coronary heart disease (CHD), diabetes mellitus and colonic cancer. Hypercholesterolemia has been identified as a prominent independent risk factor in the development of CHD. Soluble fibers, including those from psyllium husk, have been shown to augment the cholesterol-lowering effects of a low-fat diet in persons with hypercholesterolemia. Increasing dietary fiber has been recommended as a safe and practical approach for cholesterol reduction (NCEP ATPIII, 2001 and Brunner et al., 2007).
Psyllium is one of the highest sources of soluble mucilaginous dietary fiber derived from seed husk of Plantago psyllium, an annual plant grown in Mediterranean region, India, China and other regions. It is mainly marketed in two forms, an over-the-counter bulk-forming laxative under the trade name Metamucil and psyllium-enriched health foods (Singh, 2007). Psyllium is considered to be useful as an adjunct to dietary therapy (step 1 or step 2 American Heart Association (AHA) diet) in the treatment of patients with mild-to-moderate hypercholesterolemia. The earliest clinical study (Garvin et al., 1965) in a non-placebo-controlled pattern suggested a significant cholesterol-lowering effect of psyllium (Metamucil). Then the first double-blind, placebo-controlled trial of psyllium was performed and it showed its effectiveness in mild-to-moderate hypercholesterolemia (Anderson et al., 1988). Subsequently, numerous clinical trials have investigated the treatment effect of psyllium as an adjunct to dietary modification. There was a universal feeling in all these investigators that psyllium was a hypocholesterolemic agent. However, most of these studies were poorly designed and only enrolled a small number of patients; thus the results should be considered inconclusive. In 1997, Olson et al. (1997) conducted a meta-analysis involving 404 subjects with mild-to-moderate hypercholesterolemia and found a beneficial effect of psyllium-enriched cereal products on blood total cholesterol and low-density lipoprotein (LDL) cholesterol. Psyllium in a form other than cereal (for example, crackers, bulk laxative) was not included in this study. More recently, Brown et al. (1999) performed another meta-analysis to quantify the cholesterol-lowering effect of major dietary fibers including pectin, oat bran, guar gum and psyllium and found that various soluble fibers reduced total and LDL cholesterol by similar amounts, but the effect was small within the practical range of intake. The latest meta-analysis, conducted on individuals with mild-to-moderate hypercholesterolemia, demonstrated that psyllium supplementation could significantly lower serum total and LDL cholesterol concentrations in subjects consuming a low-fat diet (Anderson et al., 2000a).
Despite the wealth of information that is available on the association between the consumption of soluble dietary fiber from psyllium and a cholesterol-lowering effect, the dose–response relationship and its long-term lipid-lowering efficacy have not been defined. Thus it was of great interest to conduct a meta-analysis with more updated evidence to precisely determine the effect size of psyllium on decreasing serum lipids and explore the dose–response relationship and long-term effects of psyllium on serum total cholesterol and LDL cholesterol levels in individuals with mild-to-moderate hypercholesterolemia.
Materials and methods
Identification of previous studies
A computerized literature search was conducted on the MEDLINE, EMBASE, BIOSIS and PASCAL databases, using the terms psyllium or dietary fiber and blood lipids to identify the clinical trials involving psyllium from 1966 to August 2005. In addition, FDA document, published reviews, reference lists from clinical trials and conference abstracts were also examined.
For inclusion in the meta-analysis, studies had to meet the following criteria: (1) Trials were conducted on human adults for the treatment of mild-to-moderate hypercholesterolemia, not secondary to any recognized cause; (2) they were controlled and had either a randomized crossover or a parallel design; (3) they provided lipid changes in the treatment and control groups to permit the calculation of the treatment effect; (4) they had a minimum intervention period of 2 weeks; (5) they had a low-fat diet or controlled usual diet background. Disagreements regarding criteria for admission to meta-analysis were resolved by discussion and consensus. In all 22 of 39 studies were selected for special review. Two studies (Stoy et al., 1993; Spence et al., 1995), which met above criteria, were excluded because they did not provide enough data to compute the effect size and one (Wolever et al., 1994) was excluded because subjects in the study were already represented in another published study by the same group. As two research reports (Sprecher et al., 1993; Jenkins et al., 1997) involved two sub-groups, the data were treated as two separate studies.
Data were extracted from the published reports by two independent reviewers and disagreements resolved by discussion with the third reviewer. For each trial, the followings were documented: country of origin, study population, number and type of subjects, initial cholesterol concentration, type of psyllium preparation and used dose regimen, nature of control group, background diet, method of allocation, extent of blinding, study design (parallel or crossover), treatment length, method of analysis, laboratory method used to measure lipid levels, withdrawal and so on.
The value of total cholesterol, LDL cholesterol, high density lipoprotein (HDL) cholesterol and triglycerides were presented in units of mmol/l. Results reported in mg/dl were converted to mmol/l. For studies with parallel group designs, lipid-lowering effect size was calculated by subtracting the mean change in the control group from that in the treatment group. The mean change was mean baseline subtracted by mean end point value. For crossover studies, the effect size was represented by the difference of post-treatment lipid concentrations of the treatment and control periods. A technique described by Anne (2002) was used to pool the effect sizes from individual studies. A positive value indicated a greater lipid reduction relative to placebo.
Tests of heterogeneity were performed using Cochrane Q test to determine whether they estimated for common treatment effect. The significant level was α=0.2. But to determine whether or not the Q statistic was significant, a random-effect model was assumed.
Pooled effect size and its 95% confidence interval were estimated by a fixed-effect or random-effect model. The subgroup analyses were performed by dose (10.2 g/day), psyllium-enriched foods, background diet (step I diet), treatment period (8 weeks), design type (parallel design) and clinical trial conducted in USA. Difference of lipid-lowering effect between psyllium-enriched foods and bulk laxative was tested by meta-regression (Berkey et al., 1995; Houwelingen et al., 2002). The possibility of publication bias was detected with funnel plot (Egger et al., 1997).
When more than one dose was studied, the mean lipid change across all doses was used to provide an average effect size. However, in the dose–response analysis, each dose was represented separately. Only clinical trials with parallel group design were included in dose–response analysis, as the net change of lipid could only be estimated in this kind of trials.
A random-effect meta-regression was performed to address dose–response relationships. The long-term effect of psyllium was explored by mixed-effect model for repeated measure data.
All statistical analyses were performed using software package SAS 9.13.
Description of trials
Study characteristics of all published trials included in the meta-analysis were listed in Table 1. The studies were undertaken in following five countries: United States, United Kingdom, Canada, Australia and Mexico. They were either parallel or crossover design. Subjects were mild-to-moderate hypercholesterolemia (total cholesterol >4.78 mmol/l or LDL cholesterol >3.36 mmol/l). Four studies had no dietary lead-in period, whereas others had 2–8 weeks dietary lead-in period. Most subjects consumed either a low-fat diet or an NCEP step I or II diet during diet lead-in and therapy periods, whereas those in four studies consumed controlled usual diet. The therapy period ranged from 14 to 182 days. The form was psyllium bulk laxatives or psyllium-enriched foods with the dosage of 3–20.4 g/day. Data from 1717 persons with hypercholesterolemia were included in this meta-analysis.
Effects of psyllium on lipid
In the full dose range, psyllium significantly reduced both serum total cholesterol and LDL cholesterol concentrations: 0.375 mmol/l (95% CI: 0.257–0.494 mmol/l) and 0.278 mmol/l (95% CI: 0.213–0.312 mmol/l) respectively, among adults with mild-to-moderate hypercholesterolemia (Table 2). The heterogeneity test of total cholesterol was statistically significant, and that of LDL cholesterol was non-significant. Psyllium also significantly reduced HDL cholesterol, but by a much smaller amount: 0.0353 mmol/l (95% CI: 0.0003–0.0514 mmol/l). Psyllium intake did not significantly affect serum triacylglycerol concentrations. Figures 1 and 2 showed the net effect of consumption of psyllium on total cholesterol and LDL cholesterol levels for each study included in the meta-analysis. To explore heterogeneity, funnel plots for total cholesterol and LDL cholesterol were drawn and shown in Figure 3. Egger test displayed that asymmetry=−0.3999 (P=0.7058) for total cholesterol, suggesting that publication bias was not confirmed, but the phenomenon of ‘large effect and small sample’ could be concluded. The asymmetry of LDL cholesterol was 0.9424 (P=0.0399), indicating that both publication bias and the phenomenon of ‘large effect and small sample’ existed simultaneously.
Dose–response relation of psyllium
There was a significant dose–response relationship between doses (3–20.4 g/day) and changes of serum total cholesterol or LDL cholesterol levels. The regression between log term of dose and measurement was estimated with random-effect regression model. The regression model of total cholesterol was −0.0222+0.2061 × log (dose+1), and that of LDL cholesterol was 0.0485+0.1390 × log (dose+1). Although the two models were both statistically significant (both the P-values were smaller than 0.0001), the regression coefficients were relatively small and the independent variable was log-transformed. Thus greatly increasing dosages of psyllium might not correspondingly produce a clinically significant difference in the lipid-lowering effect across a dose range of 3–20.4 g/day. On the other hand, we fitted a weighted linear model to explore the relationship between the percent reduction of LDL cholesterol (calculated as (baseline−end point)/baseline×100%) and dose of psyllium (Figure 4). Results showed that there was a moderate correlation, and the regression equation was: 2.200+0.685 × dose (P=0.0454). According to this equation, consumption of psyllium 5, 10 and 15 g/day could result in 5.6, 9.0 and 12.5% decrease of LDL cholesterol level respectively.
Long-term effects of psyllium on lipid
There was a long-term effect of psyllium on total cholesterol and LDL cholesterol (Figure 5). The equation was fitted using a mixed-effect model for a repeated measurement. Throughout the treatment period, serum total cholesterol and LDL cholesterol levels of psyllium group reduced faster than that of the placebo group (P<0.0001). The equation of total cholesterol was 6.3640–0.0316 × treatment period, and that of LDL cholesterol was 4.3134–0.0162 × treatment period, suggesting that, along with time, total cholesterol reduced more quickly than LDL cholesterol.
Type of intervention
Psyllium, as either bulk laxatives or enriched foods, could significantly reduce serum total cholesterol and LDL cholesterol levels. The effect sizes of psyllium as bulk laxatives on total cholesterol and LDL cholesterol were 0.442 mmol/l (95% CI: 0.210–0.674 mmol/l) and 0.300 mmol/l (95% CI: 0.167–0.430 mmol/l) respectively, whereas that of psyllium as enriched foods were 0.320 mmol/l (95% CI: 0.162–0.477 mmol/l) and 0.260 mmol/l (95% CI: 0.180–0.340 mmol/l), respectively. The form of psyllium seemed to have no significant influence on lipid-lowering effects (P=0.2367 for total cholesterol and P=0.5688 for LDL cholesterol).
Effect size of each subgroup was similar to the pooled effect size of all studies in this meta-analysis (Table 3).
In this meta-analysis of 21 studies, subjects who administrated psyllium bulk laxatives or enriched foods had significantly decreased total cholesterol, LDL cholesterol and HDL cholesterol concentrations. Compared with placebo control, the mean decrease of total cholesterol, LDL cholesterol and HDL cholesterol were 0.375, 0.278 and 0.0353 mmol/l, respectively. But for triglycerides, there was little effect.
We established a dose–response relationship between psyllium and serum total and LDL cholesterol levels. However, it was a logarithmic relationship, which indicated that greatly increasing dosage of psyllium did not produce a correspondingly large effect size. Although psyllium was well tolerated, high dosage (20.4 g per day) might cause diarrhea (Neal and Balm, 1990).
Long-term intake of psyllium could cause sustained decrease in the total and LDL cholesterol levels. It had been estimated that a fall of 0.6 mmol/l in total cholesterol would reduce subsequent cardiovascular risk by 50% among those aged 40 and by 20% among those aged 70 (Betteridge, 1994). Every 1% reduction in LDL cholesterol could reduce the risk of CHD by 1.2–2.0% (Katan et al., 2003). Based on the result of this meta-analysis, after consuming psyllium for 20 weeks, serum total cholesterol level could be reduced from baseline level (about 6.36 mmol/l) to 5.73 mmol/l and LDL cholesterol concentration could be dropped from 4.31 to 4.0 mmol/l with the decrease rate as 9%. And after psyllium consumption for about 1 year and a half, the LDL cholesterol could get to 3.1 mmol/l (the upper limit of the reference range).
CHD is a major cause of death in United States and most Western countries. Blood cholesterol is a major risk factor of coronary heart disease. Dietary and pharmacological reductions in total and LDL cholesterol decrease the risk of coronary events, so dietary has been recommended as a safe and practical approach for cholesterol reduction. Psyllium is one of the most effective lipid-lowering agents. In 1998, FDA ruled that labels on certain foods containing soluble fiber from psyllium seed husk, such as certain breakfast cereals, might claim that these foods, as part of a diet low in saturated fat and cholesterol, may reduce the risk of coronary heart disease. Newer dietary approaches combining cholesterol-lowering drugs or foods may offer another option. The synthesized effects of a combination therapy including psyllium and low-dose statins (Moreyra et al., 2005) or plant sterols (Shrestha et al., 2006) have been demonstrated in clinical trials.
We believed that the results of this meta-analysis represented a conservative estimate of psyllium effect on lipid-lowering. We could conclude that there was a publication bias in this meta-analysis, based on the results of funnel plots. The publication bias was mainly caused by ‘large effect but small trial’ phenomenon. The effect size would be larger if all those small trials were published and selected into the meta-analysis.
Consuming psyllium as a bulk-forming fiber laxative is functionally equivalent to consuming psyllium-enriched foods. Clinical studies evaluating psyllium administered as bulk laxatives were conducted in a fashion similar to the studies conducted with psyllium-enriched foods, including patient selection criteria and regimen, thus efficacy should be comparable for bulk laxative and enriched foods studies. The results of this meta-analysis confirmed that the effect of these two forms on serum lipid levels has a non-significant difference. Having both available options would provide alternatives to improve the compliance of a fiber intake recommendation. Moreover, psyllium is one of the most commonly used over-the-counter drug for the treatment of chronic constipation, especially in the elders who have a high incidence of hypercholesterolemia, thus psyllium would provide another potential benefit for them due to its efficacy in lowering serum cholesterol levels at the dose in common use.
This work was partially supported by Cooperative Program of Rhone-Alps Region and Shanghai. We thank Professor Jean-Pierre Boissel and Dr Michael Cucherat (Institute of Clinical Pharmacology, Lyon University I, France) for expert advice.
About this article
Current Nutrition Reports (2014)