To more precisely quantify the effect of barley β-glucan on blood lipid concentrations in humans and to examine the factors that could affect its efficacy.
Eleven eligible randomized clinical trials published from 1989 to 2008 were identified from nine databases. Weighted mean effect sizes were calculated for net differences in lipid profile using a random effect model (RevMan 4.2).
Overall, barley and β-glucan isolated from barley lowered total and low-density lipoprotein (LDL) cholesterol concentrations by 0.30 mmol/l (95% confidence interval (CI): −0.39 to −0.21, P<0.00001) and 0.27 mmol/l (95% CI: −0.34 to −0.20, P<0.00001), respectively, compared with control. The pattern of cholesterol-lowering action of barley in this analysis could not be viewed as a dose-dependent response. There were no significant subgroup differences by type of intervention and food matrix.
Increased consumption of barely products should be considered as a dietary approach to reduce LDL cholesterol concentrations.
It is thought that barley cultivation began 8000–10 000 years ago in the ‘Fertile Crescent’ of the Middle East and thus barley has been considered to be one of the founding crops of Old World agriculture (Badr et al., 2000). Barley consumption as food is still an important component for many regions, including several areas of North Africa and the Near East, the highlands of Central Asia, the Horn of Africa, the Andean countries and the Baltic States (2005). On the other hand, many Western countries now only use barley predominantly for animal feed and malting. However, there has been renewed interest worldwide in barley as food due to its many health benefits, including its potential in reducing the risk of cardiovascular disease via cholesterol lowering (Behall et al., 2004a, 2004b; Keenan et al., 2007; Shimizu et al., 2008), and improvement of glucose tolerance (Pick et al., 1998; Behall et al., 2006; Hinata et al., 2007). The active ingredient thought to provide barley its health benefits is β-glucan. β-Glucan is a type of soluble fiber, which is made up of unbranched polysaccharides with (1 → 4) and (1 → 3) linked β-D-glucopyranosyl units in varying proportions (Charalampopoulos et al., 2002). Of cereal grains, oats and barley contain the highest level of β-glucan at 3–7 and 3–11% (dry weight basis), respectively (Charalampopoulos et al., 2002). A meta-analysis by Ripsin et al. (1992) showed that the consumption of about 3 g/day of soluble fibers from oat products lowered serum total cholesterol concentrations by 0.13–0.16 mmol/l; likewise, barley is also now emerging to have similar health benefits. To date, a number of studies have been carried out to test the efficacy of barley or β-glucan derived from barley as cholesterol-lowering agents. However, some studies have demonstrated no benefits (Keogh et al., 2003; Biorklund et al., 2005), whereas others have shown reductions in plasma lipids (McIntosh et al., 1991; Li et al., 2003; Behall et al., 2004a, 2004b; Keenan et al., 2007; Shimizu et al., 2008). The varying results in human studies may be due to factors such as dose size, food matrix, type of intervention, background diet and subjects’ characteristics.
Meta-analysis is a statistical tool that generates pooled estimates of effects from the results of randomized, controlled trials (Pai et al., 2004). Therefore, a meta-analysis could be used to more precisely quantify the efficacy of barley and its products as lipid-lowering agents. Thus, the primary objective of this meta-analysis was to quantify the effect of β-glucan from barley on total and low-density lipoprotein (LDL) cholesterol concentrations, as well as on high-density lipoprotein (HDL) cholesterol and triacylglycerol concentrations. The secondary objectives of this meta-analysis were to test for the presence of a dose–response effect, and to identify and quantify the effects of food matrix, intervention type (barley vs β-glucan from barley) and background diet on the efficacy of barley as a cholesterol-lowering agent.
Nine electronic databases, including AGRICOLA, Agris, BIOSIS, CAB Abstracts, Foodline Science, Food Science and Technology Abstracts, PubMed, Scopus and Google Scholar, were searched using barley, cholesterol and heart/coronary disease terms to July 2008. Non-English-language articles were translated when possible.
Studies conducted to examine barley and cardiovascular disease risk factors were first identified for this meta-analysis. Subsequently, all articles were reviewed to select studies with the following criteria: (1) randomized, controlled clinical trials with either a crossover or parallel design; (2) subjects from a healthy population, that is, not after myocardial infarction; (3) measured total and LDL cholesterol as outcomes; (4) subjects ingested β-glucan from barley; and (5) the intervention lasted for at least 3 weeks.
Studies were then evaluated for study quality using a custom-built tool in collaboration with Health Canada. The custom-built tool was used to evaluate studies found via a systematic literature search that was conducted in order to review scientific evidence for a potential Health Canada health claim submission for barley β-glucan as a cholesterol-lowering food. The custom-built tool allocated a certain number of points for factors including: randomization (two points), subject-inclusion criteria (one point), measure of food exposure (one point), measure of health outcome (one point), justification of subject number/power to detect (one point), description of background diet (two points), statistical analysis (two points), and accounted for confounding factors, including weight change, washout period and age (subtract up to three points). Studies were required to attain at least seven points to be deemed a pass to be included in the meta-analysis. These studies were then sent to expert reviewers for further re-evaluation. Expert reviewers were used to assess the data as per a step suggested in the Health Canada, ‘Interim Guidance Document, Preparing a Submission for Foods with Health Claims.’ (Health Canada, Bureau of Nutritional Sciences 2002). The reviewers chosen are all well-respected scientists in the field of nutrition and/or food science with emphasis on those who had an expertise in barley and soluble fiber. The reviewers were asked to comment on the following aspects: consistency of the observations and their effect, whether there is significant statistical evidence of lowering LDL cholesterol, dose–response relationship (for example—x causes y effect), feasibility of consuming the effective dose, target population that the evidence is aimed at (for example—general/hypercholesterolemic population), overall number or percentage of studies that show LDL cholesterol reduction and significance.
For studies that met the inclusion criteria and that possessed a quality score of seven or more, data were extracted for parameters related to (i) trial design, (ii) type of intervention, that is, barley vs β-glucan from barley, (iii) dose (g/day), (iv) duration of treatment, (v) food matrix, that is, food carrier, (vi) characteristics of the study population, (vii) the mean values and the s.d. of lipid levels and (viii) sample size. Two reviewers independently extracted the data.
Quantitative data synthesis
For studies that reported multiple time points for the same subjects, only end points for the longest duration of the intervention were used (Li et al., 2003; Shimizu et al., 2008). In one instance, mean values and s.e. of the serum lipids were estimated from the figures because they were not reported in the text (Keogh et al., 2003).
The primary outcomes for this meta-analysis were the differences in total and LDL cholesterol levels due to barley treatment. For parallel arm designed trials, end points of lipid concentrations among the subjects ingesting barley were subtracted from those among the subjects consuming the control group (Deeks et al., 2005). For crossover trials, the lipid concentrations at the end of the treatment period were subtracted from those at the end of the control period (Deeks et al., 2005). Within-individual changes were used when presented; otherwise, group means were used. The s.d.'s were extracted from the studies or, if not reported, derived from s.e. of the mean, paired t- or P-value as provided (Deeks et al., 2005). For a number of studies (Keogh et al., 2003; Behall et al., 2004a, 2004b) approximate paired analysis was performed using imputed correlation coefficients describing the similarity of outcomes within each subject as was previously described (Deeks et al., 2005). Within-individual correlations between the treatment and control periods were calculated from two studies (Clinical Study Report, 2005; Sundberg, 2008).
If different treatments were tested within the same trial, they were evaluated as separate strata, as is described by ‘i, ii, iii and iv’ suffixes in Tables and Figures. To obtain the pooled treatment effect size, effect size estimates and s.e. were entered into RevMan 4.2 under the ‘generic inverse variance’ outcome. Heterogeneity between trial results was tested for by using a standard χ2-test. A P-value <0.1 was used to indicate that significant heterogeneity was present (Deeks et al., 2005). Calculations used in this meta-analysis were previously presented in more detail (AbuMweis et al., 2008). Estimates of the pooled treatment effect sizes and 95% confidence intervals (CIs) were calculated by using both fixed effect and random effect models. If the test for heterogeneity was significant, we presented the results of the random effect models. Otherwise, estimated results based on a fixed effect model are presented. The presence of publication bias was examined using a funnel plot in which the s.e.'s of the studies were plotted against their corresponding effect.
Characteristics of the studies
A total of 266 articles were identified from the first search strategy. Of these, only 11 studies with 17 strata met the a priori inclusion criteria and passed the quality assessment test (Figure 1). Table 1 describes the studies included in the calculation of the effect size. Seven studies utilized a crossover design, and four used a parallel design. The duration of the intervention lasted from 4 to 12 weeks and 10–62 subjects were enrolled in the studies. Mean age ranged from 20 to 63 years and mean baseline body mass index ranged from 19 to 35 kg/m2. Five studies recruited only males and one study recruited only women, whereas the remaining studies comprised 28–52% male subjects. In five studies (McIntosh et al., 1991; Keogh et al., 2003; Clinical Study Report, 2005; Biorklund et al., 2005; Sundberg, 2008), the subjects had high blood cholesterol concentrations at baseline according to the definitions of the ATPIII (Cleeman et al., 2001). All the studies except one (Shimizu et al., 2008) reported no significant weight changes.
The median intervention dose of β-glucan given to the subjects was 5 g/day. The interventions were incorporated into different foods including liquid and solid food. In five studies, the subjects consumed their habitual diet (Newman et al., 1989; Clinical Study Report, 2005; Biorklund et al., 2005; Shimizu et al., 2008; Sundberg, 2008), and in two studies, the subjects were provided with an American Heart Association Step I diet (Behall et al., 2004a, 2004b). Two other studies were controlled feeding trials of typical Japanese (Li et al., 2003) and Western (Keogh et al., 2003) diets. In one study (Keenan et al., 2007) subjects were given instructions to follow a low saturated fat and low trans fat diet. Most of the studies did not report all data on the actual nutrient intakes by the subjects. Most control groups received wheat or rice products and therefore barley products were not isocalorically substituted for fat.
Changes in serum lipid concentrations
Overall, β-glucan from barley lowered total and LDL cholesterol concentrations by 0.30 mmol/l (95% CI: −0.39 to −0.21, P<0.00001) and 0.27 mmol/l (95% CI: −0.34 to −0.20, P<0.00001), respectively, compared with the control (Figure 2). The reductions in total and LDL cholesterol concentrations were significant in 10 and 11 strata, respectively. The ingestion of β-glucan from barley did not affect HDL cholesterol and triacylglycerol concentrations (Figure 2). The test for heterogeneity was not significant (P>0.1) except for the subgroup analysis of the effect of β-glucan from barley and use of beverage as food carrier on total cholesterol concentrations (Table 2).
Subgroup analysis based on dose size (Tables 2 and 3) showed that the reductions in total cholesterol were 0.25 mmol/l (95% CI: −0.36 to −0.14, P<0.00001), 0.45 mmol/l (95% CI: −0.69 to −0.20, P=0.0003) and 0.28 mmol/l (95% CI: −0.50 to −0.07, P=0.01) for intakes of 3–5, 5.1–7 and >7.1 g/day, respectively. The reductions in LDL were 0.22 mmol/l (95% CI: −0.31 to −0.12, P<0.00001), 0.33 mmol/l (95% CI: −0.47 to −0.19, P<0.00001) and 0.24 mmol/l (95% CI: −0.37 to −0.10, P=0.0005) for intakes of 3–5, 5.1–7 and >7.1 g/day, respectively. However, there were no significant subgroup differences, which could be due to the limited number of studies used in the analysis. When studies were analyzed according to the type of the intervention, that is, barley vs β-glucan from barley, the reduction of total cholesterol was only significant in barley, but both barley and β-glucan from barley reduced LDL concentrations. Subgroup analysis by food matrix revealed that β-glucan from barley reduced total and LDL cholesterol whether they were incorporated into a beverage or solid foods. However, results of the subgroup analyses should be interpreted with caution.
The effect of the background diet on efficacy of β-glucan from barley as a cholesterol-lowering agent was not tested due to the limited information provided by the studies. For example, studies reported that subjects were given controlled feeding diets but did not report the actual nutrient intake of the subjects. Therefore, it is not known whether subjects adhered to the dietary instructions.
On the basis of subgroup analysis, baseline cholesterol levels did not affect the cholesterol-lowering action of β-glucan from barley (P>0.05). The decreases in LDL were 0.31 mmol/l (95% CI: −0.55 to −0.07, P=0.01), 0.33 mmol/l (95% CI: −0.46 to −0.20, P<0.00001) and 0.22 mmol/l (95% CI: −0.30 to −0.13, P=0.01) for studies in which subjects had optimal, borderline high and high baseline values of serum cholesterol, respectively.
Funnel plots for total and LDL cholesterol concentrations are shown in Figure 3. Visual examination of the funnel plots shows its asymmetrical appearance with a gap in a bottom corner of the graph. Thus, it is clear that small studies remain unpublished.
This is the first meta-analysis of randomized clinical trials yielding information on the effect of β-glucan from barley on lipid profile. The meta-analysis showed that the consumption of β-glucan from barley decreased blood cholesterol concentrations in subjects with different dietary backgrounds. Owing to limited information available in the studies, we were not able to study the cholesterol-lowering effect of other dietary components present in barley, such as arabinoxylans, plant sterols and tocols.
Barley intake did not affect HDL cholesterol or triacylglycerol concentrations. Similarly, the consumption of other soluble fibers, including those from oats, psyllium and pectin, have been shown not to have an effect on HDL cholesterol and triacylglycerol concentrations (Brown et al., 1999).
The pattern of cholesterol-lowering action of β-glucan from barley in this analysis cannot be viewed as a dose-dependent response. There are various reasons that may help to explain the lack of dose-dependent response. First, the dose-dependent response may be observable with a wider range of doses, other than the one used in our analysis. In other words, the range of doses used here may have showed only a plateau effect. Second, differences in the molecular weight of β-glucan may influence the dose–response effect. It has been suggested that β-glucan characteristics including its solubility and molecular weight are important determinants of its cholesterol-lowering action (Theuwissen and Mensink, 2008). Highly water-soluble β-glucan, with moderate to high molecular weight, may reduce serum LDL cholesterol levels better than β-glucan with a low water-solubility and low molecular weight (Theuwissen and Mensink, 2008). On the other hand, sensory properties of foods enriched with β-glucan are more positively received at a lower β-glucan molecular weight (Keenan et al., 2007). In one study, the consumption of either high molecular weight or low molecular weight concentrated barley β-glucan at both 3 and 5 g doses for 6 weeks decreased LDL cholesterol compared with control (Keenan et al., 2007). The exact molecular weight of β-glucan was not reported in the majority of studies and therefore we could not assess its impact on the cholesterol-lowering action of barley products. Third, the power might have been insufficient to prove dose–response effect at the included dose range. High consumption of β-glucan (>7 g/day) did not appear to have substantially greater effects than modest consumption (3–5 g/day). Therefore, consumption of at least 3 g/day of barley β-glucan will reduce blood cholesterol concentrations.
Similarly, other studies have shown that food processing of β-glucan affect its cholesterol-lowering effect as a consequence of changes in β-glucan structure or solubility. A daily intake of a beverage providing 5 g of β-glucan from barley for 5 weeks did not reduce LDL concentrations (Biorklund et al., 2005). In another study, intake of 3 g/day of β-glucan supplied as barley powder and mixed in a beverage reduced LDL concentrations compared with control (Clinical Study Report, 2005). The consumption of about 8–12 g/day of barley β-glucan extract incorporated into baked products for 4 weeks did not improve the lipid profiles of hypercholesterolemic men (Keogh et al., 2003). While, consumption of barley based baked products reduced LDL cholesterol in other studies (Newman et al., 1989; McIntosh et al., 1991). In spite of some conflicting results from individual clinical trials, pooling data from 11 studies in this analysis showed that incorporating barley into different food products could be used as an efficacious way to increase the consumption of viscous soluble fibers in order to achieve the desired reduction in LDL cholesterol concentration.
The cholesterol-lowering effect of barley reported in this meta-analysis is greater than that reported previously for oats (Ripsin et al., 1992), 0.25 mmol/l (95% CI: −0.36 to −0.14) and 0.13 mmol/l (95% CI: −0.19 to −0.017), respectively, for barley and oats. One difference between the two meta-analyses is that Ripsin et al. included many studies that used lower doses (<3 g/day) but in this study all studies used doses 3 g. Moreover, there are differences with respect to other aspects such as the inclusion criteria and quality assessment of the trials. For example, this meta-analysis assessed the quality of the trials, whereas that by Ripsin et al. did not. Upon quality assessment of trials in this current analysis, seven trials were excluded because they were not of sufficient quality.
Another meta-analysis by Brown et al. (1999) showed that various soluble fibers from oats, psyllium or pectin reduce total and LDL cholesterol by similar amounts. We cannot directly compare our results to the result from Brown et al. analysis as the latter analysis was expressed as reduction in LDL per gram of fiber consumed. In addition, heterogeneity was evident in the Brown et al. (1999) analysis, whereas heterogeneity was not observed in this study.
Both barley and oats contain β-glucan, which is thought to be the active ingredient responsible for their cholesterol-lowering effect (Truswell, 2002). One human study has concurrently examined the effect of β-glucan from oats or barley on total cholesterol concentration (Biorklund et al., 2005). The Biorklund et al. (2005) study showed that the consumption of a beverage containing 5 g of β-glucan from oats lowered total cholesterol concentrations by 7.4% compared with a control beverage. No cholesterol-lowering effect of a beverage with 5 g of β-glucan from barley was found. The lack of cholesterol-lowering effect of β-glucan from barley in the study by Biorklund et al. (2005) was attributed to its lower molecular weight compared with that of β-glucan from oats.
In conclusion, this meta-analysis of 11 studies indicates that the consumption of barley or β-glucan from barley incorporated into different food products is associated with a significant reduction in total and LDL cholesterol concentrations. Increased consumption of barely products should be considered as a dietary approach to reduce LDL cholesterol concentrations.
We thank Tracy Lee Exley, Stephen Capelle and Camille Rhymer for their assistance in literature search and systematic review. This study was supported by the Agriculture and Agri-Food Canada and the Alberta Barley Commission. The funders had no role in the design, analysis or interpretation of the study.