To investigate side by side the effects on serum lipoproteins and postprandial glucose and insulin concentrations of beverages enriched with 5 or 10 g of β-glucans from oats or barley.
Design and setting:
An 8-week single blind, controlled study with five parallel groups carried out at two centres under identical conditions.
A total of 100 free-living hypercholesterolaemic subjects were recruited locally and 89 completed the study.
During a 3-week run-in period all subjects consumed a control beverage. For the following 5-week period four groups received a beverage with 5 or 10 g β-glucans from oats or barley and one group continued with the control beverage. Blood samples in weeks 0, 2, 3, 7 and 8 were analysed for serum lipids, lipoproteins, glucose and insulin. Postprandial concentrations of glucose and insulin were compared between control and the beverage with 5 g of β-glucans from oats or barley.
Compared to control, 5 g of β-glucans from oats significantly lowered total-cholesterol by 7.4% (P<0.01), and postprandial concentrations of glucose (30 min, P=0.005) and insulin (30 min, P=0.025). The beverage with 10 g of β-glucans from oats did not affect serum lipids significantly in comparison with control. No statistically significant effects compared to control of the beverages with barley β-glucans were found.
A daily consumption of 5 g of oat β-glucans in a beverage improved the lipid and glucose metabolism, while barley β-glucans did not.
Founded by the European Commission (QLK1-CT-2000-00535).
As early as 1963 de Groot and co-workers observed that oats lowered serum total-cholesterol concentrations in humans. Since then, many human studies have confirmed this effect and suggested that it is related to the soluble fibres in oats, the β-glucans. From a meta-analysis by Ripsin et al (1992) it was concluded that about 3 g per day of soluble fibres from oat products could lower serum total-cholesterol concentrations by 0.13–0.16 mmol/l in hypercholesterolaemic subjects. Based on this meta-analysis and other convincing studies (Braaten et al, 1994), the US Food and Drug Administration (FDA) approved a health claim, ‘that soluble fibres from whole oats, as a part of a diet low in saturated fat, cholesterol and total fat, may reduce the risk of heart disease’ (FDA, 1997). FDA implied that the amount of soluble fibre needed for an effect on cholesterol levels is about 3 g per day and, to qualify for the health claim, the whole oat-containing food must provide at least 0.75 g of soluble fibres per serving.
Even though Ripsin et al (1992) observed a dose–response relationship between the amount of soluble fibre in the diet and the degree of cholesterol lowering, it is not clear if the effect of β-glucans levels off at higher intakes. Davidson et al (1991) found that there was a significantly greater serum cholesterol reduction after an intake of 4 g of β-glucans compared with 2 g β-glucans from oat bran or oat meal, incorporated into foods such as muffins, cereals and shakes. However, a dose of 6 g of β-glucans did not result in a further reduction in serum cholesterol concentrations. In addition, whole oats (oatmeal) or oat bran has been used as sources of β-glucans in most studies. The concentration of β-glucans in oat kernels varies between 3.5 and 5.7% (Asp et al, 1991) and in oat bran the content of dietary fibre should be at least 16% (dry matter basis; AACC, 1989). Thus, even though some foods are naturally rich in soluble fibres, larger quantities need to be consumed to achieve an intake according to the FDA health claim. Therefore, it would be of benefit for a large segment of the population if a variety of food products with a concentrated content of soluble β-glucans was available. In the present study, we therefore investigated if the cholesterol-lowering effect of a beverage enriched with a high dose of β-glucans (10 g) is more pronounced compared to a beverage providing half the amount.
Barley has a similar concentration of β-glucans as oats (3.5–5.9% of the dry matter; Oscarsson et al, 1998), yet there are only few published trials on the cholesterol-lowering effect of barley in humans. McIntosh et al (1991) observed a significant fall in total- and LDL-cholesterol in hypercholesterolaemic men by incorporating 170 g of barley bran and flakes into a normal diet.
In two recent studies by Behall et al (2004a, 2004b) hypercholesterolaemic subjects were given low-fat whole grain diets daily that contained either low (0–0.4 g), medium (3 g) or high (6 g) amounts of barley β-glucans. The total- and LDL-cholesterol concentrations were significantly reduced with the high barley β-glucan diet in comparison with the diet low in barley β-glucans in both studies. However, the results were not compatible for the diets containing medium amounts of barley β-glucans. In the first study by Behall et al (2004a), the medium dose (3 g) of barley β-glucans significantly lowered the total and LDL concentrations compared to the low dose. This did not occur in the second study (Behall et al, 2004b). Keogh et al (2003) also recently investigated the serum lipoprotein profile in hypercholesterolaemic men taking barley β-glucans, but did not see any significant effect compared to control of a daily 10 g dose of barley β-glucans incorporated into foods as bread, cakes, muffins or savoury dishes. Contrary to the healthy back ground diet of Behall et al (2004a, 2004b), the barley and control diet in the study of Keogh et al (2003) had an energy- and macronutrient intake more typical of a Westernised diet. So far, no human studies have evaluated the cholesterol-lowering effects of oats and barley side by side. We therefore decided to compare the effect of products enriched with β-glucan from oats and barley on the serum lipoprotein profile. Other advantageous effects of β-glucans are a reduced postprandial glucose and insulin response (Braaten et al, 1994; Dubois et al, 1995; Liljeberg et al, 1996; Bourdon et al, 1999), and for that reason we also evaluated the effect of the oats and barley β-glucan beverages on postprandial concentrations of glucose and insulin.
Subjects and methods
The study was designed as a single blind, randomised, dose-controlled trial. To increase its external validity the trial was carried out simultaneously under identical conditions in Maastricht (the Netherlands) and Lund (Sweden). During a 3-week run-in period all subjects (n=100) consumed a control beverage enriched with rice starch daily. At week 3, they were randomly divided over five groups, stratified for gender and centre. For the following 5-week intervention period, four groups received a beverage enriched with either 5.0 or 10.0 g of β-glucans from oat or barley, while one group still received a control beverage. During the study, subjects consumed their habitual diets and were not allowed to change their physical activity pattern. The primary end point was changes in LDL-cholesterol and secondary end points were changes in serum total-cholesterol, HDL-cholesterol, plasma glucose, insulin concentrations, triacylglycerols, apolipoprotein A1 and B. Blood samples and body weight were taken at the start of the study (week 0) and twice at the end of the run-in (weeks 2 and 3) and intervention periods (weeks 7 and 8). In Lund the 10 subjects from the two groups that received the beverage with 5 g of β-glucans from oats or barley also participated in postprandial tests in weeks 3 and 8.
By advertisements in local newspapers the two centres recruited 100 healthy men and women with mildly elevated serum cholesterol concentrations. Before entering the study the subjects were screened by two blood tests and a questionnaire about their health status. At the second screening visit weight, height, glucosuria and blood pressure were measured. The inclusion criteria were: age 18–70 y, body mass index (BMI) 20–30 kg/m2, serum total-cholesterol 5.5–8.0 mmol/l and LDL-cholesterol 4.1–5.7 mmol/l. Exclusion criteria were serum triacylglycerols above 4 mmol/l, a systolic blood pressure above 160 mmHg, a diastolic blood pressure above 95 mmHg and presence of glucosuria. Individuals were also excluded if they reported to have an unstable body weight (weight gain or loss ≥3 kg in the previous 3 months), diabetes mellitus, history of coronary heart disease and/or heart failure, cardiomyopathy, history or presence of kidney-, liver- and/or pancreatic disease, serious malignancy less than 5 y ago or drug treatment for hyperlipidaemia. Further, subjects should not be pregnant or breastfeeding, abuse alcohol or drugs, use a medication or a diet known to affect lipid or glucose metabolism, use other research products up to 30 days prior the study, have a known gluten intolerance, or participate in another research study at the same time. The local ethical committees of Lund University and of Maastricht University had approved the study. All subjects gave written informed consent before participation.
Study products and diet
Oat bran from Swedish oats and polished grains of Swedish barley were used to produce fibre preparations with a high content of β-glucans (Ceba Foods AB). Oats or barley was milled, treated with enzymes and the insoluble fibres were removed. The remaining β-glucan fractions were freeze-dried and the powder that was received had a β-glucan content of 18% for the oat preparation and of 36% for the barley preparation. The powder products were mixed with water, and to get a 1% concentration of β-glucans in the beverages 5.3 g of oat fibre preparation and 2.8 g of barley fibre preparation were added. Consequently, the double amount of each fibre preparation was added for the beverages with 2% concentration of β-glucans. The mean molecular weight (MW) of the β-glucans in the fibre preparations was 200 000 for oats and 40 000 for barley. In the beverages the MW of the β-glucans was 70 000 for oats and 40 000 for barley. In the control beverage, rice starch (4.5 g/100 ml) was used instead of the fibre preparations. All beverages were flavoured with a blackcurrant fruit juice concentrate (5 g/100 ml). Rapeseed oil was used to balance the total amount of lipids, sucrose and glucose syrup powder to balance the energy content, and citric acid was used for the adjustment of pH (Table 1). The serving size of the beverages was 250 ml, and to obtain a daily intake of 5 or 10 g of β-glucans the subjects were instructed to consume two beverages (2 × 250 ml) with two main meals (breakfast, lunch or dinner). The nutrient composition and the content of β-glucans per 100 ml are described in Table 2. The beverages are hereafter referred to as oat-5, oat-10, barley-5 and barley-10 depending on their total content of β-glucans per 500 ml.
Compliance with the protocol was assessed on the basis of the subject's self reported daily intake in a diary and by returned beverages. Three times during the study the subjects filled in a questionnaire for sensory evaluation concerning taste, smell, consistency, appearance and total impression of the beverages, using a nine-graded scale that ranged from ‘dislike very much’ (grade 1) to ‘like very much’ (grade 9).
In the diary the subjects also recorded any signs of illness and medication used. At the end of the run-in and intervention periods, the subjects were requested to report if they had experienced any side effects related to the study products, such as stomach problems, flatulence, bloating or abdominal pain. Additional side effects and participitans' health status were checked three times during the study by measuring standard haematological parameters (number of white and red blood cells, number of platelets) and clinical chemistry for liver function (ALAT, ASAT, γGT and total bilirubin), kidney function (creatinin) and inflammation (C-reactive protein).
A dietary registration was made in the last week of both the run-in and the intervention period to assess the subjects' intakes of energy, protein, fat, carbohydrates, alcohol and fibre during the two study periods. At the centre, in Lund a 3-day food record was used (Menu Book, Swedish National Food Administration), in which the subjects recorded their food intake during two weekdays and one weekend day. A dietitian checked the food records in the presence of the subjects. The food items were coded and the composition of the diets was calculated using the food database from the Swedish National Food Administration (PC-Kost 1_99, SLV, Uppsala, Sweden) and the software programme Stor MATs 4.05 SR-3 (Rudans Lättdata AB, Västerås, Sweden). At the centre in Maastricht the subjects recorded their food intake at the same intervals by filling in food frequency lists, consisting of 100 items. A dietitian checked the food frequency lists immediately in the presence of the subjects. Items were coded and the composition of the diets was calculated according to the Dutch Food Composition Table (NEVO tabel, Nederlands voedingstoffenbestand. Den Haag: Voorlichtingsbureau voor de voeding, 1998). The study products were not included in these calculations.
Blood sampling and analyses
Blood samples were taken after an overnight fast at the start of the study (week 0) and at the end of the run-in (weeks 2 and 3) and intervention periods (weeks 7 and 8). The blood was sampled with the subject in recumbent position with a minimum of stasis using the vacuum system, in a 10 ml clotting tube, a 10 ml EDTA tube and a 5 ml fluoride tube. Serum and plasma were obtained by low-speed centrifugation within 1 h after venipuncture and stored at −20°C until analysis. Blood collection, subsequent handling and plasma storage was performed according to standardised procedures.
The blood samples were transported on dry ice at the end of the study to the department of Clinical Chemistry at the University Hospital in Lund, which was responsible for the analysis. Plasma glucose, serum total-cholesterol, HDL-cholesterol, triacylglycerol and apolipoproteins A1 and B were measured on Hitachi Modular-P4 and P2 equipment, according to standardised procedures (Roche, Hitachi Modular). LDL-cholesterol concentrations were calculated using the Friedewald equation (Friedewald et al, 1972). Serum insulin was measured by using the Beckman Access Immunoassay system. The postprandial serum samples were analysed for triacylglycerols, glucose and insulin using the same methods as described above. The haematological parameters and clinical chemistry were analysed at each centre immediately after blood sampling. The measurements were carried out on a Beckman Synchron CX System (Beckman Instruments Inc., Palo Alto, CA, USA) according to standardised procedures (Synchron CX Clinical Systems Operating Instructions Manual, Beckman Instruments Inc., 1995).
In Lund 10 subjects from the oat-5 group and 10 subjects from the barley-5 group volunteered to participate in two postprandial tests. All 20 subjects performed the first test in week 3 with the control beverage, and the second test in week 8 with either the oat-5 or barley-5 beverage. The test meals consisted of the total daily amount of the beverage (500 ml) accompanied with bread and butter. Owing to the lower protein content of the control and barley-5 beverages ham was also added to these meals. The nutrient content of the meals was calculated using the database from Swedish National Food Administration and a computerised calculation programme (Dietist, Kost och näringsdata AB, Bromma, Sweden) and are presented in Table 3.
The two tests took place in the morning with the subjects in a fasting state. At the first test the subjects were interviewed about their food intake and physical activity pattern the day and morning before, and they were instructed to repeat this at the second test. Before the test meal a catheter was placed in an antecubital vein and fasting blood samples were collected. Satiety was assessed before the test meal using an analogue scale with grades from −8 (extreme hunger) to 8 (extreme satiety). Blood sampling was repeated at every 30-min interval until 180 min after the test meal parallel with continuous assessment of satiety. Serum and plasma were separated and stored at −20°C until analysis.
Responses to the study products were calculated as the difference between the two mean values obtained at the end of the run-in and intervention period. Differences in responses between groups were evaluated using analysis of variance (ANOVA). Tukey's post hoc test was used when required. To compare changes within a group a paired samples t-test was used. Postprandial tests, the satiety VAS analogue scale and the sensory questionnaires were evaluated using Wilcoxon's nonparametric tests for pairwise comparisons. All tests were two-tailed and P<0.05 was considered to be statistically significant. This study was calculated to have the power to detect a 6% decrease in LDL-cholesterol with an α value of 5% and a β risk lower than 20%. Results are presented as mean±s.d., unless otherwise indicated. All statistical analysis was performed using SPSS 11.5 for Windows.
In total, 100 subjects were included in the study, of which 89 (45 females and 44 males) completed the 8-week protocol. The baseline characteristics of the subjects (mean±s.d., n=43) from Lund were: age 59±7 y, BMI 25.3±3.2 kg/m2, total-cholesterol 6.26±0.7 mmol/l, LDL-cholesterol 4.20±0.6 mmol/l, HDL-cholesterol 1.44±0.4 mmol/l and triacylglycerols 1.29±0.7 mmol/l. In Lund five subjects dropped out because they disliked the beverages (oat-10 and barley-10), one due to reasons not related to the study protocol, while one participant was excluded from the statistical analysis because of elevated blood glucose concentrations. In Maastricht, the baseline characteristics of the subjects (n=46) were: age 53±12 y, BMI 25.3±3.3 kg/m2, total-cholesterol 6.70±1.12 mmol/l, LDL-cholesterol 4.46±1.0 mmol/l, HDL-cholesterol 1.46±0.4 mmol/l and triacylglycerols 1.69±0.9 mmol/l. Four subjects in Maastricht dropped out during the run-in period because of illness not related to the study products. In the final per-protocol analysis the mean values (±s.d.) at baseline for the 89 subjects (44 men and 45 women) who completed the study were: age 56±10 y, BMI 25.2±3.3 kg/m2, total-cholesterol 6.49±1.0 mmol/l, LDL-cholesterol 4.35±0.8 mmol/l, HDL-cholesterol 1.46±0.4 mmol/l and triacylglycerols 1.50±0.8 mmol/l. No significant differences for the baseline parameters were observed among the groups, that is, oat-5, oat-10, barley-5, barley-10 and control (Table 4).
Study products and diet
The daily average intake of the beverages for all groups was 496 (range: 417–500) ml during the run-in and 494 (range: 402–500) ml during the intervention period, which corresponds to a compliance of 99% during both periods. There was no difference in the average intake between the groups during the intervention period. Some subjects reported gastrointestinal discomfort during the study. The major complaints, including bloating, flatulence and diarrhoea, were reported both for the control beverages and the beverages enriched with β-glucans. The gastrointestinal problems were somewhat more frequent in the oat-10 group (11 complaints) compared to the other groups (7–8 complaints), but the problems decreased gradually for all subjects after 1–2 weeks of consumption.
In week 4, the mean (±s.d.) total impression of the beverages on a nine-graded scale was 5.4±1.8 for the oat-5 beverage and 5.5±2.1 for the barley-5 beverage. This was significantly higher than that for the oat-10 (3.3±1.9) and barley-10 beverage (3.6±2.4). In week 8, the total impression of the beverages was slightly higher in comparison with week 4. The largest differences for the sensory parameters were seen for the appreciation of the consistency of the beverages. In week 4, this was significantly higher for the oat-5 (5.4±1.9) and the barley-5 beverage (5.5±2.5) than for the oat-10 (2.3±1.9) and the barley-10 beverage (2.6±1.8). In weeks 4 and 8, evaluation scores for the control group and the oat-5 and barley-5 groups were comparable.
In Lund the registered mean daily energy intake during the run-in (range: 6.7–8.1 MJ) and intervention period (range: 6.8–8.4 MJ) remained similar for all groups throughout the study. The intakes of protein, fat, carbohydrates, alcohol and fibre were also similar, except in the control group. In this group the mean daily fat intake was significantly lower (60.1±21.5 g) during the run-in period compared to the intervention period (68.9±21.8 g; P=0.04). Consequently, the carbohydrate intake in this group was significantly different between the periods and decreased from 51.5±5.2 percent of energy (En %) during the run-in to 49±6.5 En % (P=0.02) during the intervention period. In Maastricht the registered mean daily energy intake also remained similar with a range from 8.9 to 10.5 MJ during the run-in and from 8.8 to 10.1 MJ during the intervention period. In the barley-5 group, the mean daily protein intake was significantly higher during the run-in (107.9±23.4 g) compared to the intervention period (98.5±22.1 g; P=0.01). In the barley-10 group, the energy (10.3±2.2 MJ) and fat intakes (97.8±33.3 g) during the run-in period were significantly higher compared to intake in the intervention period (9.2±2.5 MJ, P=0.03; 81.6±25.4 g fat, P=0.05).
BMI, lipids, glucose and insulin
In neither of the groups, the BMI changed between the run-in and intervention periods and there were no significant differences between the groups. Compared to the control group, the oat-5 group lowered the LDL-cholesterol concentrations by 0.29 mmol/l or 6.7% and the oat-10 group beverage by 0.16 mmol/l or 3.7% (Table 4). Corresponding reductions for LDL-cholesterol in the barley-5 and the barley-10 groups were 0.08 mmol/l (1.8%) and 0.17 mmol/l (4.0%). The changes in LDL-cholesterol during the intervention period did not differ significantly between the groups. For the oat-5 group, the LDL-cholesterol in the intervention period was significantly lower compared to the mean value after the run-in period (P=0.01). For total-cholesterol, the reductions compared to the control group were 0.49 mmol/l (7.4%) for the oat-5 group, 0.29 mmol/l (4.5%) for the oat-10 group, 0.20 mmol/l (3.1%) for the barley-5 group and 0.27 mmol/l (4.2%) for the barley-10 group. Changes in total-cholesterol were significantly different between groups and post hoc test (Tukey's) showed that the reduction for the oat-5 group was significantly different compared to the control group (P<0.01) and the barley-5 group (P=0.04). Triacylglycerols concentrations were slightly reduced in the oat-5, oat-10 and barley-10 groups. In the barley-10 group, the triacylglycerol concentrations did not change but in the control group they slightly increased. These changes were not significant between the groups and neither were the changes for apolipoproteins A1 and B, glucose and serum insulin.
In the group where the postprandial effects of oats was evaluated the postprandial glucose concentrations after the oat-5 meal were 19% lower at 30 min (P=0.005) and 16% lower at 60 min (P=0.066) in comparison with the control meal (Figure 1). The insulin response after the oat-5 meal was also significantly lower at 30 min (33%, P=0.025) compared to control (Figure 2). In the other group that compared the postprandial effect of barley and control, the mean postprandial glucose and insulin concentrations were almost identical after both the barley-5 and the control meal (Figures 3 and 4). There was a significant difference in glucose concentrations between the oat-5 and barley-5 meal at 30 min (P=0.003). The satiety score before and after the test meals did not differ for the oat-5 or the barley-5 meal compared to the control meal.
The present study showed that a daily intake of a beverage providing 5 g of β-glucans from oats significantly lowered serum total-cholesterol concentrations by 7.4% compared with a control beverage containing rice starch. Effects of the barley beverages were less pronounced. Keogh et al (2003) also reported a poor cholesterol-lowering effect of a daily supplement of 10 g of barley β-glucans compared to a control diet and conclude that the lack of effect may be a consequence of structural changes in β-glucan that result from food processing or storage of the barley products. Furthermore, in this study there was a greater variability in lipid profile than predicted between and within subjects that could have masked small improvements in total-cholesterol. In contrast, Behall et al (2004a, 2004b) in two recent studies with similar design observed that total- and LDL-cholesterol was significantly reduced after a diet with 6 g of barley β-glucans per day in comparison with a diet containing 0–0.4 g barley β-glucans. All diets in the two studies by Behall et al (2004a, 2004b) reduced serum lipid compared to prestudy concentrations, which indicates that general healthy dietary changes (low fat content, high total dietary fibre content and whole grain foods) have a positive effect on blood lipids, but can be even more augmented when adding soluble fibre. Regarding β-glucan from oats, results are also conflicting, although the majority of studies have shown a cholesterol-lowering effect (Önning, 2004). However, in some studies the reductions were small, that is less than 5% for LDL-cholesterol, in comparison to control groups (Törronen et al, 1992; Whyte et al, 1992; Poulter et al, 1994; Lovegrove et al, 2000; Kerckhoffs et al, 2003). These varying results in human trials may be due to factors such as the dose of β-glucans, food processing, solid or liquid study products, the initial cholesterol concentrations of the subjects, or the study design. Although the cholesterol-lowering mechanisms of oats and barley β-glucans are still not fully understood, their effects seem to relate to an increased viscosity in the intestine, which leads to a decreased cholesterol absorption (Mälkki et al, 1992; Lia et al, 1995; Bourdon et al, 1999). The viscosity in the intestine not only depends on the amount of β-glucan ingested, but also on the solubility, structure and MW of the β-glucan (Åman et al, 2004). Decreasing the MW of the β-glucan will reduce the viscosity and consequently its cholesterol-lowering effect. The mean MW of the β-glucans in the barley beverages was almost half of the oat beverages, and this may explain the weaker effect of barley but not the lack of dose–response relationship. Food processing also has an important effect on the mean MW. Åman et al (2004) showed that the mean MW of β-glucan in oats is retained in rolled oats, oat bran and different types of oat bran concentrate (OBC), while bread making markedly degraded the β-glucan and the mean MW. Likewise, Kerckhoffs et al (2003) demonstrated that a daily intake of 5 g of oat β-glucans incorporated into bread did not decrease the cholesterol concentrations compared to when β-glucans are incorporated into beverages. Önning et al (1999) demonstrated significant cholesterol reductions of beverages with oat β-glucans (3.8 g per day) compared to rice starch as control. The processing of the oat and barley preparations in our study was comparable with the methods used in the study of Önning et al (1999).
In addition, there may be a difference in solubility of the β-glucans in oats and barley, and food processing can further influence this. Lambo et al (2005) analysed the solubility of barley and oat and found that only 15–20% of the barley β-glucans were soluble, while almost 70% of the oat β-glucans. Amundsen et al (2003) analysed the solubility of β-glucans in different test products that was used in a study, providing the subjects a daily intake of 5.1 g of β-glucans. The test products in both the oat and control (wheat or rye) diet were muesli, extruded breakfast flakes, bread, teacakes, muffins, fresh pasta, macaroni and apple drink. The mean solubility of the β-glucans in the test products was 50%, with a large variation from 22% in macaroni to 70% in the apple drink. Even if this meant that the intake of soluble β-glucan was only 2.7 g per day, it still decreased the total- and the LDL-cholesterol levels significantly compared to the control diet. We have not analysed the solubility of the β-glucan in the products used in our study, but probably this was high since the process to isolate the β-glucan-rich powder also filters away the insoluble dietary fibre components. However, the solubility of the β-glucans in the 10-g beverages may have been reduced during storage since this is more likely to happen in a concentrated solution, but whether this explains the lack of the dose–response relationship is not known.
In the postprandial part of our study the beverage enriched with 5 g of β-glucans from oats significantly lowered postprandial glucose and insulin concentrations compared to control, which indicates that β-glucans from oats can have an overall positive metabolic effect. Previous studies (Anderson et al, 1990; Hallfrisch & Behall, 2000; Keogh et al, 2003) have shown that the delayed glucose absorption probably is due to the high viscosity of β-glucans. Bourdon et al (1999) demonstrated that there was no significant difference in the glucose response between a meal with enriched barley pasta (5 g of β-glucans) and wheat pasta, but the insulin response differed. Though this indicates that carbohydrate from the meal with β-glucan enriched pasta was digested and absorbed more slowly, the effect was not enough to lower the glycaemic response. The authors hypothesised that pasta per se gives a low glycaemic response and that this was more important than the difference in amount of soluble fibre between these pasta meals. However, a sufficient amount of soluble fibre in a meal is needed to get an effect on the glucose and insulin response. Liljeberg et al (1996) showed that 2 g of soluble fibres in a meal with oat or barley porridges had no influence on the postprandial glucose tolerance, but 8 g of soluble fibre in a high-fibre barley porridge had.
Since we used different methods to assess dietary intakes in each centre, results were analysed separately. There were, however, only slight differences in dietary intake between the two study periods, which cannot explain the observed effects on the serum lipoprotein profile (Mensink et al, 2003). At both centres the subjects reported a low energy intake compared to their BMI and energy requirements. This indicates that subjects possibly underreported food intakes, which is a common problem of dietary assessment. However, the aim of this was not to evaluate the food intake per se, but to establish that the diet intake did not change significantly between the run-in and intervention periods.
In conclusion, this trial showed that a beverage containing 5 g of β-glucans from oats significantly lowered the total-cholesterol concentrations compared to a control beverage and compared to a beverage with 5 g of β-glucans from barley. The beverage enriched with 5 g of β-glucans from oats also gave a significantly lower postprandial glucose and insulin response compared to control, but the barley beverage did not. This trial was unable to show a dose–response effect of 5 g compared with 10 g of β-glucans from oats and barley. Thus, our study showed that the amount of β-glucan in a food product does not necessarily predict its effect on serum cholesterol concentrations. This suggests that physiological effects of β-glucans incorporated into new products and into different type of foods need to be carefully evaluated before entering the market.
We thank Tapani Suortti and Marjatta Salmenkallio-Marttila, VTT Biotechnology, Espoo, Finland for analysis of the molecular weight of the β-glucans in the study products, and Angeliki Öste-Triantafyllou, Ceba AB, Lund, Sweden for helping with the formulation of the study products. This work was carried out with financial support from the Commission of the European Communities, project QLKI-CY-2000-00535 ‘Design of foods with improved functionality and superior health effects using cereal β-glucans’. It does not necessarily reflect its views and in no way anticipates the Commission's future policy in this area.