Fibre supplementation for the prevention of type 2 diabetes and improvement of glucose metabolism: the randomised controlled Optimal Fibre Trial (OptiFiT)

The effectiveness of preventing type 2 diabetes mellitus by moderate alterations in lifestyle has been proven in two previous trials [1, 2]. Additionally, large epidemiological studies have identified dietary fibre as one of the most effective nutrient components for diabetes prevention [3]. Fibre intake is currently far below the recommended levels in most countries that consume Western-style foods [4]. According to epidemiological data, the preventive effect of fibres is most evident with cereal but not at all with fruit, legumes and vegetable fibres [5, 6]. Cereals contain predominantly insoluble fibres from the coat region and a smaller fraction of soluble fibres from the endosperm and aleurone layers of the grain. The insoluble fraction is composed of cellulose, hemicelluloses and lignin, and is only modestly fermented, such that the greater part is excreted with the faeces [7]. This modest fermentation leads to little gas production in the intestine, making insoluble fibres more acceptable to people who are sensitive to bowel symptoms.

Mechanisms linking insoluble fibres to glucose homeostasis in humans point to a rapid, overnight improvement in meal-related insulin sensitivity without significant changes in insulin secretion when using either insoluble fermentable or non-fermentable fibres or whole grain [8, 9]. Using cereal fibre extracts identical to those used in the current study, we have shown an improvement in euglycaemic–hyperinsulinaemic-clamp-measured whole-body insulin sensitivity after 3 days and 6 weeks in humans [10, 11], as well as beneficial metabolic effects in a long-term controlled study in C57BL/6J mice [7].

Long-term prospective randomised data on fibre supplementation in humans are sparse, and available data do not show striking changes in glycaemic variables. A 1 year study involving 14 participants failed to demonstrate beneficial effects, possibly due to relatively small differences in fibre intake [12]. So far, no evidence for an interventional effect of soluble fibre on glucose levels has been provided [13, 14, 15].

Recent diabetes prevention trials have shown that lifestyle intervention reduces the risk of incident diabetes by 40–60% compared with placebo [1, 2, 16, 17]. These studies targeted primarily weight loss and increased physical activity. Other studies have investigated the effect of a reduced intake of saturated and total fat and increased fibre intake as part of daily nutrition. However, relevantly increased fibre intake was usually not achieved [18]. Therefore, supplementation appears to be a more efficient way to increase fibre consumption. Here, we hypothesised that a continuous increase in fibre supplementation of everyday food intake would lead to a significant reduction in the incidence of type 2 diabetes in a high-risk cohort. In order to select and design foods containing effective fibre types, and to estimate the effect size of increased fibre intake, proof-of-principle studies are needed. We therefore decided to perform a long-term prospective randomised placebo-controlled trial investigating whether insoluble cereal fibres would improve glucose metabolism over 12 and 24 months in individuals with impaired glucose tolerance (IGT).

During the 12 months, mean body weight decreased by 2.0 ± 5.2 kg in the fibre group and by 2.6 ± 5.2 kg in the placebo group (total difference 0.6 ± 0.7 kg; not significant). Variables such as BMI and waist and hip circumference significantly decreased in both groups, fat mass (BIA) decreased in the placebo group only, whereas blood pressure and WHR did not change in either group (see electronic supplementary material [ESM] Table 1).

The change in HbA_1c levels was significantly different between the two groups (−0.2 ± 4.6 mmol/mol [−0.0 ± 0.0%; not significant] vs +1.2 ± 5.2 mmol/mol [+0.1 ± 0.0%; not significant]; total difference 1.4 ± 0.7 mmol/mol [0.1 + 0.0%]; p = 0.018, adjusted for weight change).

In the fibre and placebo group, respectively, we found significantly decreased 2 h plasma glucose levels (−0.78 ± 1.88 mmol/l [p ≤ 0.001] vs −0.46 ± 1.80 mmol/l [p = 0.020]; total difference 0.32 ± 0.29 mmol/l; not significant) and AUC for glucose (−71.4 ± 184.0 mmol/l × min [p ≤ 0.001] vs −59.4 ± 156.7 mmol/l × min [p ≤ 0.001], total difference 12.0 + 27.0 mmol/l × min; not significant) without a significant difference between the groups. Fasting blood glucose did not change (Fig. 1a–c).

Open image in new window

Fasting insulin levels significantly decreased only in the fibre group (−9.1 ± 22.2 pmol/l [p = 0.001] vs −6.3 ± 29.9 pmol/l [p = 0.064]; total difference 2.8 ± 4.2 pmol/l [p = 0.515]). The insulin response after the glucose load was similar in the two groups. Additionally, 2 h C-peptide values fell significantly in both groups (−0.2 ± 0.8 nmol/l [p = 0.018] vs −0.3 ± 1.1 nmol/l [p = 0.045]), although the 3-h C-peptide value was only significantly reduced in the fibre group (−0.2 ± 0.6 nmol/l [p = 0.013] vs −0.1 ± 0.7 nmol/l [p = 0.504]).

Fasting indices such as HOMA-IR and dynamic indices, combining insulin and glucose values during the OGTT, such as glucose effectiveness (Sg; data not shown) and Cederholm index, improved equally in the two groups (Fig. 1d, e). HIC improved similarly in both groups (Fig. 1f).

Further secondary outcomes revealed significant changes in C-reactive protein (CRP) (within both groups), IL-18 and chemerin (placebo group only), and leucocyte count and leptin (fibre group only), but the only significant difference between the groups was for leucocyte count (see electronic supplementary material [ESM] Table 1). During the first year of the intervention, the change in AUC_Glucose (r = 0.199, p = 0.114), HOMA-IR (r = 0.192, p = 0.125) and Cederholm index (r = −0.189, p = 0.135) did not correlate with the weight change in the fibre group, but showed significant correlations in the placebo group: AUC_Glucose (r = 0.250, p = 0.043), HOMA-IR (r = 0.284, p = 0.021) and Cederholm index (r = −0.369, p = 0.002).

Analysis for interaction with sex did not reveal significant results, but sex-specific analyses showed that, in women, only the fibre group achieved significant changes in 2 h plasma glucose (−0.88 ± 1.59 mmol/l [p ≤ 0.001] vs −0.22 ± 1.52 mmol/l [p = 0.311]), AUC_Glucose (−65.4 ± 175.1 mmol/l × min [p = 0.007] vs −28.8 ± 158.2 mmol/l × min [p = 0.242]), insulin levels (fasting, 2 h, 3 h and AUC; data not shown) and HOMA-IR (−0.30 ± 1.04 [p = 0.034] vs −0.17 ± 1.06 [p = 0.261]). Of these, only the improvement in 2 h plasma glucose levels was significantly different between the fibre and placebo groups in women (total difference 0.67 ± 0.31 mmol/l; p = 0.015, adjusted for weight change; Fig. 2). The relatively small and less balanced subgroup of men was not analysed separately.

Open image in new window

One of the main goals of the lifestyle intervention course was a healthy high-fibre diet. The individual change in diet was assessed using food records, and the evaluation showed that, without the fibre supplement, the participants in both groups were not able to increase their fibre intake (+0.9 ± 8.0 g vs +0.5 ± 7.0 g) despite advice to do so (Fig. 3). However, the participants in the fibre-supplemented group raised their average daily fibre intake from about 11.5 ± 9.3 g to 34.4 ± 8.3 g (p ≤ 0.001). The total energy intake as well as the fat intake of the participants decreased during the dietary intervention in both groups. In the fibre group, also the mean protein and carbohydrate intake was significantly lower after the first 12 months (Table 2).

Open image in new window

Table 2

 Changes in dietary habits during the first year of the intervention

Variable	Placebo (n = 59)		Fibre (n = 58)
	Baseline	12 months	Baseline	12 months
Total energy intake (kJ/day)	8862 ± 2536	8205 ± 2339*	8795 ± 2059	7598 ± 1619***
Fat intake (g/day)	82 ± 28	71 ± 29*	80 ± 26	66 ± 22*
Protein intake (g/day)	84 ± 23	81 ± 22	81 ± 24	71 ± 20*
Carbohydrate intake (g/day)	225 ± 72	217 ± 56	225 ± 56	206 ± 44*
Total dietary fibre intake (g/day)	23 ± 8	23 ± 7	23 ± 7	22 ± 7
Insoluble	15 ± 5	15 ± 5	15 ± 5	15 ± 5
Soluble	7 ± 2	5 ± 4***	7 ± 5	5 ± 3***
Alcohol (g/day)	9 ± 15	7 ± 11	10 ± 16	6 ± 8***

Data are mean ± SD

Nutrient intake was calculated from 4 day food records

Significant change within the group: *p ≤ 0.05, ***p ≤ 0.001. Individual variables did not differ significantly between groups

After the second year of the study, 2 h plasma glucose values were reduced in the fibre group by −0.48 ± 1.81 mmol/l (p = 0.020) and in the placebo group by −0.22 ± 2.25 mmol/l (p = 0.366) compared with baseline values (difference not significant). Surprisingly, the improvement in insulin sensitivity and HIC that could be observed after 1 year remained, only slightly diminished, in both groups after 2 years (Fig. 4), with no significant difference between the groups.

Open image in new window

The main finding of our study is a significantly stronger reduction in HbA_1c and 2 h glucose levels (in women) after 1 year of supplementation with insoluble dietary fibre in individuals with IGT. Additionally, the incidence of diabetes was 42% lower in the fibre group, although this did not reach statistical significance. Although significant improvements in HbA_1c and glucose levels did not persist during the second year, the reduced incidence of diabetes in the fibre group was maintained.

Prevention of diabetes by lifestyle intervention has been investigated for several decades. Currently, the predominant approach is based on weight loss and increased physical activity. In these paradigms, dietary counselling includes low dietary total and saturated fat consumption, reduced cholesterol intake and increased fibre intake. With these approaches, recent lifestyle intervention programmes have been able to reduce the incidence of diabetes by 40–60% [1, 2]. Their counselling programmes were usually successful only with regard to weight loss and physical activity, whereas goals for fibre intake were achieved by as little as 25% of participants [2, 15, 28]. Thus, improving the intake of dietary fibre is one of the most taxing problems, as they are present in foods with high caloric density, such as whole-grain bread, or are poorly accepted by participants due to taste (vegetables) or gastrointestinal side effects (grains and legumes).

Notably, the metabolic role of dietary fibre is controversial, as only insoluble fibre has been shown in epidemiological studies to be associated with a decreased risk of type 2 diabetes [3, 5, 6, 29]. This is the first randomised intervention study demonstrating an improvement in HbA_1c over 1 year, a lower incidence of diabetes over 2 years and significantly reduced blood glucose levels upon intake of insoluble cereal fibre over 1 year in women. Similar studies have not been performed with either insoluble or soluble fibre. Therefore, the importance of solubility and fermentability of dietary fibre and the role of the microbiome in long-term improvements in blood glucose levels requires targeted studies [7].

We found a similar improvement in HIC and HOMA-IR in both groups. Weight loss tended to be slightly higher in the placebo group (without a statistically significant difference between the groups) and was almost non-existent at the end of the study, thereby excluding the hypothesis that the observed effects of insoluble cereal fibre were linked to weight changes [29]. Also, metabolic improvements did correlate with weight loss in the placebo group, but not in the fibre group, supporting independence of antidiabetic fibre effects from weight loss. These findings contradict recent studies showing greater weight loss with high-fibre diets, especially in individuals with impaired fasting glucose [30]. The metabolic effects were more pronounced and significant in the female subgroup, which may indicate sex-specific responses. Alternatively, the male group might have been too small.

As a strength of our study, the control group received the same lifestyle interventions known to improve glucose metabolism, but the significant effects of fibre supplementation could nevertheless be demonstrated. Evaluating dietary compliance, individuals in both groups were able to only slightly increase their fibre intake by conventional dietary improvements besides the supplement. Only 14.3% in the placebo group and 11.2% in the fibre group were able to achieve an average daily fibre load at the recommended level (15 g/4200 kJ). When the supplement was included, 67.7% of the fibre group participants managed to achieve the recommendation. Therefore the significant additional improvement in the cereal fibre group supports the use of insoluble fibres as one component of diabetes prevention. In fact, insoluble fibres are widely available byproducts of flour production and could safely be added to numerous foods, thereby improving the population’s cereal fibre intake. The absence of gastrointestinal side effects is another positive characteristic of this type of fibre, which underlines that fermentation is most likely not necessary for the beneficial effects on glucose metabolism [8, 29].

We experienced a dropout rate of 33%, with an even distribution between the two study groups, which is an excellent outcome for a 2 year supplementation study and is comparable to other lifestyle intervention studies [1, 2, 6, 16, 17].

Long-term observation after the end of dietary counselling demonstrated that dietary patterns again deteriorated rapidly and weight maintenance fell in most of the participants. Two-hour glucose levels increased again in both study groups, although supplementation was continued during the 1 year follow-up period. We therefore conclude that a 1 year follow-up after the end of dietary counselling was sufficient to cover the main time frame of dietary rebound. Surprisingly, insulin sensitivity and HIC remained similar to the year before.

A placebo supplementation, especially in case of a study focusing on fibre intake, is very hard to develop. We decided to provide two supplements, which were equal in colour, digestibility, taste and caloric content to assure blinding over the entire study period and to avoid side effects by placebo or fibre supplement. It was considered the best option to use the slowly digestible carbohydrate isomaltulose as the main component of the placebo drink.

Limitations of the present study need mentioning. First, as is usual in most dietary trials, the sex distribution is skewed towards women, leading to a relatively small male subgroup, which then tends to be statistically unbalanced regarding body weight and glycaemic variables. However, by analysing the sexes separately, we were able to demonstrate that the pronounced effect in terms of glycaemic improvement was statistically significant in the large female subgroup, which was highly balanced regarding baseline anthropometric values.

Second, dietary compliance in terms of regular food intake and supplementation was controlled by self-reporting questionnaires. No biomarker is available to validate the reported fibre intake since alkylresorcinols (used to monitor cereal consumption) are not generated from insoluble cereal fibre. However, in general, we noticed an excellent correlation between reported energy intake and body weight changes, with an even distribution and only a few under- and over-reporters. We therefore assume that the dietary counselling was capable of stimulating acceptable dietary documentation from most volunteers.

Finally, there is no randomised control group, which would allow a precise estimation of the cumulative incidence of diabetes if there were no intervention at all. However, previous successful diabetes prevention trials have rendered the use of control groups without intervention unethical.

In summary, we cannot clearly support major benefits from a 2 year randomised placebo-controlled supplementation with insoluble low-digestibility dietary fibre on glycaemic metabolism, as all predefined outcomes failed to reach statistical significance. Effects on secondary outcomes, such as HbA_1c or AUC_Glucose, were pronounced in the first year, particularly in the more balanced female subgroup. Metabolic improvements in 2 h glucose levels were connected to improved fasting insulin sensitivity, insulin secretion and HIC. Fasting glucose levels did not change significantly.

Diabetes incidence was, non-significantly, 42–43% lower in the fibre group over the entire study period. This effect size is comparable with that of other lifestyle studies, which compared dietary counselling with sham intervention. Dietary counselling alone was not effective in increasing fibre intake in a significant and relevant manner, which highlights the need for fibre supplementation in order to achieve relevantly higher cereal fibre intake.

Supplementation did not lead to relevant gastrointestinal side effects, was accepted with moderate to high compliance and was associated with dropout rates comparable with those of other large lifestyle intervention studies with less intense dietary treatment [1, 2]. Additionally, as a result of this, the participants were unable to identify their treatment allocation by self-observation of gastrointestinal side effects.

In modern developed societies, foods for which health claims are made are often more expensive and consumed predominantly by people with a higher awareness of healthy eating. Thus, people with low income or lower educational levels are not only at higher risk of becoming diabetic, but are also less likely to eat healthy food. It therefore seems both plausible and useful not to advertise fibre as an especially healthy component, but to supplement various types of food—cheap and expensive, for snacks and for regular eating—with dietary fibre.

Further studies are required to assess the need for and effectivity of supplementation with insoluble dietary fibre, as it is well accepted owing to its low rate and intensity of side effects and may still prove to be a promising way to decrease the incidence of diabetes on a population level.