Hostname: page-component-7c8c6479df-5xszh Total loading time: 0 Render date: 2024-03-18T15:00:55.729Z Has data issue: false hasContentIssue false

Probiotics and dietary counselling contribute to glucose regulation during and after pregnancy: a randomised controlled trial

Published online by Cambridge University Press:  19 November 2008

Kirsi Laitinen*
Affiliation:
Department of Biochemistry and Food Chemistry, University of Turku, 20014Turku, Finland Functional Foods Forum, University of Turku, 20014Turku, Finland
Tuija Poussa
Affiliation:
Stat-Consulting, 33230Tampere, Finland
Erika Isolauri
Affiliation:
Department of Paediatrics, University of Turku, 20014Turku, Finland Department of Paediatrics, Turku University Central Hospital, 20520Turku, Finland
*
*Corresponding author: Dr Kirsi Laitinen, fax +358 2 333 6862, email kirsi.laitinen@utu.fi
Rights & Permissions [Opens in a new window]

Abstract

Balanced glucose metabolism ensures optimal fetal growth with long-term health implications conferred on both mother and child. We examined whether supplementation of probiotics with dietary counselling affects glucose metabolism in normoglycaemic pregnant women. At the first trimester of pregnancy 256 women were randomised to receive nutrition counselling to modify dietary intake according to current recommendations or as controls; the dietary intervention group was further randomised to receive probiotics (Lactobacillus rhamnosus GG and Bifidobacterium lactis Bb12; diet/probiotics) or placebo (diet/placebo) in a double-blind manner, whilst the control group received placebo (control/placebo). Blood glucose concentrations were lowest in the diet/probiotics group during pregnancy (baseline-adjusted means 4·45, 4·60 and 4·56 mmol/l in diet/probiotics, diet/placebo and control/placebo, respectively; P = 0·025) and over the 12 months' postpartum period (baseline-adjusted means 4·87, 5·01 and 5·02 mmol/l; P = 0·025). Better glucose tolerance in the diet/probiotics group was confirmed by a reduced risk of elevated glucose concentration compared with the control/placebo group (OR 0·31 (95 % CI 0·12, 0·78); P = 0·013) as well as by the lowest insulin concentration (adjusted means 7·55, 9·32 and 9·27 mU/l; P = 0·032) and homeostasis model assessment (adjusted means 1·49, 1·90 and 1·88; P = 0·028) and the highest quantitative insulin sensitivity check index (adjusted means 0·37, 0·35 and 0·35; P = 0·028) during the last trimester of pregnancy. The effects observed extended over the 12-month postpartum period. The present study demonstrated that improved blood glucose control can be achieved by dietary counselling with probiotics even in a normoglycaemic population and thus may provide potential novel means for the prophylactic and therapeutic management of glucose disorders.

Type
Full Papers
Copyright
Copyright © The Authors 2008

Early pregnancy is characterised by normal tolerance to glucose and insulin. In late pregnancy, in contrast, an increase is observed in the serum insulin concentration accompanied by insulin resistance. These metabolic adaptations aim to promote fetal growth by shunting metabolic fuels to the fetus instead of the mother, as well as preparation for breast-feeding. In some pregnant women these adaptive processes are exaggerated and lead to impaired glucose tolerance. Such individuals are predisposed to gestational diabetes mellitus and consequently to type 2 diabetes(Reference Ben-Haroush, Tyogev and Hod1). In the case of the child, impaired maternal glycaemia predisposes toward macrosomia(Reference Östlund, Hanson and Björklund2) and impaired glucose tolerance(Reference Plagemann, Harder and Kohlhoff3), which may develop even when maternal estimates are within normal reference ranges, i.e. not classified as gestational diabetes mellitus(Reference Clausen, Burski and Oyen4). A higher-than-optimal glucose level, now acknowledged to be more common than anticipated, may involve long-term effects not only on the mother but also on the child(Reference Godfrey and Barker5Reference Ahrén7). Indeed, the effects on the infant of maternal nutrition during pregnancy may initiate a cascade of metabolic and immunoinflammatory conditions manifested in later life.

We chose a combined dietary counselling and probiotic intervention to target maternal glucose metabolism, in view of the importance to maintain normoglycaemia throughout pregnancy. Previous dietary interventions with primarily reduced energy and fat intakes as well as increased fibre intakes have resulted in improved glucose tolerance test results(Reference Yamaoka and Tango8). Recent experimental evidence, on the other hand, points to a role for the gut microbiota composition in the harvesting and storage of nutrients(Reference Bäckhed, Ding and Wang9Reference Bäckhed, Manchester and Semenkovich11). The approach may also be justified by the demonstration that diet and microbiota may exert their effects via similar signalling pathways in regulating immune responses(Reference Goldblum, Grann and Ding12Reference Laitinen, Hoppu and Hämäläinen14). Immunoinflammatory processes and prevailing systemic low-grade inflammation may contribute to the metabolic conditions affecting glucose metabolism(Reference Shoelson, Lee and Goldfine15). In a randomised clinical trial from early pregnancy, 256 women were allocated to three groups: modification of dietary intake according to current recommendations with probiotics or placebo and a control group receiving placebo only. The women were followed clinically and their glucose metabolism repeatedly evaluated from early pregnancy up to 12 months postpartum.

Methods

Participants

Altogether 256 pregnant women were recruited to participate in a randomised, prospective, parallel-group, combined dietary counselling and probiotics intervention study from April 2002 to November 2005 (NCT00167700; section 3, http://www.clinicaltrials.gov). The overall aims of the study were to optimise maternal dietary intake and metabolism to advance maternal health and thus to reduce the risk of disease in the child. The present report explores the impact of intervention on maternal nutrition with the main focus on glucose metabolism. The subjects were informed about the study by leaflets distributed during their first visit to maternal welfare clinics in the city of Turku and neighbouring areas in South-West Finland. Interested recipients contacted the research nurse, who gave further information on the study and scheduled their first visit to the study clinic in Turku University Central Hospital. Women were eligible for participation if they were at less than 17 weeks' gestation and had no metabolic or chronic diseases such as diabetes. The study complies with the Declaration of Helsinki as revised in 2000. Written informed consent was obtained from the participants and the study protocol was approved by the Ethics Committee of the Hospital District of South-West Finland.

Intervention and study conduct

Study visits took place three times during pregnancy (at 13·9 (sd 1·6), 23·8 (sd 1·4) and 33·9 (sd 1·4) weeks of gestation) and at 1, 6 and 12 months postpartum. At baseline, subjects were randomly assigned to three study groups (Fig. 1) according to computer-generated block randomisation of six women: dietary counselling with probiotic capsules (diet/probiotics); dietary counselling with placebo (diet/placebo); controls (control/placebo). The randomisation list was generated by a statistician (T. P.) who was not involved in recruitment or study visits. Sealed envelopes contained subject numbers corresponding to numbered probiotics and placebo containers and information on whether the subject would receive dietary counselling. All pregnant women participating in the study also attended communal well-women clinics.

Fig. 1 Flow chart of the study.

At the first study visit the envelopes were opened by the research nurse and nutritionist in the presence of each study subject in their order of recruitment. The random allocation sequence was thus concealed until interventions were assigned. Research nurses and researchers ensured that capsules with corresponding numbers were given to the subjects and that appropriate dietary counselling intervention was carried out. The capsule containers were numbered according to the randomisation list by a member of the research group not involved with the conduct or reporting of the study. The trial data were collected on printed case record forms and the members of the research group performed data entry. All data were kept confidential.

Randomisation to receive probiotics (Lactobacillus rhamnosus GG, ATCC 53 103, Valio Ltd, Helsinki, Finland and Bifidobacterium lactis Bb12, Chr. Hansen, Hoersholm, Denmark, 10Reference Turnbaugh, Ley and Mahowald10 colony-forming units/d each) or placebo (microcrystalline cellulose and dextrose anhydrate; Chr. Hansen, Hoersholm, Denmark) in the dietary counselling groups took place in a double-blind manner, while the control group received placebo in single-blind manner (Fig. 1). The choice of the probiotic combination was based on in vitro results(Reference Ouwehand, Isolauri and Kirjavainen16), and previous clinical intervention studies suggesting that L. rhamnosus GG promotes a bifidogenic microbiota and bifidobacteria dominate the microbiota of healthy breast-fed infants(Reference Benno, He and Hosoda17, Reference Kalliomäki, Kirjavainen and Eerola18), with high bifidobacteria levels linked to the risk of allergy(Reference Kalliomäki, Salminen and Arvilommi19). Probiotics and placebo capsules and contents looked, smelled and tasted identical. Dosing with standard content capsules commenced at the first study visit and lasted until the end of exclusive breast-feeding. All capsules were stored at +5°C and the viability of the probiotic capsules was confirmed by regular analysis of blind in the laboratory under Professor S. Salminen. Compliance in consumption of study capsules was assessed by interview.

Dietary counselling given by a dietitian at each study visit aimed to modify dietary intake to conform with that currently recommended(20, Reference Becker, Lyhne and Pedersen21), particular attention being paid to the quality of dietary fat. Achievement of the recommended diet was supported by providing participants with readily available food products of favourable fat composition (for example, rapeseed oil-based spreads and salad dressing) to be consumed at home. The counselling and food products provided have been described in detail elsewhere(Reference Piirainen, Isolauri and Lagström22). Dietary intake was assessed at each trimester using 3 d food diaries. Energy and nutrient intakes were calculated with a Micro-Nutrica® computerised program (version 2.5; Research Centre of the Social Insurance Institution, Turku, Finland).

At baseline, background information concerning education and parity was collected by interview. Total gestational weight gain was calculated by subtracting self-reported pre-pregnancy weight from that recorded at a prenatal visit or at hospital within 1 week before delivery. Information regarding children's birth weights and heights and the course of pregnancy was obtained from hospital records. On the morning of each visit, a 10 h overnight fasting blood sample was drawn from the antecubital vein for the analysis of glucose and glycated Hb A1C on the day of sampling, whilst serum was stored at − 70 °C for the group analysis of insulin.

Analytical methods

On the day of sampling, plasma glucose concentration was measured by an enzymic method utilising hexokinase in a Modular P800 automatic analyser (Roche Diagnostics GmbH, Mannhein, Germany) and blood glycated Hb A1C was measured by ion-exchange HPLC by the Bio-Rad VariantTM II Haemoglobin A1C Program (Bio-Rad Laboratorioes, Marnes-la-Coquette, France). Serum insulin concentration was measured by immunoelectrochemiluminometric assay in a Modular E170 automatic analyser (Roche Diagnostics GmbH). To evaluate insulin sensitivity, the quantitative insulin sensitivity check index (QUICKI) was calculated as described by Katz et al. (Reference Katz, Nambi and Mather23). Homeostasis model assessment (HOMA) was calculated using a formula devised by Matthews et al. (Reference Matthews, Hosker and Rudenski24). All personnel who handled or analysed blood samples were blind to the intervention. Glucose challenge screening tests were taken from those performed in well-women clinics at 26 to 28 weeks of gestation according to standard procedures for women fulfilling the criteria for at-risk pregnancy: pre-pregnancy BMI over 25 kg/m2; age over 40 years; gestational diabetes mellitus in a previous pregnancy; previous delivery of a child weighing more than 4500 g; detection of glucose in the urine or suspicion of a macrosomic fetus in the present pregnancy.

Plasma glucose concentrations above 4·8 mmol/l during pregnancy and 5·6 mmol/l in the non-pregnant state, the percentage of glycated Hb in total Hb above 6·5 % and a serum insulin concentration above 26 mU/l were considered raised. Results of the glucose challenge tests were considered pathological if an increased fasting glucose value (4·8 mmol/l) was combined with at least one abnormal post-glucose measurement (blood glucose >10·0 mmol/l at 1 h or >8·7 mmol/l at 2 h) according to reference values in Turku University Central Laboratories. A higher QUICKI and lower HOMA were taken to indicate better insulin sensitivity in comparison of differences amongst the groups.

Outcome measures

The primary outcome measure to explore the effects of intervention on the mother was glucose metabolism, characterised by plasma glucose concentration, blood glycated Hb A1C, serum insulin and HOMA and QUICKI indices. The measurements were made at the first trimester (baseline) and third trimester of pregnancy, and at 1, 6 and 12 months postpartum, the primary time points being the third trimester of pregnancy and 12 months postpartum. Secondary outcomes were dietary energy-yielding nutrients assessed from food diaries, which were analysed to explain changes in glucose metabolism.

Statistical analyses

Data were analysed with SPSS (version 14.0; SPSS Inc., Chicago, IL, USA) by a statistician (T. P.) independent of clinical evaluations. The primary sample size calculations were based on infant sensitisation assessed by skin prick testing at the age of 12 months. We estimated that the required sample size for analyses of glucose metabolism was sixty-six per group to detect a difference in blood glucose of 0·20 mmol/l between groups statistically significant with a 0·05 two-sided significance level and 90 % power. We assumed the common sd to be 0·35. Thus the same fixed sample size ensured that the power was sufficient also for analyses of glucose metabolism.

The baseline and clinical variables were analysed using the χ2 test, ANOVA or the Kruskal–Wallis test (5 min Apgar). Missing values for glucose metabolism (at most one during pregnancy and one during the postpartum period) were computed using the group mean or geometric mean, as linear extrapolation or interpolation methods were not appropriate due to the substantial inherent non-linear within-subject fluctuation. Serum insulin and HOMA were skewed to the right and were logarithmically transformed before analysis.

Comparison of glucose metabolism between the three study groups at the third trimester of pregnancy or at 12 months postpartum was made by analysis of covariance (ANCOVA) and that in the postpartum period (1, 6 and 12 months) was analysed using ANCOVA for repeated measurements. In both cases the baseline was included as a continuous covariate. The results are given as baseline-adjusted means or geometric means with 95 % CI or standard deviations. Paired group comparisons were Bonferroni-adjusted. The proportions of subjects with elevated glucose concentrations ( ≥ 4·8 mmol/l during pregnancy, ≥ 5·6 mmol/l postpartum) were compared between the study groups using the χ2 test. Results of group comparisons are given as OR with 95 % CI. For dietary intake the study groups were compared at the third trimester, during the postpartum period and at 12 months postpartum using the same methods as described for glucose metabolism.

The analyses were based on the intention-to-treat population, apart from the twenty-three women who were pregnant again by the end of the follow-up of 12 months and were excluded, a new pregnancy being considered to be a strong confounder.

Results

The participants (Table 1) were Caucasian. The majority had higher college or university education (79 % in the diet/probiotics, 69 % in the diet/placebo and 79 % in the control/placebo groups; P = 0·210) and were expecting their first child (65 % in the diet/probiotics, 51 % in the diet/placebo and 57 % in the control/placebo groups; P = 0·197). The infants were delivered at term and their mean heights and weights were within population reference ranges (Table 1). The mean duration of exclusive breast-feeding and thus the duration of the probiotics/placebo intervention did not differ amongst the study groups, nor did the groups diverge with regard to pregnancy weight gain. Of the total participating women, 99·5 % (216 out of 217), 99 % (209 out of 212) and 95 % (195 out of 205) at the second, third and fourth study visit, respectively, reported without significant difference between groups that they consumed the capsules regularly daily. An adverse event was not the reason for non-compliance in any case. On initiation of capsule consumption 7 % (five out of seventy-three) of the women in the diet/probiotics group, 8 % (six out of seventy-five) in the diet/placebo group, 3 % (two out of sixty-nine) in the control/placebo group and 6 % (thirteen out of 217) in all three groups together reported gut-associated adverse events including flatulence, loose stools or constipation. Subsequently the prevalence of reported symptoms was reduced to 2 % and 0·5 % at subsequent study visits.

Table 1 Characteristics of the women and their infants

(Mean values and standard deviations)

* ANOVA.

Kruskal–Wallis test.

Of the recruited women, 81 % (208 out of 256) were followed up till 12 months postpartum (Fig. 1). The reasons for discontinuing were representative of a normal population of pregnant women. Additionally, twenty-three women were pregnant again by the end of the follow-up and were excluded from the postpartum analysis.

Although energy intakes did not differ amongst the groups, dietary counselling resulted in changes in the intakes of energy-yielding nutrients compared with controls (Table 2). Particularly the intakes of MUFA and PUFA as a proportion of energy intake were highest in the diet/placebo group; thus, the intake of SFA as a proportion of energy intake was lowest in both intervention groups compared with controls. At the same time the intakes of energy-yielding nutrients among women receiving probiotics (diet/probiotics) and placebo (diet/placebo) did not differ. This allowed comparisons of the effect of the probiotics according to the study design, except for that of the intake of protein as a proportion of energy intake, this being lowest in the diet/probiotics group at the third trimester of pregnancy, but not during the postpartum period.

Table 2 Daily intake of energy and energy-yielding nutrients and dietary fibre at first trimester (baseline) and third trimester of pregnancy, during the postpartum period (mean of 1, 6 and 12 months) and at 12 months postpartum in the study groups*

(Mean values and standard deviations or baseline-adjusted means and 95 % confidence intervals)

ANCOVA, analysis of covariance.

* Number of subjects in analysis 232 (75 in diet/probiotics, 81 in diet/placebo and 76 in control/placebo) at first trimester of pregnancy, 209 (70, 71 and 68) at third trimester of pregnancy, 203 (68, 71 and 64) during the postpartum period and 173 (58, 60 and 55) at 12 months postpartum.

Baseline-adjusted univariate ANOVA.

The group comparisons (diet/probiotics v. control/placebo; diet/probiotics v. diet/placebo; diet/placebo v. control/placebo) are given Bonferroni-corrected.

§ Baseline-adjusted mean of measurements at 1, 6 and 12 months.

Impact of the intervention on glucose metabolism

Glucose concentrations decreased from the first trimester to the third and increased during the 12 months' postpartum period in all study groups alike (Fig. 2). The levels were lowest in the diet/probiotics group throughout the follow-up period and thus results are presented adjusted for baseline, i.e. any differences before randomisation cannot explain the outcome. The difference between the study groups was significant during pregnancy, when the baseline-adjusted means were 4·45, 4·60 and 4·56 mmol/l in the diet/probiotics, diet/placebo and control/placebo groups, respectively (P = 0·025). The same was noted at 12 months after delivery (adjusted means 4·93, 5·22 and 5·06 mmol/l; P = 0·060) and over the 12-month postpartum period (adjusted means 4·87, 5·01 and 5·02 mmol/l; P = 0·025). The diet/probiotics group was distinguishable from the diet/placebo group at the third trimester of pregnancy (P = 0·026), at 12-months postpartum (P = 0·054) and over the entire postpartum period (P = 0·066), and further, from the control/placebo group over the postpartum period (P = 0·048).

Fig. 2 Plasma glucose concentrations (mmol/l) during and after pregnancy in diet/probiotics (n 66; Δ), diet/placebo (n 70; ●) and control/placebo (n 60; ○) groups. Values are means, with 95 % CI represented by vertical bars. There were significant differences amongst the groups at the third trimester of pregnancy (P = 0·025), 12 months postpartum (P = 0·060) and over the postpartum period (P = 0·025) by analysis of covariance, where the baseline was taken as the continuous covariate. Only women with values for each time-point were included in the analysis. Bonferroni-corrected group comparisons for diet/probiotics v. diet/placebo and diet/probiotics v. control/placebo were P = 0·026 and P = 0·165 for the third trimester, P = 0·054 and P = 0·878 for 12 months postpartum, and P = 0·066 and P = 0·048 for over the postpartum period, respectively. The inset shows the proportion (%) of abnormal glucose concentrations in the diet/probiotics (▒), diet/placebo (■) and control/placebo (□) study groups. tr., Trimester. There were differences amongst the groups in the third trimester of pregnancy (P = 0·013), 12 months postpartum (P = 0·189) and over the postpartum period (P = 0·096) by logistic regression, where the baseline was taken as a covariate. Only women with values for each time-point are included.

Although in these healthy pregnant women, the mean plasma glucose concentrations were within normal reference ranges in all study groups, the risk of elevated concentrations was reduced in the diet/probiotics group throughout the study period (Fig. 2, inset). During the third trimester, the diet/probiotics intervention (OR 0·31 (95 % CI 0·12, 0·78); P = 0·013), unlike the diet/placebo intervention (OR 1·26 (95 % CI 0·59, 2·69); P = 0·553), had the capacity to reduce the risk of elevated plasma glucose concentrations compared with the control/placebo treatment. In sequence, over the postpartum period the risk of elevated plasma glucose concentrations remained lower in the diet/probiotics group, albeit not statistically significantly (OR 0·46 (95 % CI 0·14, 1·50); P = 0·197), but not in the diet/placebo group (OR 1·55 (95 % CI 0·61, 3·95); P = 0·360), both compared with the control/placebo group.

Altogether 45 % of the subjects underwent a glucose challenge test in well-women clinics during pregnancy. The prevalence of pathological test results was lowest in the diet/probiotics group (37 % of subjects) compared with the diet/placebo (58 %) and control/placebo (57 %) groups. However, the relative risk was not statistically significantly lowered (OR 0·44 (95 % CI 0·14, 1·38) in the diet/probiotics group and OR 1·03 (95 % CI 0·41, 2·61) in the diet/placebo group compared with the control/placebo group).

Glycated Hb A1C remained within normal ranges throughout the study and was comparable amongst the study groups at the third trimester of pregnancy and at 12 months postpartum, but there was a tendency towards lowered glycated Hb A1C in the diet/probiotics group compared with the diet/placebo group over the postpartum period (Table 3).

Table 3 Blood glycated Hb A1C (percentage of total Hb), serum insulin (mU/l) and insulin sensitivity indices homeostasis model assessment (HOMA) and quantitative insulin sensitivity check index (QUICKI) at the first trimester (baseline) and third trimester of pregnancy, during the postpartum period (mean of 1, 6 and 12 months) and at 12 months postpartum in the study groups*

(Mean values and standard deviations or baseline-adjusted means and 95 % confidence intervals)

* Number of subjects in analysis 232 (76 in diet/probiotics, 81 in diet/placebo and 75 in control/placebo) at first and third trimesters of pregnancy and 196 (66, 70 and 60) during the postpartum period and at 12 months.

Baseline-adjusted univariate ANOVA or ANOVA for repeated measurements.

The group comparisons (diet/probiotics v. control/placebo; diet/probiotics v. diet/placebo) are given Bonferroni-corrected.

§ Baseline-adjusted marginal mean or geometric mean, based on repeated measurements at 1, 6 and 12 months.

Data are presented as geometric mean and standard deviation or baseline-adjusted geometric mean and 95 % CI for insulin and HOMA.

Number of subjects 45 in diet/probiotics group, 44 in diet/placebo group and 34 in control/placebo group.

Impact of the intervention on serum insulin and insulin sensitivity indices

Insulin concentrations as well as insulin resistance, evaluated by the HOMA index, were increased and insulin sensitivity, evaluated by the QUICKI index, was reduced towards the third trimester of pregnancy in all groups. The opposite was noted after delivery, insulin concentration and HOMA index being reduced and QUICKI index increased. Mean serum insulin concentrations, insulin resistance and insulin sensitivity were found to differ amongst the groups throughout the study period (Table 3). This difference, at the third trimester of pregnancy and over the postpartum period, was explained by the lowering effect on serum insulin of the combined dietary and probiotics intervention (diet/probiotics group), which was especially pronounced when compared with controls (control/placebo group). The HOMA index was lowest and QUICKI index highest, indicating improved insulin sensitivity in the diet/probiotics group.

Discussion

Balanced glucose metabolism during pregnancy reduces the risk of pregnancy-related complications(Reference Crowther, Hiller and Moss25) and confers long-term health benefits on both the mother and the child(Reference Plagemann, Harder and Kohlhoff3, Reference Löbner, Knopff and Baumgarten26). In the present study, throughout the study period, combined dietary counselling and probiotics intervention yielded consistently improved glucose metabolism and insulin sensitivity in healthy women, providing the first clinical evidence of an active dialogue between host and microbiota in glucose metabolism. These results in a general normoglycaemic population call for further research in at-risk populations.

Combined dietary counselling and probiotic intervention with L. rhamnosus GG and B. lactis Bb12 moderated plasma glucose concentrations and afforded glycaemic control in healthy young females during and after pregnancy. Previous evidence regarding the effects of probiotics on glucose metabolism has been limited to experimental studies, specifically in mice with existing alterations in glucose metabolism. A diet enriched with L. rhamnosus GG has resulted in improved glucose tolerance test results as well as in reduced blood glycated Hb(Reference Tabuchi, Ozaki and Tamura27) and L. casei administration in reduced glucose levels(Reference Matsuzaki, Nagata and Kado28) in diabetic mice. Likewise the Indian fermented milk product dahi supplemented with L. acidophilus and L. casei (Reference Yadav, Jain and Sinha29) or Lactococcus lactis (Reference Yadav, Jain and Sinha30) delayed the disturbance of glucose metabolism in diabetic rats. The present study, indeed, provided the first evidence of consistently improved glucose metabolism in humans. The impact of the intervention extended several months beyond the period of probiotic consumption, which has also been shown in other clinical studies, for example, the effect of probiotic intervention lasting for up to 7 years in reducing the risk of atopic eczema(Reference Kalliomäki, Poussa and Salminen31). This is probably due to a relatively long duration of probiotic consumption, from early pregnancy until the end of exclusive breast-feeding, and occurring in the critical period of the maturing infant, thus inducing an enduring small change in intestinal microbiota composition, sufficient to stimulate the metabolic change observed in blood glucose metabolism.

The microbes may impact on the glucose metabolism by processing dietary polysaccharides, indigestible by human enzymes, adding to the pool of gastrointestinal absorbable glucose(Reference Bäckhed, Ding and Wang9). The gut microbiota may also enhance glucose storage in adipose tissue by suppressing the fasting-induced adipocyte factor gene transcription with ensuing enhanced lipoprotein lipase activity(Reference Bäckhed, Ding and Wang9). Thus, as a consequence of probiotic consumption, a less saccharolytic microbiota in the gastrointestinal tract may diminish both fermentation of polysaccharides and induction of fasting-induced adipocyte factor gene transcription. In addition to the effects seen as lowered blood glucose levels in our study, it has been recently reported that differences in the gut microbiota content may predict overweight in children, bifidobacterial content and composition being determinants of normal weight(Reference Kalliomäki, Collado and Salminen32). Fermentation of dietary fibre in the gastrointestinal tract is known to be associated with improved glucose metabolism(Reference Delzenne and Cani33), but is not a likely explanatory factor here since dietary intake of fibre did not differ in women having received probiotics or not. Alternatively, the mechanism may be independent of energy harvest and storage as shown by resistance to diet-induced obesity and glucose intolerance(Reference Bäckhed, Manchester and Semenkovich11, Reference Membrez, Blancher and Jaquet34).

We suggest that the observed pronounced effect of probiotics on glucose metabolism is most probably attributable to their immunoregulatory properties. Probiotics elicit powerful anti-inflammatory capabilities by inhibiting the NF-κB pathway, which mediates microbial activation of the immune system through toll-like receptors(Reference Shi, Kokoeva and Inouye35). Regulation of inflammatory pathways by probiotics may be of particular importance due to the fundamental involvement that inflammation plays in insulin resistance(Reference Shoelson, Lee and Goldfine15). The concomitance of elevated blood glucose concentrations, insulin resistance and dyslipidaemia with activation of inflammation pathways(Reference Shi, Kokoeva and Inouye35) is related to an enhanced risk of a range of metabolic disorders, including obesity and CVD(Reference Stampfer, Hu and Manson36, Reference Weiss, Dziura and Burgert37). Indeed, alterations in gut microbiota composition have recently been documented in obesity, providing a target for probiotic intervention(Reference Turnbaugh, Ley and Mahowald10, Reference Kalliomäki, Collado and Salminen32, Reference Ley, Turnbaugh and Klein38). Here we need to acknowledge that specific probiotic strains may influence the microbiota composition in a manner favouring lower circulating lipopolysaccharide levels, possibly ensuing via the CD14 receptor(Reference Cani, Neyrinck and Fava39), that associate with lower insulin resistance and blood glucose levels(Reference Creely, McTernan and Kusminski40, Reference Erridge, Attina and Spickett41). Thus a universal presence of microbes or microbiota per se, may not be crucial in determining glucose-regulating effects but rather specific compositions may form the key factors.

Intriguingly, probiotics were shown here to provide a more profound glucose-lowering effect than dietary counselling alone. This notwithstanding, the impact of dietary counselling, focused on fat composition(Reference Piirainen, Isolauri and Lagström22), is most probably also afforded by the specific regulatory properties of fats, beyond their nutritional value(Reference Calder42). In fact, the microbes and fatty acids engage the same signalling channels through toll-like receptor 4(Reference Shi, Kokoeva and Inouye35) and soluble CD14(Reference Laitinen, Hoppu and Hämäläinen14) of innate immunity. The innate immune system, again, apart from microbial recognition, has been demonstrated to participate in the regulation of glucose metabolism and insulin resistance(Reference Katz, Nambi and Mather35). Furthermore the composition of the gut microbiota has recently been proven instrumental in energy metabolism. High-fat feeding is associated with lower intestinal Bifidobacterium content in mice, and increase in Bifidobacterium content positively correlated with improved glucose tolerance(Reference Cani, Neyrinck and Fava39). It may be projected that the clinical effects are under the same regulatory mechanisms. Thus the present study calls for the precise characterisation of the mechanisms involved in the combined regulatory properties of probiotics and specific dietary compounds.

Modification of gut microbiota composition by probiotics, thereby altering the intestinal immunological milieu, may be seen as a novel means of attaining regulation of glucose metabolism. This dietary approach would offer a cost-effective tool for both prophylaxis and therapy in the metabolic disorders that constitute the metabolic syndrome. The benefit is expected to be most pronounced during the critical period of human development in view of the programming of later diseases by events in the uterus(Reference Godfrey and Barker5).

Acknowledgements

The present study was supported by grants from the Social Insurance Institution of Finland, the Sigrid Juselius Foundation and the Academy of Finland. Provision of food products was by Raisio plc (Raisio, Finland), B. lactis Bb12 by Chr. Hansen (Hoersholm, Denmark) and L. rhamnosus GG by Valio Ltd (Helsinki). We would like to thank Professor Seppo Salminen, University of Turku, for academic assistance and continual organisation of the microbial content analysis of probiotic capsules. The authors declare that there is no personal or financial conflict of interest associated with this paper. The authors' responsibilities were as follows: K. L. and E. I. were responsible for the design of the study, organisation of data collection, and for analysing and reporting the data. T. P. conducted the statistical analysis. All authors contributed to writing and revising of the paper.

References

1Ben-Haroush, A, Tyogev, Y & Hod, M (2004) Epidemiology of gestational diabetes mellitus and its association with type 2 diabetes. Diabet Med 21, 103113.Google Scholar
2Östlund, I, Hanson, U, Björklund, A, et al. (2003) Maternal and fetal outcomes if gestational impaired glucose tolerance is not treated. Diabetes Care 26, 21072111.Google Scholar
3Plagemann, A, Harder, T, Kohlhoff, R, et al. (1997) Glucose tolerance and insulin secretion in children of mothers with pregestational IDDM or gestational diabetes. Diabetologia 40, 10941100.Google Scholar
4Clausen, T, Burski, TK, Oyen, N, et al. (2005) Maternal anthropometric and metabolic factors in the first half of pregnancy and risk of neonatal macrosomia in term pregnancies. A prospective study. Eur J Endocrinol 153, 887894.Google Scholar
5Godfrey, KM & Barker, DJP (2000) Fetal nutrition and adult disease. Am J Clin Nutr 71, 344352.Google Scholar
6Danaei, G, Lawes, CMM, Vander Hoorn, S, et al. (2006) Global and regional mortality from ischaemic heart disease and stroke attributable to higher-than-optimum blood glucose concentration: comparative risk assessment. Lancet 368, 16511659.Google Scholar
7Ahrén, B (2007) Insulin secretion and insulin sensitivity in relation to fasting glucose in healthy subjects. Diabetes Care 30, 644648.Google Scholar
8Yamaoka, K & Tango, T (2005) Efficacy of lifestyle education to prevent type 2 diabetes. Diabetes Care 28, 27802786.Google Scholar
9Bäckhed, F, Ding, H, Wang, T, et al. (2004) The gut microbiota as an environmental factor that regulates fat storage. Proc Natl Acad Sci U S A 101, 1571815723.Google Scholar
10Turnbaugh, PJ, Ley, RE, Mahowald, MA, et al. (2006) An obesity-associated gut microbiome with increased capacity for energy harvest. Nature 444, 10271031.Google Scholar
11Bäckhed, F, Manchester, JK, Semenkovich, CF, et al. . (2007) Mechanisms underlying the resistance to diet-induced obesity in germ-free mice. Proc Natl Acad Sci U S A 104, 979984.Google Scholar
12Goldblum, SE, Grann, TW, Ding, X, et al. . (1994) Lipopolysaccharide (LPS)-binding protein and soluble CD14 function as accessory molecules for LPS-induced changes in endothelial barrier function, in vitro. J Clin Invest 93, 692702.Google Scholar
13Yu, B, Hailman, E & Wright, SD (1997) Lipopolysaccharides binding protein and soluble CD14 catalyze exchange of phospholipids. J Clin Invest 99, 315324.Google Scholar
14Laitinen, K, Hoppu, U, Hämäläinen, M, et al. (2006) Breast milk fatty acids may link innate and adaptive immune regulation: analysis of soluble CD14, prostaglandin E2 and fatty acids. Pediatr Res 59, 723727.Google Scholar
15Shoelson, SE, Lee, J & Goldfine, AB (2006) Inflammation and insulin resistance. J Clin Invest 116, 17931801.Google Scholar
16Ouwehand, A, Isolauri, E, Kirjavainen, PV, et al. . (2000) The mucus binding of Bifidobacterium lactis Bb12 is enhanced in the presence of Lactobacillus GG and Lact. delbrueckii subsp. bulgaricus. Lett Appl Microbiol 30, 1013.Google Scholar
17Benno, Y, He, F, Hosoda, M, et al. . (1996) Effects of Lactobacillus GG yogurt on human intestinal microecology in Japanese subjects. Nutr Today 31, Suppl., 911.Google Scholar
18Kalliomäki, M, Kirjavainen, P, Eerola, E, et al. . (2001) Distinct patterns of neonatal gut microflora in infants in whom atopy was and was not developing. J Allergy Clin Immunol 107, 129134.Google Scholar
19Kalliomäki, M, Salminen, S, Arvilommi, H, et al. . (2001) Probiotics in primary prevention of atopic disease: a randomised, placebo-controlled trial. Lancet 357, 10761079.Google Scholar
20Nordic Working Group on Diet and Nutrition (1996) Nordic Nutrition Recommendations. Scand J Nutr 40, 161165.Google Scholar
21Becker, W, Lyhne, N, Pedersen, AN, et al. (2004) Nordic nutrition recommendations 2004 – integrating nutrition and physical activity. Scand J Nutr 48, 178187.Google Scholar
22Piirainen, T, Isolauri, E, Lagström, H, et al. (2006) Impact of dietary counseling on nutrient intake during pregnancy: a prospective cohort study. Br J Nutr 96, 10951104.Google Scholar
23Katz, A, Nambi, SS, Mather, K, et al. (2000) Quantitative insulin sensitivy check index: a simple accurate method for assessing insulin sensitivity in humans. J Clin Endocrinol Metab 85, 24022410.Google Scholar
24Matthews, DR, Hosker, JP, Rudenski, AS, et al. (1985) Homeostasis model assessment: insulin resistance and β-cell function from fasting plasma glucose and insulin concentrations in man. Diabetologia 28, 412419.Google Scholar
25Crowther, CA, Hiller, JE, Moss, JR, et al. (2005) Effect of treatment of gestational diabetes mellitus on pregnancy outcomes. New Engl J Med 352, 24772486.Google Scholar
26Löbner, K, Knopff, A, Baumgarten, A, et al. (2006) Predictors of postpartum diabetes in women with gestational diabetes mellitus. Diabetes 55, 792797.Google Scholar
27Tabuchi, M, Ozaki, M, Tamura, A, et al. (2003) Antidiabetic effect of Lactobacillus GG in streptozotocin-induced diabetic rats. Biosci Biotechnol Biochem 67, 14211424.Google Scholar
28Matsuzaki, T, Nagata, Y, Kado, S, et al. (1997) Effect of oral administration of Lactobacillus casei on alloxan-induced diabetes in mice. APMIS 105, 637642.Google Scholar
29Yadav, H, Jain, S & Sinha, PR (2007) Antidiabetic effect of probiotic dahi containing Lactobacillus acidophilus and Lactobacillus casei in high-fructose fed rats. Nutrition 23, 6268.Google Scholar
30Yadav, H, Jain, S & Sinha, PR (2006) Effect of dahi containing Lactococcus lactis on the progression of diabetes induced by a high-fructose diet in rats. Biosci Biotechnol Biochem 70, 12551258.Google Scholar
31Kalliomäki, M, Poussa, T, Salminen, S, et al. . (2007) Probiotics during the first 7 years of life: a cumulative risk reduction of eczema in a randomized, placebo-controlled trial. J Allergy Clin Immunol 119, 10191021.Google Scholar
32Kalliomäki, M, Collado, MC, Salminen, S, et al. . (2008) Early differences in fecal microbiota composition in children may predict overweight. Am J Clin Nutr 87, 534538.Google Scholar
33Delzenne, NM & Cani, PD (2005) A place for dietary fibre in the management of the metabolic syndrome. Micronutrients, neutraceutics and functional foods. Curr Opin Clin Nutr Metab Care 8, 636640.Google Scholar
34Membrez, M, Blancher, F, Jaquet, M, et al. . (2008) Gut microbiota modulation with norfloxacin and ampicillin enhances glucose tolerance in mice. FASEB J 22, 24162426.Google Scholar
35Shi, H, Kokoeva, MV, Inouye, K, et al. (2006) TLR4 links innate immunity and fatty acid-induced insulin resistance. J Clin Invest 116, 30153025.Google Scholar
36Stampfer, MJ, Hu, FB, Manson, JE, et al. (2000) Primary prevention of coronary heart disease in women through diet and lifestyle. New Engl J Med 343, 1622.Google Scholar
37Weiss, R, Dziura, J, Burgert, TS, et al. (2004) Obesity and the metabolic syndrome in children and adolescents. New Engl J Med 350, 23622374.Google Scholar
38Ley, RE, Turnbaugh, PJ, Klein, S, et al. (2006) Human gut microbes associated with obesity. Nature 444, 10221023.Google Scholar
39Cani, PD, Neyrinck, AM, Fava, F, et al. (2007) Selective increases of bifidobacteria in gut microflora improve high-fat-diet-induced diabetes in mice through a mechanism associated with endotoxaemia. Diabetologia 50, 23742383.Google Scholar
40Creely, SJ, McTernan, PG, Kusminski, CM, et al. . (2007) Lipopolysaccharide activates an innate immune system response in human adipose tissue in obesity and type 2 diabetes. Am J Physiol Endocrinol Metab 292, E740E747.Google Scholar
41Erridge, C, Attina, T, Spickett, CM, et al. . (2007) A high-fat meal induces a low-grade endotoxemia: evidence of a novel mechanism of postprandial inflammation. Am J Clin Nutr 86, 12861292.Google Scholar
42Calder, PC (2006) Polyunsaturated fatty acids and inflammation. Prostaglandins Leukot Essent Fatty Acids 75, 197202.Google Scholar
43Cani, PD, Neyrinck, AM, Fava, F, et al. . (2007) Selective increases of bifidobacteria in gut microflora improve high-fat-diet-induced diabetes in mice through a mechanism associated with endotoxaemia. Diabetologia 50, 23742383.Google Scholar
Figure 0

Fig. 1 Flow chart of the study.

Figure 1

Table 1 Characteristics of the women and their infants(Mean values and standard deviations)

Figure 2

Table 2 Daily intake of energy and energy-yielding nutrients and dietary fibre at first trimester (baseline) and third trimester of pregnancy, during the postpartum period (mean of 1, 6 and 12 months) and at 12 months postpartum in the study groups*(Mean values and standard deviations or baseline-adjusted means and 95 % confidence intervals)

Figure 3

Fig. 2 Plasma glucose concentrations (mmol/l) during and after pregnancy in diet/probiotics (n 66; Δ), diet/placebo (n 70; ●) and control/placebo (n 60; ○) groups. Values are means, with 95 % CI represented by vertical bars. There were significant differences amongst the groups at the third trimester of pregnancy (P = 0·025), 12 months postpartum (P = 0·060) and over the postpartum period (P = 0·025) by analysis of covariance, where the baseline was taken as the continuous covariate. Only women with values for each time-point were included in the analysis. Bonferroni-corrected group comparisons for diet/probiotics v. diet/placebo and diet/probiotics v. control/placebo were P = 0·026 and P = 0·165 for the third trimester, P = 0·054 and P = 0·878 for 12 months postpartum, and P = 0·066 and P = 0·048 for over the postpartum period, respectively. The inset shows the proportion (%) of abnormal glucose concentrations in the diet/probiotics (▒), diet/placebo (■) and control/placebo (□) study groups. tr., Trimester. There were differences amongst the groups in the third trimester of pregnancy (P = 0·013), 12 months postpartum (P = 0·189) and over the postpartum period (P = 0·096) by logistic regression, where the baseline was taken as a covariate. Only women with values for each time-point are included.

Figure 4

Table 3 Blood glycated Hb A1C (percentage of total Hb), serum insulin (mU/l) and insulin sensitivity indices homeostasis model assessment (HOMA) and quantitative insulin sensitivity check index (QUICKI) at the first trimester (baseline) and third trimester of pregnancy, during the postpartum period (mean of 1, 6 and 12 months) and at 12 months postpartum in the study groups*(Mean values and standard deviations or baseline-adjusted means and 95 % confidence intervals)