The Probiotics in Pregnancy Study (PiP Study): rationale and design of a double-blind randomised controlled trial to improve maternal health during pregnancy and prevent infant eczema and allergy

Worldwide, allergic diseases are the largest group of non-communicable diseases (NCD) with an increasing prevalence in both the developed and developing world [4]. They are also the NCD with earliest onset, children bearing much of the burden of these diseases [2]. The prevalence of atopic eczema has increased two to three fold in the last three decades with 15–30 % of children worldwide [4], and up to 40 % of infants in New Zealand having eczema by 15 months of age [5]. Sixty percent of children developing eczema will do so within the first year of life [4]. About half of the children who develop eczema early in life become sensitised to allergens by 2 years of age [4].

A significant body of research examining the use of probiotics to prevent allergic disease already exists. A meta-analyses of randomised controlled trials (RCT) shows benefits of using probiotic supplements during pregnancy and early infant life to prevent the development of atopic dermatitis [6]; however, a Cochrane review found that the benefit is not significant for immunoglobulin E (IgE) associated atopic dermatitis [7].

A more recent subgroup meta-analysis concluded that pre and post-natal supplementation is effective (OR = 0.61, 95 % confidence interval (CI) 0.52–0.71, p < 0.001) while there was no evidence for post-natal interventions-only being effective (OR = 0.95, 95 % CI 0.63–1.45, p = 0.82) [8]. Meta-analyses also indicate both treatments with Lactobacillus alone or Lactobacillus with Bifidobacterium appear to be protective (OR = 0.70, 95 % CI 0.54–0.89, p = 0.004; OR = 0.62, 95 % CI 0.52–0.074, p < 0.001) [8]. This is consistent with our own previous work in a RCT of 474 infants which showed that Lactobacillus rhamnosus HN001 (HN001) 6×10⁹ cfu/day given daily to mothers from 35 weeks gestation, continuing until 6 months post-partum if breastfeeding and from birth until 2 years in the infant was associated with a significant 50 % reduction in the prevalence of eczema at age 2 [9], 4 [10] and 6 years [11]. While not evident early, by 6 years there was also a significant reduction in skin prick sensitisation in the HN001 group (HR = 0.69, 95 % CI 0.48–0.99) [11].

Almost without exception [12] previous probiotic trials of allergic disease with a pre-birth intervention have commenced at some time in the final 2 months of pregnancy [6]. In our current trial we commence the intervention from 14 to 16 weeks gestation and continue it throughout pregnancy and for 6 months post-partum while breastfeeding, the probiotic being only given to the mother, not directly to the infant. Our justification for an early probiotic intervention is based on evidence showing that fetal production of IgE antibodies occurs before the end of the first trimester and allergen-specific IgE antibodies towards the end of the second trimester [13]. There is also evidence that maternal allergy alters the regulation of antigen-specific responses during pregnancy, with non-allergic mothers showing down-regulation of their (already lower) Th2 responses to specific allergen from mid to late gestation [14]. This down-regulation was absent in allergic mothers. Epidemiological support for the importance of intervention in early pregnancy comes from a longitudinal study which shows that maternal exposure to pollen during the first trimester of pregnancy increased the risk of food sensitisation in the child [15]. The majority of probiotic trials, using a late pregnancy intervention (from 32 to 35 weeks gestation), may therefore have missed the critical window to influence fetal immune responses and thus the later development of allergic disease. This may explain the general lack of effect of probiotics on infant sensitisation. The only study, by Huurre et al. [12], that did show a protective effect of probiotics on sensitisation in the infant used an early pregnancy intervention and the effect was limited to those with sensitised mothers (OR = 0.34, 95 % CI 0.13–0.88). In that study [12], the group of infants with non-sensitised mothers had a significantly increased risk of sensitisation but this finding was not reported in the article.

A RCT with a late pregnancy intervention [16] showed a reduction in child sensitisation also among children with allergic mothers (defined according to the presence of disease not atopic sensitisation). There was no effect among children of allergic fathers, highlighting the relative importance of the mother in influencing fetal immune development. In contrast, two probiotic trials [17, 18] have shown increased rates of sensitisation in all children taking probiotics but neither of these studies used Lactobacillus rhamnosus, and one intervened in infants only [18]. Confirmation of the role of probiotics in the development of atopic sensitisation in a larger study with an early pregnancy intervention may allow the targeting of a probiotic intervention to those who are most likely to benefit, i.e. in those with maternal sensitisation, while avoiding the possible increased risk of sensitisation among those without maternal sensitisation. As sensitisation is associated with more severe and persistent eczema [19], a probiotic intervention from early pregnancy, if found to protect against sensitisation, may also reduce the prevalence of clinically important eczema.

Previous pre and/or post-natal intervention studies also vary according to who received the probiotic intervention after birth: mother or infant or both. There have been two studies with an intervention only in the mothers (both from 36 weeks gestation and during breastfeeding) which have both shown an effect on eczema at 2 years that is as strong as that seen when the probiotics were also administered directly to the infant [20, 21]. Alteration in breast milk cytokine levels associated with allergic outcomes in those receiving probiotics suggest that immune modulation may also occur through this pathway [12, 21‐23], and this indicates that post-natal maternal supplementation while breastfeeding may also be important. In contrast to many of the previous probiotic studies our current study administers the probiotics directly to women only, and if proven effective provides an intervention that is easier to administer as it does not require administration of probiotics to new-born infants. This would make the intervention more easily adopted into practice.

Accompanying the worldwide trends in obesity, the rate of gestational diabetes mellitus (GDM) is also increasing in both the developed and the developing world [24]. Using the International Association of Diabetes and Pregnancy Study Group (IADPSG) [25] diagnostic criteria (fasting plasma glucose ≥5.1 mmol/l, or 1-hour post-75 g load ≥10.0 mmol/l, or 2-hour post 75 g load ≥8.5 mmol/l), 18 % of pregnant women in the United States develop GDM during pregnancy [24]. GDM is associated with short and long-term adverse outcomes for both women and infants, including maternal gestational hypertension, polyhydramnios, preeclampsia, delivery of large-for-gestation infants, instrumental or caesarean delivery, and maternal death [24, 26]. Adverse infant outcomes include preterm birth, shoulder dystocia, macrosomia, congenital defects, and neonatal complications such as hypoglycaemia, jaundice and respiratory distress [24]. In addition, in the longer term, women with GDM are at increased risk of metabolic syndrome [27], type 2 diabetes, and cardiovascular disease. Offspring of women with GDM have an increased risk of diabetes, obesity and metabolic issues with evidence of altered insulin secretion and lipid profiles regardless of the infant’s weight [28].

Lifestyle interventions to prevent GDM relating to diet, weight loss and exercise are often unsuccessful [24, 29]; therefore primary prevention of GDM could provide substantial multigenerational health and economic benefits. In a Finnish study [30, 31], among those receiving intensive dietary counselling, probiotic use (Lactobacillus rhamnosus GG and Bifidobacterium lactis Bb12 10¹⁰ cfu/day each) from the first trimester of pregnancy until the end of exclusive breastfeeding was associated with beneficial outcomes for GDM. The diagnostic test used in the Finnish study was a 75 g glucose OGTT with one value exceeding any of the following cut points being considered positive: fasting glucose value ≥4.8 mmol/l, or 1 h blood glucose ≥10.0 mmol/l, or 2 h blood glucose ≥8.7 mmol/l. Using these criteria, the prevalence of GDM was dramatically decreased, 13 % in women given dietary advice plus probiotics compared with 36 % in a group given dietary advice only and 34 % in a control group with no intervention (p = 0.003). The authors suggest that this effect may be due to probiotics contributing to glucose regulation during pregnancy [30]. In this same study population, probiotics taken from the first trimester were associated with half the risk of maternal adiposity, defined as having a waist circumference ≥80 cm, at 6 months post-partum (p = 0.03) [32]. A different study using Lactobacillus rhamnosus GG supplementation from 36 weeks gestation to 6 months postnatally to breastfeeding mothers or their infant found no significant change in birth weight adjusted mean BMI in the offspring at 4 and 10 years of age [33]. A more recent study [34] using a short probiotic intervention from 24 to 28 weeks gestation and different probiotic species (Lactobacillus salivarius UCC188 10⁹ cfu/day) in obese pregnant woman did not alter fasting glucose or other maternal outcomes. These findings may indicate that the probiotic species and strain as well as gestation at commencement of intervention, duration of intervention and concurrent diet contribute to the prevention of GDM. Our current study will examine the impact of HN001 supplementation from early pregnancy without altering baseline diet.

The maintenance of healthy vaginal microbiota is important for optimal pregnancy outcomes. Vaginal coliform and streptococcal colonisation occurs by intestinal microbes ascending from the perineum, and a healthy vaginal flora contains a predominance of organisms from the Lactobacillus genus [35]. Lactobacilli protect the vagina from pathogenic organisms by producing antimicrobial agents such as hydrogen peroxide and bacteriocins, competing for nutrients, adhering to the epithelial surfaces, maintaining the vaginal pH through lactic acid production, and by immune modulation [36, 37]. Both bacterial vaginosis (BV) and Group B Streptococcus (GBS) colonisation are associated with depleted vaginal lactobacillus populations [38, 39] and are associated with negative pregnancy outcomes.

Internationally the prevalence of BV is high e.g. 25 % in pregnant women in USA [36]. BV is associated with preterm labour, premature rupture of membranes, spontaneous abortion, and chorioamnionitis [39]. Premature birth predisposes the infant to a range of other serious health issues including respiratory distress syndrome, intraventricular haemorrhage, leukomalacia, retinopathy, necrotising enterocolitis and prolonged hospitalisation with the associated costs to the health system [40]. Eighty percent of preterm deliveries result from premature rupture of the membranes and spontaneous preterm labour [40]. Maternal infections are associated with 30–50 % of preterm labours [40].

Antibiotic therapy (metronidazole) is recommended as treatment for BV, yet a large placebo-controlled trial did not find that metronidazole reduced the occurrence of preterm delivery or other adverse perinatal outcomes [41]. However a BV treatment study showed that a combination of orally administered Lactobacillus rhamnosus GR-1 and Lactobacillus reuteri RC-14 and metronidazole doubled the cure rate compared to metronidazole alone [42]. Efficacy of current antibiotic treatments for BV is variable and recurrence is common (40 % at 3 months) [37]. In addition, antibiotic resistance in vaginal pathogens is an increasing concern [37]. Exploration of the role of probiotics in the prevention of BV-related adverse outcomes in pregnancy is in its infancy. A Cochrane review of probiotics for preventing preterm labour found an 81 % reduction in the risk of genital infection with the use of probiotics (RR 0.19; 95 % CI 0.08 to 0.48); however, there were insufficient trials to determine the effect on preterm birth and other complications [40]. Another review [43] of probiotics in the treatment and prevention of BV suggests a role for a range of Lactobacillus species in managing urogenital infections but studies with outcomes among pregnant women were absent.

GBS is a commensal bacterium found in the gastrointestinal and genitourinary tracts of 30 % of healthy adults [44]. Worldwide GBS vaginal colonisation in pregnant women varies with rates of between 4 and 36 % in European countries, and most countries having rates higher than 20 % [44‐46]. Usually maternal GBS is asymptomatic, however it can cause endometritis, chorioamnionitis, and bacteraemia in pregnant women, and may cause stillbirth [47], and is the leading cause of early onset Group B Streptococcal septicaemia and meningitis in infants [44, 48]. Up to 50 % of babies born to colonised women acquire the infection and 1–2 % of colonised infants become seriously ill [49]. Despite the low rates of early onset infant GBS illness (1–4 cases/1000 live births) the consequences are potentially fatal, including sepsis, bacteraemia, pneumonia and meningitis with associated long term neurodevelopmental defects [44, 49]. Most countries use screening to detect vaginal GBS colonisation in pregnant women at 35–37 weeks. Those colonised receive intra-partum antibiotics to reduce the risk of vertical transmission to the infant during birth.

Lactobacilli have been shown to have inhibitory effects on GBS growth in vitro [50, 51] and vaginal Lactobacillus counts from pregnant women are inversely related to GBS colonisation [38]. Although the popular literature supports the use of probiotics in the prevention of GBS, there has been only one feasibility study examining an oral probiotic supplementation effect on GBS, and while non-blinded and not fully powered this study did find reduced GBS colony counts in the participants taking oral probiotic supplements [35].

Studies [43], including our own [9], have shown lactobacilli survive passage through the gastrointestinal tract, indicating that oral delivery of lactobacilli is feasible and may be expected to impact the composition of vaginal flora. In addition, HN001 has been demonstrated to produce bacteriocins [52], and appears not to have genes commonly associated with resistance to peroxide [GenBank Acc No. NZ_ABWJ00000000] so we anticipate that oral administration of this organism may favourably influence vaginal flora.

There is a growing literature on how gut microbiota might influence anxiety, depression and cognition via the microbiota-gut-brain axis [53‐56]. Much of this work has been done in preclinical animal trials by intentional manipulation of the animal’s gut microbiota (such as using germ free mice, or treating with probiotics, antibiotics or pathogenic bacteria). Reviews of these studies demonstrate that alterations in anxiety-like or depressive behaviours in animals have been documented in response to manipulation of their gut microbiota [54, 56]. In particular, one study has shown that probiotic supplementation with Lactobacillus rhamnosus decreased anxiety-like and depressive-like behaviours in healthy mice [57].

A range of potential mechanisms by which the gut microbiota affects central nervous system function have been proposed. These include altered microbial composition, immune inactivation, vagal nerve activation, tryptophan metabolism, gut hormone response, and through production of neuro active substances or other metabolites [53, 56].

There is very limited published work describing mood or cognitive outcomes of probiotic interventions in humans. In a double blind, placebo randomised controlled trial; healthy subjects were given probiotics (Lactobacillus helveticus R0052 and Bifidobacterium longum R0175) or placebo for 30 days. The probiotic group had significantly less anxiety and depression than the controls [58]. In a similar study those subjects who initially scored in the lowest third for depressed mood showed significant improvement in symptoms after probiotic treatment [59].

A recent study provides the first direct evidence that probiotics alters brain activity in humans [60]. Daily intake of a mixture of four probiotic strains (Bifidobacterium animalas subsp Lactis, Streptococcus thermophiles, Lactobacillus bulgaricus and Lactococcus lactis subsp Lactis) over four weeks reduced brain activity to an emotional attention task in the regions of the brain that influence the processing of sensory information and emotion. Brain activity was assessed using functional magnetic resonance imaging. No changes in commensal bacteria were observed. The authors argued that the probiotics might interact with the host microbiota to alter their metabolic activity resulting in the production of metabolites that influence brain activity.

The prevalence of postpartum depression is variously estimated to be 10–15 % [61], and women with postpartum depression experience a range of symptoms including general anxiety and dissatisfaction with life, emotional lability, insomnia, confusion, guilt, and suicidality [62]. In addition maternal depression can interfere with mother-infant interactions and adversely affect the infant’s psychological and developmental trajectory at a time of vulnerability [62‐64]. There have been no human studies examining whether probiotics given during pregnancy and breast feeding influence mood. With data emerging to indicate that probiotic effects may be mediated by the gut-brain axis, this study provides an excellent opportunity to assess this further by assessing postpartum depression and anxiety in our study population.

In summary the Probiotics in Pregnancy study aims to investigate if maternal supplementation by HN001 prevents infant eczema and atopic sensitisation by one year and improves maternal health during pregnancy by reducing GDM, BV and GBS, and in the early post-natal period by improving mood. The study design has an early maternal intervention at 14–16 weeks gestation continuing post-natally until 6 months after birth, if breastfeeding, which provides the unique opportunity to study these outcomes, all of which merit consideration on their own.

The PiP Study is a two-centre randomised double blind placebo controlled trial in Wellington and Auckland, New Zealand designed to test the hypothesis that probiotic supplementation by Lactobacillus rhamnosus HN001 (6×10⁹ cfu/day) administered to mothers from 14 to 16 weeks gestation until delivery and continuing until 6 months post-partum, if breastfeeding, will reduce (1) the prevalence of infant eczema and atopic sensitisation by age 12 months and (2) the prevalence of maternal GDM, BV and GBS vaginal colonisation before birth. Table 1 provides an overview of primary, secondary and related outcomes being examined in the study.

The distribution of the Edinburgh Postnatal Depression Scale (EPDS) from Growing up in New Zealand Study [68] was used. With a sample size of 200 in each group and 13 % drop-out rate, using a Wilcoxon Rank-Sum test, the study will have 79 % power to detect a 26 % reduction in EPDS at a 5 % level of significance.

A range of recruitment strategies are used to notify pregnant women about the study and to invite them to contact the study centres to discuss the study and be assessed for study eligibility. These include: enclosing study brochures in information packs routinely given to women by clinicians and midwives in early pregnancy, emailing employees of large organisations, using web-based approaches such as Facebook, arranging for sonographers and phlebotomists to hand out study brochures at the time of early scans and the first antenatal blood collection, radio advertising, newspaper articles as well as pamphlets and posters placed at hospitals, clinics, crèches, libraries and pharmacies and inviting clinicians (midwives, obstetricians and general practitioners) to refer women to the study.

Pregnant women who contact the study centres are screened for eligibility by phone and if eligible are invited to attend an enrolment visit at the study centre when they reach between 14 and 16 weeks gestation.

Table 2 provides a summary of the study timeline and investigations. Women enrolled in the study are provided with a study calendar to assist them to prospectively record data relevant to the study questionnaires. Research staff members are trained in standardised administration of all study questionnaires and data collection procedures.

Questionnaires collect a range of data including confounders (such as previous GDM, parity etc.), effect modifiers (such as use of antibiotics, yoghurt etc.), outcome data, and adverse events.

Our previous study using HN001 in women from 35 weeks gestation until 6 months post birth and in their infants for 2 years found no adverse effects [76]. Similarly other studies of safety using differing strains of Lactobacillus rhamnosus [77], and studies with interventions commencing from the first trimester [31, 78] have found no adverse effects on maternal (gestation length, miscarriage, caesarean delivery) or infant (Apgar score, birth weight and length, head circumference) outcomes.

Adverse event data are collected in all questionnaires. In addition, study doctors review the notes of any participant where there is any event that may require review, such as if the infant is delivered at less than 37 weeks’ gestation; if an infant requires resuscitation at birth; if the infant has any malformations (apart from tongue tie); if the infant requires admission to neonatal unit for more than 48 h for any reason; if the woman or baby had any hospital admission other than the birth admission; if the women or infant have any serious infection (e.g. septicaemia) or any other serious health issues.

An intention-to-treat analysis will be used to assess the effect of HN001 on the 12 month cumulative prevalence of eczema and SCORAD ≥10 using hazard ratios from a Cox’s proportional hazards model, and the point prevalence of atopy at age 12 months using relative rates and a chi-squared test. Chi-square test will also be used to compare the proportion in each study group with detectable levels of immunological markers in breast milk and for those with detectable levels, t-tests will be used to compare the groups if the data is log-normally distributed. If the data is not log-normally distributed a Wilcoxon Rank-Sum test will be used. Where there are important differences between groups, these variables will be added to the models to assess whether these variables mediate any effect of HN001 on eczema and/or atopic sensitisation (using a generalised linear model with a log link and binomial distribution). The association of HN001 with the presence of GDM, BV and GBS will be assessed using relative rates and chi-squared tests. EPDS and State Anxiety Inventory will be compared with a Wilcoxon Rank-Sum test. The effect of weight gain will be investigated as an intermediate variable by adding it to the model (a generalised linear model with a log link and binomial distribution) in the GDM analysis. Any baseline differences that occur will be adjusted for. If commercially available non-study probiotics have been used, a sub-group analysis will be performed after excluding this group, which we do not expect to be large. Reasons for withdrawal from the study will be recorded and examined for differences between study groups. Adjustment will be made for differences in antibiotic use between study groups. SAS version 9.4 will be used.

The PiP Study will assess the efficacy of HN001 supplementation from 14 to 16 weeks gestation in the prevention of GDM, BV, and vaginal GBS colonisation in pregnant women, and development of eczema and atopic sensitisation in their infants at one year and maternal depression and anxiety at 1–2 months postpartum. A unique strength of the study is the range of maternal and infant outcomes examined as a result of a single intervention. If successful the intervention could be adopted into practice with relative ease. This is in contrast to other probiotic eczema prevention strategies which also require direct infant supplementation. Through analysis of biological samples collected prospectively the study may also provide insights into the mechanisms of probiotic action in prevention of gestational diabetes, infant eczema and infant atopic disease. In addition the study will provide further valuable information of the safety of the HN001 supplementation in these populations.

The PiP Study is ongoing, and final outcome data are due to be collected by mid-2016.