FormalPara Key Points

The use of Lactobacillus rhamnosus with or without other probiotics when administered to infants pre- and postnatally has a positive effect in reducing the incidence of eczema/atopic dermatitis in children when evaluated at 2 years out and 6–7 years out.

1 Introduction

Atopic dermatitis (AD), also known as eczema, is a chronic inflammatory skin disease that has recently been recognized as a leading cause of, or precursor to, other atopic conditions such as food allergy and asthma [1, 2]. The age of onset of eczema and the severity of the symptoms has been directly correlated to the risk of future atopic conditions [3]. It is estimated that 30% of infants are diagnosed with AD based on the rates of prescribed medications [4].

Various factors are involved in the development of AD, including genetic predisposition such as filaggrin mutations, a decline in barrier function of the skin, environmental factors and microbial dysbiosis [5]. This dysbiosis extends beyond the skin itself.

For several diseases, including AD and food allergy, patterns in the infant gut microbiome during its developmental stages have been detected [6]. These patterns include differences in overall microbial diversity, the relative prevalence of different phyla and the presence of specific strains of bacteria [7].

Studies have shown that the infant gut microbiome is seeded by maternal transfer of bacteria to offspring beginning in pregnancy [8, 9]. Bacterial DNA can be detected in amniotic fluid, in placental and fetal membranes and in umbilical cord blood [10,11,12]. Maternal transfer of bacteria also occurs during the birthing process and is directly affected by the make-up of the maternal microbiome [8]. Separately, the infant microbiome in early life has been repeatedly shown to alter infant uptake of breastmilk or formula, production of gastrointestinal metabolites and immune regulation [13, 14].

Since the 1990s, the use of probiotic supplementation from the prenatal period through early infancy has been studied as a method to support or optimize gut microbial composition and alter the risk of infant allergic disease. Probiotics are defined as “live microorganisms which when administered in adequate amounts confer a health benefit on the host” [15].

Previous meta-analyses suggest no benefit of oral probiotics in the treatment or prevention of atopic disease [16]. However, the high degree of heterogeneity between studies, which includes differences in probiotic strains, probiotic combinations, probiotic dosages, study populations (maternal vs infant vs both), duration of treatment, stage of intervention, lack of continuation of treatment throughout the perinatal period and outcomes measured impedes direct comparison of studies. This has been a highly debated topic for a number of years in the medical community.

Further, previous systematic reviews have come to a wide range of differing conclusions when probiotics are compared with a placebo in infants with high risk for allergy. One review found that the administration of probiotic microorganisms during pregnancy up until delivery did not have an effect on any outcomes evaluated (including AD) [17]. Similarly, two analyses found that the administration of probiotics may reduce the risk of AD but the evidence was judged to have a low quality [17, 18]. Another meta-analysis combined prenatal plus postnatal with postnatal only administration and found a statistically significant reduction in eczema but not in atopic eczema, with no definition of eczema or atopic eczema provided [19].

When considering bacterial strains, one review found significant risk reduction in atopy with the use of lactobacilli strains (various strains) as monotherapy during pregnancy and lactation [20]. Another analysis concluded that the administration of various genuses of probiotics administered during pregnancy only, in infants early life only or both had a risk reduction in AD when all were combined in a meta-analysis [21]; however, the findings were not statistically significant. Other systematic reviews have found that the administration of various genuses of probiotics administered prenatally and postnatally in infants had a positive and statistically significant effect of reducing the incidence of atopy [20].

These findings can be confusing to interpret as they seemingly contradict each other despite often including the same research. However, probiotic supplementation to a mother prenatally, when the infant’s and mother’s immune systems are effectively combined, is different from probiotic supplementation postnatally when the infant’s gut and immune system are developing independently [18]. Further, it is well understood that the effect of different bacterial strains, even within the same family, may differ [22]. Lastly, and perhaps most importantly, treatment of a disease is fundamentally different than prevention. Therefore, we postulated that the heterogeneity of conclusions in previous meta-analyses resulted largely from too broad a combination of administration protocols, bacterial strains and outcome measures.

Lactobacillus rhamnosus is the most extensively studied strain to date in the treatment of AD [23] and has also been explored for its potential in prevention of AD [24]. The intention of this systematic review and meta-analysis is to build upon a prior systematic review [20] and to focus specifically on the effect of L. rhamnosus when administered both prenatally and postnatally (in order to help remedy the above issues) on eczema, which to the author’s knowledge has not been examined to date. Additionally, this review will also assess the longitudinal effect of L. rhamnosus (± other probiotics) on AD over various timeframes.

2 Methods

This analysis followed the PRISMA guidelines for systematic reviews and meta-analyses [25] (see Appendix A—PRISMA checklist in the electronic supplementary material [ESM]). A systematic review of the literature was undertaken to identify randomized controlled trials (RCTs) of the oral administration of L. rhamnosus either alone or in conjunction with other probiotics during pregnancy and post-pregnancy in mothers and infants (with probiotic exposure via breast milk or oral supplementation) in order to determine its effect on AD and on adverse events.

The definition used for atopic eczema/dermatitis was extracted from the Mayo Clinic website and from the American Academy of Allergy, Asthma, and Immunology (AAAAI), and includes presence of dry skin, red to brownish patches on the body, small raised bumps which may leak fluid when scratched, thickened cracked scaly skin and raw, swollen skin from scratching (collectively a local inflammation of the skin) [26, 27]. As well, AD outcomes were included if they were moderate to severe only (as defined in each included study) by the Nottingham Eczema Severity Score (NESS) [28] and Eczema Area and Severity Index (EASI) [29] scoring. Studies that included other scoring systems (e.g. SCORAD) were also included where moderate to severe dermatitis was assessed. If any part of these definitions existed in the studies identified, they were included in the meta-analysis. Mild eczema/dermatitis was excluded due to the definition used for EASI (mild being barely perceptible) and the definitions found in each of the papers evaluated (AD—pruritis, chronic relapsing; excluding trivial rash, visible eczema, facial and extensor involvement). All of the definitions found in the studies more clearly mapped to moderate and severe eczema.

The following databases were searched from inception through December 8, 2021: PubMed, Cochrane Central Database of Controlled Trials (CENTRAL) and Cochrane Reviews using the search terms (((((((((((((probiotic) AND randomized) AND trial) AND infant) AND eczema)) AND placebo)) AND pregnancy)) AND Lactobacillus) AND rhamnosus)) AND HN001. Subsequent to this, the references of identified studies were hand searched for additional RCT publications.

Data collection was performed by two independent reviewers using Cochrane characteristics and risk of bias forms and then reviewed collectively to determine inclusion and exclusion of studies. Each assessed risk of bias independently and then convened to discuss and review their risk of bias assessments. Where differences existed, the more conservative risk assessment for bias was made (e.g. low to unclear, unclear; unclear to high, high; low to high, high). A PRISMA diagram was used to depict the distillation of included trials. Review manager (Version 5.3) from the Cochrane Collaboration was used in both the qualitative and quantitative assessments made in the current analysis [30].

Risk of bias assessment using Cochrane methodology was undertaken by one author, reviewed by another and then agreed to. The domains in risk of bias that were assessed included bias arising from the randomization process; bias due to deviations from intended interventions (allocation concealment); bias due to missing outcome data (attrition); bias in measurement of the outcome and who was aware of treatment allocation (blinding); bias in selection of the reported result; and bias related to potential conflicts of interest. Publication bias was assessed via funnel plots [31].

Studies were combined for meta-analytic purposes if two or more examined the same outcome during the same timeframe [32]. Statistics used in the analysis (for dichotomous outcomes, e.g. presence or absence of atopic dermatitis) was the Cochrane-Mantel-Haenszel random effects method (for combining results across studies), which is a statistical technique that generates an estimate of an association between an exposure and an outcome, after adjusting for or taking into account confounding [32]. The effect measure was evaluated using risk ratios. Additionally, if future studies examined patients longitudinally, the original study only was referenced.

Heterogeneity (diversity in outcomes) across studies was measured using the I2 statistic. Substantial heterogeneity was noted if the I2 statistic exceeded 50%. If high heterogeneity existed, sensitivity analysis was performed in order to determine which study(s) affected it and the possible reasons why the study was different from the others.

As alluded to above, studies on the evaluation of atopic eczema/dermatitis over time were grouped based on common timeframes for the evaluation on this outcome.

Lastly, a Grading of Recommendations, Assessment, Development and Evaluations (GRADE) was undertaken to assess the quality of the evidence. GRADE is a transparent framework for developing and presenting summaries of evidence and provides a systematic approach for making clinical practice recommendations [33].

3 Results

After duplicates were removed from database searching and hand searching of relevant references, 182 records were screened (i.e. abstracts reviewed). Of these, 59 articles were assessed for eligibility with 48 of these excluded with reasons (i.e. follow up longer term studies of Kalliomäki et al. [2001] [34], Wickens et al. [2008] [35] and Kukkonen et al. [2007] [36], total of six; six systematic reviews and meta-analyses which examined various probiotics on health; and 36 which examined other forms of Lactobacillus or other probiotics, where L. rhamnosus was not included). The PRISMA flow chart can be found in supplementary Fig. 1 (see ESM).

A total of 11 randomized, double-blind, placebo-controlled trials were identified (Table 1), reporting on the incidence of AD following prenatal and postnatal use of L. rhamnosus. Of these, ten studies reported on atopic dermatitis up to 2 years out (N = 2572 mother/infants), three studies up to 4–5 years (N = 1278), three studies up to 6–7 years (N = 588) and two studies up to 11 years (N = 999). Of the studies identified, five took place in Finland [34, 36,37,38,39], two in Norway [40, 41], two in New Zealand [37,[42], one in Germany [43] and one in Taiwan [44]. Five studies used L. rhamnosus solely compared with placebo [34, 35, 42,43,44] and six used L. rhamnosus combined with other probiotics versus placebo [36,37,38,39,40,41].

Table 1 Summary of included studies

Of the 11 studies included, eight studies focused on mother–infant dyads with a family history of atopy/allergies. The other three studies included a general population; however a majority of parents had a history of atopy/allergies, with a range of 70–80% [37, 40, 41].

There was a low risk of bias in five of the seven domains assessed: randomization of sequence generation (11/11); bias in measurement of the outcome (11/11) and who was aware of treatment allocation (11/11); non-selective reporting (11/11); and other biases such as conflicts of interest (10/11) (supplementary Figs 2 and 3, see ESM). There was a high risk of bias in incomplete outcome data (attrition of patients) in 9 out of 11 studies. For allocation concealment (e.g. patients entering treatment almost immediately following randomization), it was unclear in 10 of the 11 studies. Supplementary Fig. 4 is a funnel plot of studies examining the outcome of AD at 2 years. There is a noticeable symmetry in the scatter of the studies indicating a lack of publication bias. Appendix B in the ESM shows the risk of bias assessments for each study.

Fig. 1
figure 1

Forest plot incidence of atopic eczema/dermatitis, < 2 years out

3.1 Outcome of Incidence of Atopic Eczema/Dermatitis

Figure 1 shows the forest plot for incidence of AD out to 2 years in the ten studies examined [34,35,36,37,38,39,40, 42,43,44]. The use of L. rhamnosus during pregnancy and thereafter in infants in the 2-year cohort demonstrated a statistically significant reduction in the incidence of AD (RR 0.60, 95% CI 0.47–0.75); < 0.00001; I2 = 48%).

The use of L. rhamnosus during pregnancy and thereafter in infants in three studies [34,35,36] demonstrated a statistically significant reduction in the incidence of AD (RR 0.74, 95% CI 0.55–1.00; = 0.05; I2 = 61%) 4–5 years out (Fig. 2). The use of L. rhamnosus during pregnancy and thereafter in infants in three studies [34, 35, 41] demonstrated a statistically significant reduction in the incidence of AD (RR 0.62, 95% CI 0.51–0.75; < 0.00001; I2 = 0%) 6–7 years out (Fig. 3). The use of L. rhamnosus during pregnancy and thereafter in infants in two studies [35, 36] did not demonstrate a statistically significant reduction in the incidence of AD (RR 0.68, 95% CI 0.37–1.27; = 0.23; I2 = 74%) out to 11 years (Fig. 4).

Fig. 2
figure 2

Forest plot incidence of atopic eczema/dermatitis, 4–5 years out

Fig. 3
figure 3

Forest plot incidence of atopic eczema/dermatitis, 6–7 years out

Fig. 4
figure 4

Forest plot incidence of atopic eczema/dermatitis, 11 years out

In a post-hoc analysis examining the modes of ingestion of probiotics, mothers and infants received probiotics via the following routes during the perinatal period: mothers only (prenatally and with infants receiving probiotics via breast feeding postnatally) [37,38,39,40,41,42]; mothers and infants both prenatally and postnatally [35, 42, 43, 53, 55]; and mothers prenatally and infants only postnatally [34, 36, 44, 50,51,52]. In a subgroup analysis of modes of ingestion for infants, the following was found: in infants who received probiotics prenatally and via breast milk postnatally, there was a significant reduction in AD (RR 0.52, 95% CI 0.28–0.72; < 0.0001; I2 = 49%) (Supplementary Fig. 5, see ESM); in infants who received probiotics prenatally then ingested them postnatally via diet, there was a significant reduction in AD (RR 0.66, 95% CI 0.46–0.95; = 0.02; I2 = 42%) (Supplementary Fig. 6, see ESM); and in infants who received probiotics prenatally and both the mother and infant continued to ingest them via diet postnatally, there was no statistical difference (RR 0.89, 95% CI 0.64–1.25; = 0.51; I2 = 0%) (Supplementary Fig. 7, see ESM).

In a further post-hoc subgroup analysis of single-strain L. rhamnosus versus mixed strain at 2 years, the RR was 0.58 (95% CI 0.41–0.82; p = 0.002; I2 = 58%; Supplementary Fig. 8, see ESM) and 0.56 (95% CI 0.39–0.81; p = 0.002; I2 = 25%; Supplementary Fig. 9, see ESM), respectively.

3.2 Adverse Events

No adverse events were noted in three of the four studies where adverse events were evaluated [36, 38, 40]. One study noted gastrointestinal symptoms in 39% of the infants during the first 2 months of life in the probiotic arm and 34% in the placebo arm (= 0.44) [39]. Eight of the studies did not report on adverse events.

Table 2 shows the GRADE profile. Overall, the quality of the evidence was low to moderate for each timeframe examined (≤ 2 years, 4–5 years, 6–7 years, 11 years); with a moderate finding owing mainly to a high attrition rate of patients in the studies. The quality of the evidence was determined to be low in one of the timeframes (11 years out) due to imprecision.

Table 2 GRADE evidence profile: Incidence rate of atopic eczema with use of L. rhamnosus vs placebo

4 Discussion

Overall, L. rhamnosus as a monotherapy or when used in conjunction with other probiotic strains during pregnancy through post-pregnancy demonstrated a significant risk reduction in AD in offspring over time; specifically at 2 years and 6–7 years. This finding is an extension of a prior systematic review on the use of monotherapy lactobacilli (various strains) and its risk reduction of AD in infants [20].

This finding is striking in light of previous meta-analyses on the use/administration of probiotics perinatally and their effect on atopic disease in offspring. The previous meta-analyses differ from this study by combining studies with probiotics used in pregnancy only, in infants post-pregnancy only and during pregnancy plus post-pregnancy by mother or infant [17, 20, 21, 23, 45,46,47,48,49]. The previous meta-analyses had also taken a broad perspective on probiotics, combining studies using Lactobacillus rhamnosus, L. acidophilus, L. paracasei, L. reuteri, L. salivarius, B. lactis, B. bifidum, B. longum and B. animalis. Specifically as it relates to ‘like’ meta-analyses, the findings herein are different than those of Szajewska and Horvath [50], who studied L. rhamnosus GG for the prevention of eczema in children and found that L. rhamnosus did not reduce the risk of eczema. Their meta-analysis of five studies included a study which evaluated the administration of L. rhamnosus GG during pregnancy only and a study which examined the administration of L. rhamnosus to infants only. Again, the current analysis included administration of L. rhamnosus with or without other probiotics during the perinatal period. Further, the current analysis builds on a prior 2012 meta-analysis which examined Lactobacilli [20] and not specifically the L. rhamnosus strain. The difference between the Doege et al. meta-analysis [20] and the current one is that Doege et al. combined years 2–7 (whereas the current study broke down the specific timeframes) and the Doege study included L. reuteri, whereas the current analysis only examined L. rhamnosus. Additionally, 11 studies were included in the present analysis whereas Doege et al. only examined four studies.

Even within this limited scope, the daily dosage of L. rhamnosus in billions of CFU varied considerably between studies, as noted in Table 1. Some of the highest dosages of L. rhamnosus were within probiotic preparations containing relatively large counts of other bacterial strains. To date, there have been no independent studies of the effects of interactions between bacterial strains within supplements, or of the minimum or maximum daily dosages of bacterial strains, including L. rhamnosus [48]. There was no noted difference in outcomes when considering correlating daily dosage of L. rhamnosus when used as a monotherapy versus a combination preparation, and thus the results are reported combined.

The 61% heterogeneity in the Fig. 6 analysis (4–5 years out) was examined in sensitivity analysis. If the Kukkonen et al. study [36] were excluded from this meta-analysis, the heterogeneity statistic was 0%. The main difference in the studies included in this 4–5-year out meta-analysis was that the participants in Kukkonen et al. [36] were treated with a combination/mixture of probiotics (2 lactobacilli, bifidobacterial and propionibacteria) while those in the other two studies (Kalliomäki et al. [34] and Wickens et al. [35]) were treated with Lactobacillus strains only. Additionally, by removing Kukkonen et al. [36], this 4–5-year timeframe becomes statistically different on the outcome of AD favoring L. rhamnosus (RR 0.63, 95% CI 0.47–0.84; = 0.001; I2 = 0%). Further, the 74% heterogeneity in Fig. 8 was examined. The differences in the studies included a longer treatment period in Wickens et al. [35] of 2 years versus 6 months in Kukkonen et al. [36], and the use of Lactobacillus versus a combination/mixture of probiotics (2 lactobacilli, bifidobacterial and propionibacteria) in Kukkonen et al. [36].

The limitations of this analysis include a high attrition rate of those entered into the trials. However, as it relates to other potential biases in the trials, it was considered low. The GRADE assessments timeframes (< 2 years, 6–7 years) were considered moderate in nature due to this attrition rate and low due to attrition and imprecision in the 4–5-year and 11-year timeframes. Thus, the true effect is likely to be close to the estimated effect in the < 2-year and 6–7-year timeframes. There is a possibility that the true effect may be different [33]. However, the effects in new studies would need to be quite large in order to change the relative effect and confidence intervals included herein. Over 2500 patients were included in the cohort of < 2 years. Thus, a significant number of new patients would be required in future studies to affect these findings, even if the effects were smaller in nature.

The overall implications of this meta-analysis in comparison with previous research are that probiotics may have useful clinical effects, however, the administration protocol and the specific strains utilized should be carefully considered. Targeted inclusion criteria for those at higher risk of atopic disease might improve efficacy while reducing the need for longer interventions. More research is needed in these areas as the understanding and possibilities of probiotic supplementation continue to evolve.

5 Conclusion

Based on this analysis, the use of L. rhamnosus either solely or in conjunction with other probiotics during pregnancy and post-pregnancy in infants likely has a positive effect in reducing the incidence of AD. This finding was found in the 2-year and 6–7-year timeframes evaluated.