Bifidobacterium infantis 35624 and other probiotics in the management of irritable bowel syndrome. Strain specificity, symptoms, and mechanisms

Given the prevalence of irritable bowel syndrome (IBS), its impact on quality-of-life for some sufferers, as well as recent evidence implicating the gut microbiota in the pathogenesis of IBS, there has been keen interest in identifying whether particular strains of bacteria may be of benefit in its management. Thus, while previous systematic reviews have identified that probiotics, in general, are efficacious^1,2, these same authors felt unable, because of a lack of comparative data, to evaluate the relative efficacy of various strains and formulations. In an earlier systematic review, Brenner et al.³ concluded that B. infantis 35624 appeared superior to other strains studied up to that time; since then, however, several other strains and formulations have been subjected to clinical trial. Incorporating more recent data in an up-to-date meta-analysis, Zhang et al.⁴ found that probiotic preparations containing a single probiotic strain were effective in providing improvements in symptoms and QoL. In a new meta-analysis in this issue, Yuan et al.⁵ set out to assess the efficacy of a single strain, B. infantis 35624, in IBS; a laudable goal given the dearth of systematic data on individual strains. They concluded that, while this strain did not improve symptoms in IBS when administered alone, based on their meta-analysis of studies by Charbonneau et al.⁶, O’Mahony et al.⁷, and Whorwell et al.⁸, when they assessed what they believed to be the same strain, it did appear effective in combinations either with other probiotic bacteria⁹ or in a symbiotic mix that also included prebiotics¹⁰.

So far, so good. However, a closer analysis of the Yuan et al.⁵ study reveals some major concerns that seriously undermine the integrity of their conclusions. First, although their stated and highly praiseworthy goal was to do what others had failed to do, namely, focus on a single strain, their analysis involves at least three strains. Although Kim et al.⁹ and Cappello et al.¹⁰ studied combinations which included B. infantis, these combinations did not include the 35624 strain. Specifically, Kim et al.⁹ administered VSL#3, which contains B. infantis SD 5220, and Cappello et al.¹⁰ administered Probinul, which contains BI02. This issue alone defies the basic premise of their approach and contradicts the very title of the paper; this is not a meta-analysis of the efficacy of Bifidobacterium infantis 35624. Furthermore, very recent genomic analyses have reclassified B. infantis 35624 as B. longum subsp. longum 35624^11,12. This further underscores the importance of using strain designations (35624) rather than genus/species to identify individual strains, and highlights that the Yuan et al.⁵ meta-analysis does not, therefore, even include probiotic organisms belonging to the same sub-species. The need to understand and adhere to conventions surrounding bacterial nomenclature was highlighted in a recent commentary: “Probiotic nomenclature matters”¹³. Detailed genomic analysis and comparisons with other bacterial genomes now form the basis of modern taxonomy and, as we learn more of bacterial evolutionary relationships, reclassification of strains has become commonplace. However, it is important to emphasize that a name change does not imply any loss of efficacy or functional attributes for a relabelled strain.

Second, the studies in which B. infantis 35624 was administered as a single strain employed different doses, as noted by Mazurak et al.¹⁴. Charbonneau et al.⁶ employed B. infantis 35624 with a mean count of 1 × 10⁹ colony forming units (cfu’s), O’Mahony et al.⁷ administered 1 × 10¹⁰ cfu’s, while Whorwell et al.⁸ administered either 1 × 10⁶, 1 × 10⁸, or 1 × 10¹⁰ live bacterial cells.

Third, although Yuan et al.⁵ set out to compare the same primary outcomes (“abdominal pain, bloating/distension, and bowel habit”), in one of the three studies that evaluated B. infantis 35624 as a single strain (Charbonneau et al.⁶; coincidentally, the negative study), the primary outcome measure was fecal excretion of the probiotic microbe assessed by quantitative PCR (qPCR); this study was not powered for clinical end-points. A glance at Figures 2 and 3 within this paper⁵ suggests that the exclusion of the Charbonneau et al.⁶ study would have yielded a very different conclusion.

The study by Yuan et al.⁵ adds further fuel to the ongoing debate on whether meta-analysis can reliably be used to combine data from different probiotic strains or combinations. For example, in his editorial, Whelan¹⁵, while acknowledging that “a meta-analysis is a powerful tool to combine individual small trials to improve the power to detect the direction, size, and consistency of an effect”, also emphasized the question marks that surrounded “the use of meta-analysis to combine data from different probiotic species, strains, or combinations” and “recommended that all future meta-analyses of probiotics, in any clinical setting, perform sub-group analysis on specific species/strains and specific combinations¹⁵”. At the very least, therefore, caution must be exercised in interpreting the conclusion of Yuan et al.⁵ that a specific probiotic strain will work better when combined with other probiotic strains. Not only did the combinations that they studied include different strains, but, as Yuan et al.⁵ and others¹⁶ note, it is difficult to tease out what contribution(s) a single strain such as B. infantis 35624 provides when administered as a component of a probiotic cocktail containing multiple strains. More information is needed on interactions between different strains within probiotic cocktails-not just in the product, but also within the gut, and with constituents of the indigenous commensal microbiota. In the meantime, one cannot assume that strains will have additive effects when combined; they may indeed cancel each other out.

The microbiota could contribute to the pathophysiology of IBS through a number of mechanisms⁵; similarly, numerous hypotheses have been invoked to explain probiotic benefits in IBS. These include anti-inflammatory effects (as mentioned by Yuan et al.⁵), modulation of intestinal transit, motility and sensation, as well as alterations in the intra-luminal milieu via deconjugation of bile acids, generation of short chain fatty acids and gases; c.f. review by Camilleri¹⁷. It should also be noted, in addition, that bacteria can produce neuroactive compounds¹⁸. Of these, serotonin may be of particular interest, given biopsy evidence of defects in serotonin signalling in the gut among IBS patients¹⁹, as well as evidence that B. infantis 35624 can increase plasma concentrations of tryptophan, the precursor of serotonin²⁰. Although meta-analyses suggest that selective serotonin reuptake inhibitors do not significantly improve IBS symptoms²¹, serotonin delivered locally by bacteria²² could be more effective.

Animal models have been valuable in identifying the molecular basis of various factors (such as visceral hypersensitivity and stress responses) involved in IBS pathophysiology, and have also provided insights into the ability of probiotic bacteria to beneficially modulate such pathways; c.f. review by Moloney et al.²³. Unfortunately, actions or effects observed in pre-clinical models do not always translate to humans; some work^24–26, some do not^27,28. While pre-clinical studies are useful in initial strain selection, only mechanisms of action studies in humans, as well as high quality clinical trials, can be used to support clinical claims. The heterogeneity of the IBS phenotype makes a unifying pathophysiological explanation most unlikely (e.g. it is frequently associated with psychiatric comorbidity²⁹, which may impact upon treatment response); pending the delineation of coherent sub-types, one cannot expect any one therapeutic approach to be universally successful.

In attempting to rationalize the burgeoning and often conflicting literature on probiotics in IBS it is tempting to combine data in the hope that a coherent message will emerge; this approach may be misguided. The meta-analysis reported in this issue by Yuan et al.⁵ should serve as a warning for those who wander into this minefield. Bigger does not mean better; combining data from studies involving different strains, study populations, and study designs may generate more heat and confusion than light.

Transparency

Declaration of funding

This commentary received no funding.

Declaration of financial/other relationships

The authors have no financial/other relationships to declare.

Acknowledgments

The APC Microbiome Institute is a research centre funded by Science Foundation Ireland, through the Irish Government's National Development Plan. The authors and their work are supported by Science Foundation Ireland (grant number SFI/12/RC/2273) and through EU GRANT 613979 (MYNEWGUT FP7-KBBE-2013-7). The first author is supported by the Health Research Board through Health Research Awards (grant number HRA-POR-2-14-647; GC) and by the Health Service Executive through the Healthy Ageing Award (HaPAI/2015/GC). These funding bodies did not play any role in the writing of this editorial.